Professional Documents
Culture Documents
Applied Soft Computing: Yu-Jun Zheng, Sheng-Yong Chen, Hai-Feng Ling
Applied Soft Computing: Yu-Jun Zheng, Sheng-Yong Chen, Hai-Feng Ling
Applied Soft Computing: Yu-Jun Zheng, Sheng-Yong Chen, Hai-Feng Ling
a r t i c l e i n f o a b s t r a c t
Article history: Effective planning and scheduling of relief operations play a key role in saving lives and reducing dam-
Received 28 December 2013 age in disasters. These emergency operations involve a variety of challenging optimization problems, for
Received in revised form 30 August 2014 which evolutionary computation methods are well suited. In this paper we survey the research advances
Accepted 28 September 2014
in evolutionary algorithms (EAs) applied to disaster relief operations. The operational problems are classi-
Available online 5 October 2014
fied into five typical categories, and representative works on EAs for solving the problems are summarized,
in order to give readers a general overview of the state-of-the-arts and facilitate them to find suitable
Keywords:
methods in practical applications. Several state-of-art methods are compared on a set of real-world emer-
Disaster relief
Emergency operational problems
gency transportation problem instances, and some lessons are drawn from the experimental analysis.
Evolutionary algorithms (EAs) Finally, the strengths, limitations and future directions in the area are discussed.
Optimization © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA
license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
∗ Corresponding author. Tel.: +86 571 85290085. 1. Provide readers a general overview of the major developments
E-mail address: yujun.zheng@computer.org (Y.-J. Zheng). emerged throughout the years.
http://dx.doi.org/10.1016/j.asoc.2014.09.041
1568-4946/© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).
554 Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566
m
n
Fig. 1. The numbers of research articles on EAs for disaster relief operations in recent Although this problem can be solved by specialized methods
ten years (the data of 2014 is up to May).
that are more efficient than standard linear programs [10], EAs can
provide much better performance on very large problem instances
2. Show major strengths and shortcomings of the state-of-the-arts, [11]. Moreover, real-world problems typically have to deal with
and help practitioners to find valuable approaches that can be heterogenous commodities and transportation modes, involving
referred in the practice of disaster relief. additional (and often nonlinear) objectives and constraints [12].
3. Discuss potential directions for future research, and stimulate Such problems should be characterized by more complex integer
more interest in this cross-disciplinary field. programming (IP) or mixed integer programming (MIP) models,
which are much better suited for evolutionary optimization than
In the remainder of the paper, we first classify optimization standard optimization method.
problems in disaster relief operations into five typical categories Furthermore, transportation in emergency logistics [13] is often
in Section 2, and describe recent advances in EAs for solving the subject to uncertainty and randomization, has very limited time
problems in Section 3. Section 4 presents the experiment of six typ- constraints, focuses more concern on timeliness and/or risk rather
ical EAs on a set of real-world emergency transportation problem than transportation cost. Taking these extra properties into consid-
instances. Section 5 discusses the strengths, limitations and future eration, much more efforts should be put into the design of efficient
directions in the area, and finally Section 6 concludes. algorithms.
2. Problem classification
2.2. Location problems
Disaster relief operations involve activities including estab-
lishing emergency facilities, searching and rescuing survivors,
The problems of locating emergency facilities, such as fire sta-
providing health and medical assistance, distributing relief sup-
tions, medical services, shelters, etc., based on the topography of
plies, transferring injuries, scheduling rescue forces, etc., and the
potential facilities, can be divided into two sub-classes: continu-
coordination of these activities across organizations [8,9]. Thus they
ous facility location problems, where facilities are allowed to be
are associated with a variety of operational problems that often lie
sited on every point in a planning area, and discrete facility loca-
outside the scope of the conventional optimization methods. In this
tion problems, where there is a discrete set of candidate locations.
paper, we classify the problem into the following five categories:
Generally speaking, the first class often uses formulation of LP and
nonlinear programming (NLP) problems, while the second class is
• General transportation planning problems, which are to make up
often modeled by classical covering problems and their variants,
programs for delivering relief supplies from distribution centers such as set covering, vertex/egde covering, covering location, and
(sources) to demand points (targets). But we do not place detailed partial covering [14,15].
path planning and vehicle routing in this category. The simplest form of the facility location problem, which con-
• Facility location problems, which are to arrange emergency facil-
siders providing a single facility to cover a maximum number of
ities on appropriate locations to serve the demand points. demand points, has efficient exact solution methods [16,17]. How-
• Routing problems, which include planning routes for vehicles,
ever, involving multiple facilities leads to NP-hard problems, such
rescuers, and evacuees. as the p-median problem (2) and the maximum covering problem
• Roadway repair problems, which are to repair damaged roadways
(3):
and rehabilitate the lifelines to demand areas.
• Integrated problems, which need to solve a set of above individual
problems under one or more common objectives. min f = cij yij
s.t. xj = p
As a classic problem in OR, the basic transportation problem
j∈F
(2)
considers delivering a homogeneous commodity from a set of m yij = 1, ∀i ∈ D
sources to a set of n targets. Suppose the supply of source i is ai , j∈F
the demand of target j is bj , the cost for transporting one unit of xj ∈ {0, 1}, ∀j ∈ F
commodity from source i to target j is cij , then the problem is to
determine the commodity amount xij from each source i to each yij ∈ {0, 1}, ∀i ∈ D, ∀j ∈ F
Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566 555
max f = wi zi 3. Evolutionary optimization methods for relief
operational problems
i∈D
Table 4 Table 5
The characteristics of ACO-based approaches. The characteristics of BBO-based approaches.
Table 6 Table 7
The characteristics of AIS-based approaches. The characteristics of other heuristic search approaches.
Other heuristic search methods including simulated annealing 3.8. Hybrid algorithms
(SA), tabu search (TS), harmony search (HS), variable neighborhood
search (VNS), etc., are effective particularly in combinatorial opti- Different optimization methods have different design principles
mization problems, and have also been reported in the literature and application areas. By exploiting the strengths of two or more
in for disaster relief. SA [89] is a classical global search heuris- solution methods, we have a chance to obtain a powerful approach
tic that simulates the physical annealing process in the field of that is much more competitive than any individual method. For
optimization. Jahangiri et al. [90] presented a problem for opti- solving a large-scale multi-depots VRP in relief work, Peng et al.
mizing signal timing and increases the outbound capacity of the [100] proposed a method combining GA with ACO and SA. Using a
network for emergency evacuation, and applied a SA algorithm feedback loop, the best result of GA is used to improve ant systems
for searching the optimal solution to the problem. Ma et al. [91] and vice versa, in order to take advantage of both the fast conver-
designed a genetic-SA algorithm for emergency service facility gence of GA and the preciseness of ACO, and the SA procedure is
location problem. Experiments showed that algorithm can clearly incorporated to avoid the local optimal. Zhang et al. [101] devel-
gets the afflicted points that are covered by the selected facilities oped a hybrid AIS and ACO method for a route choice problem in
and thus improve searching efficiency. emergency logistics. The main idea is to obtain pheromone distribu-
TS uses an adaptive memory named tabu list to record the vis- tion from AIS, and optimize the solutions by basic ACO. Simulation
ited solutions and thus improves the local search to avoid local results show that the method has more effective timeliness than
optima [92,93]. Xie and Turnquist [94] solved a lane-based evacu- basic ACO.
ation network optimization problem that integrates lane reversal Hu et al. [102] studied the allocation of earthquake emergency
and crossing elimination strategies by using a relaxation and TS shelters which involves multiple sites and strict constraints, and
algorithm, and applied the approach to optimize a regional evacu- developed a new discrete PSO which uses a feasibility rule to handle
ation network for a nuclear power plant. Zheng et al. [95] designed the constraints, and embeds a SA for potentially escaping from local
a multiobjective TS for an emergency equipment maintenance optima. The approach has been applied to the allocation of earth-
scheduling problem, which aims to achieve a good balance between quake emergency shelters in the Zhuguang Block of Guangzhou
operational capability, reserved maintenance capability, as well as City, China.
cost-effectiveness. The approach has demonstrated its effective- Zhang and Fei [103] designed a novel approach that integrated
ness in several real-world operations. ACO with another SI method, fish swarm algorithm [104], to solve
HS is another memory-based algorithm inspired from a jazz a weak economy emergency logistics path optimization problem,
music orchestra when playing and improvising in their attempt the aim of which is to gain a win–win result of the lowest cost
to achieve a progressively better harmony [96]. Landa-Torres et al. on condition of meeting all affected points time requirements. The
[97] proposed a multi-objective HS algorithm for the efficient dis- hybrid method, namely the fish swarm ant colony algorithm, takes
tribution of 24-h emergency units. This paradigm is essentially a the advantage of the structure of the crowded degree factor in the
facility location problem that involves determining the optimum ant colony for improving the model of fish swarm in the ability of
locations of 24-h emergency resources and the optimum assign- global search. A similar research was conducted by Fei et al. [105],
ment of patients to the resources through the existing medical care where an ACO based on SA is proposed for path optimization in
infrastructure. Their algorithm redefines the grouping encoding to emergency logistics.
reduce the dimension of the search space and employs the nondom- For solving the problem of post-disaster infrastructure man-
inated sorting for solution evaluation. Simulations in a real scenario agement, Kallioras et al. [106] proposed an integrated method,
based on geographic data of medical centers over the provinces of where an improved HS algorithm is used for emergency inspection
Guadalajara and Cuenca, Spain, demonstrate the robustness and scheduling, and an ACO algorithm is employed for solving the asso-
applicability of the proposed approach. ciated routing problem. Comparative results show the proposed
Ozdamar and Yi [98] used greedy neighborhood search to method outperforms other algorithms including PSO, DE, the basic
develop a fast constructive heuristic, named Path-Builder, to HS, and the pure random search.
support relief and evacuation operations. The method extends Becker et al. [107] introduce a new multi-agent algorithm for
the neighborhood structure to suit the problem’s special needs, search and rescue operations for exploration of unknown terrain.
and selects partial paths to append to vehicle routes in the The hybrid method uses the flood algorithm to a fast exploration
operations, and thus exploits foreseeable opportunities within of the unknown terrain, and utilize the path optimizing features
the vehicles’ limited neighborhood. Considering the allocation of ACO for constructing short paths from points of interest back to
of scarce resources to repair a rural road network, Duque and the base. The feasibility of the approach is demonstrated by agent-
Sörensen [99] used GRASP and VNS to maximize the accessibility based simulations.
of as many people as possible to the main cities or regional centers Considering emergency railway wagon scheduling between
where the economic and social infrastructure is usually located. multiple source, center, and target stations, Zheng et al. [80] pre-
The efficiency of the approach is demonstrated by applying it to a sented an integer programming model, the calculation of objectives
large real-life motivated instance. and constraints of which further include a set of nonlinear subprob-
Table 7 summarizes the characteristics of typical heuristic meth- lems. For efficiently tackling the problem, the authors proposed a
ods other than the EAs mentioned in the previous sections. hybrid BBO algorithm which uses a local ring topology to avoid
Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566 561
premature convergence, employs the DE mutation to enhance • For two feasible solutions, the one with smaller objective function
exploration, and takes some problem-specific mechanisms to fine- value is fitter.
tuning the search. Computational experiments and a real-world • For two infeasible solutions, the one with lower constraint viola-
case study demonstrate the effectiveness and the scalability of the tion is fitter.
algorithm.
The algorithms are tested on a set of 10 problem instances
4. Experimental analysis on an integrated transportation from real-world relief operations occurred in China since 2008, the
problem dimensions of which range from 45 to 1080. The experiments are
conducted on a computer of Intel Core i5-2520M CPU and 8 GB
In this section we test the performance of some typical EAs memory. For each instance, a maximum running time limit Tmax
on an emergency problem that integrates transportation plan- (in minutes) is set for all the EAs according to the response time
ning depicted in Section 2.1 and vehicle routing in Section 2.3. The requirement in the emergency, and each algorithm is run 30 times
problem extends model (1) with heterogeneous commodities and with different random seeds. Table 8 summarizes the experimental
vehicle capacity constraints as follows: results.
The first two instances are respectively derived from the opera-
m
n
K
tions of two moderate earthquakes (the 2010 mag 4.6 earthquake
min f = wjk t(xijk ) in Ali, Tibet autonomous region and the 2009 mag 4.7 earthquake
i=1 j=1 k=1 in Baicheng, Jilin province). The relief demands are not strong and
m
the delivery tasks are not heavy. As we can see, on instance #1
s.t. xijk ≥ bjk , j = 1, . . ., n, k = 1, . . ., K all the EAs always achieve the same result; on instance #2 four
i=1 EAs except SaDE and B-BBO always achieve the same result on #2.
n (4) Post-hoc analysis shows that the results are real optimal solutions.
xij ≤ aik , i = 1, . . ., m, k = 1, . . ., K Note SaDE and B-BBO also frequently reach the optimum of #2. This
j=1 shows that there are no significant performance differences among
n
K
these popular EAs on the two small instances.
xijk ≥ ci , i = 1, . . ., m Instance #3 also represents a moderate earthquake relief oper-
j=1 k=1 ation (the 2011 mag 4.8 earthquake in Anqing, Anhui province),
xijk ≥ 0, i = 1, . . ., m, j = 1, . . ., n, k = 1, . . ., K but the earthquake occurred in a more developed region, where
the transportation network is much more complex, and thus it
where K is the number of types of relief commodities, bjk and aik
is quite difficult to find the optimal relief delivery plan. Within
are the demand and supply of commodity k of target j and source i
the limited running time, none of the algorithms reach the opti-
respectively, ci is the maximum vehicle capacity available at source
mum; EBO achieves the best result among the six EAs, but it has no
i, and wjk is the importance weight of demand item bjk . xijk , the
significant difference from GPSO, SaDE and SPO 2007.
amount of commodity k from source i to target j, is the decision vari-
Instance #4 represents the operation of the 2011 mag 5.8 earth-
able, while t(xijk ) is the corresponding arrival time to be calculated
quake in Yingjiang, Yunan province. Although the demand is not
based on a vehicle routing subproblem for each source i. In real-
strong, the operation involves much more supply centers, which
world relief operations, the integrated problem and its variations
significantly enlarges the solution space and increases the num-
occur much more frequently than the two individual subproblems
ber of local optima. On this instance, GPSO exhibits the best mean
[52,108].
performance and EBO ranks second, and the two methods show sig-
The evaluation of each solution to the above main problem
nificant performance improvement over the other four EAs, mainly
involves solving a VRP instance and thus is computationally expen-
because their exploration abilities help to jump out of the local
sive. An algorithm with high search ability can effectively reduce
optima. An interesting finding is that all the EAs have reached the
the number of function evaluations and improve the response
optimal solution among the 30 runs, which shows the random seed
speed. Here we comparatively run the following six state-of-the-art
also affects the algorithm performance to a certain degree on this
EAs on the main problem:
instance.
Instance #5, which represents the operation of the 2010 mag
• The standard PSO 2007 [109] which equips the basic PSO with a
6.0 earthquake in Chuxiong, Yunan province, is similar to #4 in
local random topology and a few fine-tuning adjustments. the huge number of local optima. Here EBO exhibits the best
• The self-adaptive DE (SaDE) [110] whose mutation schemes and
mean performance and GPSO ranks second, showing significant
control parameters are self-adapted by learning from previous improvement over the other four. Except them, PSO 2007 can also
generations. occasionally reach the optimum among the 30 runs.
• The Blended BBO (B-BBO) [111] which improves the basic BBO
For instance #6 (the 2010 Zhouqu mudslides), the relief demand
with a blended migration operator. is strong and the solution space is large. However, the number of
• The GA combined with three fast improvement procedures [31].
local optimal is not so huge, mainly because the demand points are
• The GPSO [112] that combines PSO with a gradient-based local
located in a very small area. Within the limited time, GPSO and EBO
search for accurate local exploration. achieve much better results than the other EAs. However, it is found
• The EBO [82] which enhances BBO with a local topology and the
that, when given enough running time (about 20–30 min), all the
combination of a global migration and a local migration. EAs can reach very close to the same optimum, which is a unique
phenomenon among the test instances.
In most real-world relief operations, the average number of tar- The last four instances are all large-scale operations with quite
gets assigned for each source is relatively small, and thus we employ large solution space and huge number of local optima. On #7
the branch-and-bound method [113] for solving VRP for all the EAs (the 2010 mag 7.3 earthquake in Yushu, Gansu province), GPSO
for a fair comparison. We also use the following constraint handling and EBO exhibit significant performance improvement over the
method in fitness comparison in all the EAs: others. The mean result of EBO is the best, but it has no sig-
nificant difference from GPSO. While on the more difficult #8
• A feasible solution is always fitter than an infeasible one. (the 2013 mag 6.6 earthquake in Dingxi, Gansu), EBO exhibits
562 Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566
Table 8
The experimental results of six EAs on the test problem instances (“mean” denotes the average result objective value over the 30 runs, “std” the corresponding standard
deviation, and “best” the best objective value among the 30 runs).
significant performance advantage over all the other EAs including range of benchmark problems, exhibits mediocre performance
GPSO. on our test instances, mainly because it requires feedback from
Instance #9 represents the operation of the 2008 mag 8.0 previous generations to adjust the search mechanism, but the max-
earthquake in Wenchuan, Sichuang province – the most serious imum running time is too limited to allow enough feedback. In
disaster in China in the recent decade. Among all the EAs, EBO comparison, EBO and GPSO balance exploration and exploitation
also exhibits the best performance. Nevertheless, according to the much more effectively and thus exhibit much better performance.
decision-maker/experts of the relief operation, even the mean However, on extremely large instances such as #10, the required
result of EBO is inefficient; in fact, only the best result of EBO running time (or number of generations) for producing quality solu-
(reached thrice among the 30 runs of the algorithm) is considered tions is contradictory to the limited response time, which is an open
efficient. problem faced by current research.
Instance #10, the operation of the 2013 mag 7.0 earthquake in From the experiments on the typical problem, we can draw the
Ya’an, Sichuang province, is even more difficult than #9 because following lessons:
the disaster response system in China has been much improved
since 2008, and thus the event involves much more relief demands • Generally the balance between exploration and exploitation is
and forces. On this instance, in general, none of EAs can reach a a key to achieve promising performance. Particularly, under
solution considered efficient according to post-hoc analysis. The emergency conditions, exploration ability (such as provide by
best solution of EBO (reached twice among the 30 runs) is accept- PSO and DE) can be more important in producing acceptable
able, but not so satisfactory to the decision-maker. In fact, we solutions within limited response time, while the design of effec-
can obtain an efficient solution (with an objective value around tive yet efficient exploitation mechanisms can be much more
3.50E+10) after about 60 EBO runs, each with 1 hour running difficult.
time. • Self-adaptive mechanisms, which are useful for a wide vari-
In summary, EBO and GPSO generally perform better the other ety of problems, may be not effective for emergency problems,
four EAs on the test instances, and their performance advantages because the response time often limits the number of algorithm
increase with the problem size/difficulty. On relatively large-size generations and thus cannot allow enough feedback to enhance
instances #7–#10, the performance of B-BBO and GA becomes performance.
much inferior, because their global search abilities are not strong • For different types of optimization problems in disaster relief
and thus have difficulty in exploring a very large solution space. operations, there is no a magic bullet algorithm that can be uni-
PSO 2007 has a fast convergence speed, but its results are not formly better than any other algorithm (known as the “No Free
very satisfactory on large-size instances due to the limited local Lunch Theorem” [114]). Thereby, it will be very helpful to obtain
exploitation ability. SaDE, an EA that is competitive on a wide a priori knowledge about the problem in disaster preparedness,
Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566 563
and utilize them in the design of problem-specific mechanisms to algorithm [118], etc., have been proposed and demonstrated their
facility the search for quality solutions, especially for very large- effectiveness and efficiency on a variety of problems. There is
scale operations. plenty of room for these new metaheuristics as well as their
hybridization with existing methods in the area of disaster
5. Strengths, limitations and future perspectives relief.
• Pareto-based multiobjective optimization and decision support.
From the survey of recent literature, we can clearly see that MOEA embodies an exciting opportunity to realize the full poten-
EAs represents highly competitive alternatives to conventional tial of the EA concept [119], and it is expected that studies on the
approaches for complex disaster relief operational problems where optimization of disaster relief operations would benefit greatly
these latter methods are often unaffordable or require unaccept- from the exploitation of the power of MOEA techniques. On the
able assumptions. The main strengths of EAs for these critical other hand, multi-criteria decision making techniques including
applications can be summarized as follows: preference articulation, TOPSIS [120] and Pareto-front visualiza-
tion, should be effectively utilized to help decision makers to
understand the resulting Pareto-optimal solutions and choose the
• EAs are general metaheuristics that can be easily applied or
most appropriate one(s) for implementation.
adapted to a large variety of operational problems, in particular • Hyper-heuristics. Defined as “heuristics to choose heuristics”,
those new problems with little prior knowledge.
hyper-heuristics [121] are high level-heuristics that perform
• Empirically, EAs are capable of finding high quality solutions
search over a set of low-level heuristics, and can often achieve not
for most problems in consideration, although theoretically their
only better solutions but also much higher flexibility and robust-
convergence to global optima is only guaranteed in a weak prob-
ness on very difficult problems. Hyper-heuristics have been
abilistic sense.
successfully applied to a wide range of planning and scheduling
• Population-based search mechanism and elaborately designed
problems, and their potential in emergency optimization prob-
evolutionary operators advance solutions in a quite efficient
lems remains to be explored.
manner, resulting in considerable reduction in computational • Test suites and real-world applications. It is of vital importance
resource and thus meeting stringent response time requirements
to determine to which degree the solutions of EAs would meet
for problem solving.
the stringent quality and safety requirements of relief oper-
• EAs are particularly effective in handing problems with various
ations. However, there are few benchmark problems publicly
and often incompatible types of objectives and/or constraints,
available for this purpose. It is desirable for the community
which are very common in emergency problems under stressful
to provide more standard test suites, test case generators, as
disaster conditions.
well as evaluation tools for assessment and analysis of dif-
• EAs perform well on problems with noise and uncertain param-
ferent EAs. More importantly, proposed solution approaches
eters, and produce solutions robust and adaptable to changing
are required to be fully implemented and verified on cho-
environments.
sen real-world problems to demonstrate their practicality and
• The optimization processes of EAs are flexible, and can be tuned
effectiveness.
to a great extent to the problems at hand.
6. Conclusion
However, the application of EAs to disaster relief operational
problems, in comparison with other common operational prob-
The paper gives a survey of research advances in evolu-
lems, still has many limitations and disadvantages. First, GAs are
tionary optimization applied to disaster relief operations. The
in the majority of EAs used for emergency operational problems;
survey is not intended to be all embracing. The main purpose
Many state-of-the-art methods (in particular those hybrid algo-
is to help readers have a general overview of the state-of-the-
rithms) which have been demonstrated to be much more efficient
arts and easily refer suitable methods in practical applications,
on benchmark problems, have been less explored.
given that the works described here are indicative of the breadth
Second, although MOEAs have been proved to be singularly
of emergency operational problems where EAs have proved
successful in effectively addressing MOO problems, most research
profitable.
on MOO in emergency operations still simply combines multiple
As a clear finding of the survey, EAs have found a growing num-
objectives into a single one, the results of which are much less
ber of applications in disaster relief operations in recent years.
comprehensive to support decision makers who have to resolve
However, in comparison with other typical areas in OR, the poten-
a divergence of interest.
tial of evolutionary computing in emergency operational problems
Third, most computational experiments are conducted on test
is far from being realized. The true challenge that we must over-
problems generated randomly. Only a few research uses real-world
come, and this is true of soft computing in general, is to convince
data for testing. More importantly, reports on successful applica-
sceptical decision makers that the new methods are capable of pro-
tions of EAs to real-world relief operations are very scarce. Only in
ducing results worthy of their trust. To achieve this, the methods
very recent years, several works on bio-inspired methods for trans-
have to be actually applied to and demonstrate their performance
portation planning in the 2011 Yingjiang earthquake [52] and the
on more real-world operations.
2013 Lushan earthquake [80] and repair scheduling in the 2013 Dixi
earthquake [81] have been reported.
The future use of evolutionary computation in disaster relief Acknowledgements
operations is tied to both the increasing difficulty of practical prob-
lems and the advances of EAs themselves. In view of the current This work was supported by Project 61020106009, 61105073,
limitations in this area, the following aspects deserve attention in 613250211 and 61473263 supported by Natural Science Founda-
future development initiatives: tion of China.
Table 9
The summary of the research works classified by problem category.
GA Tuson et al. [26] Wang and Zhang [32] Ma et al. [39] Sato and Ichii [48] Chang et al. [51]
Feng and Wen [27] Kongsomsaksakul et al. [33] Lin et al. [40] Chen and Tzeng [49] Zheng and Ling [52]
Na and Zhi [28] Jia et al. [34] Zhang et al. [41] Zhang and Lu [50]
Nolz et al. [29] Han and Zhang [35] Hamedi et al. [42]
Yang et al. [30] Chai et al. [36] Chung et al. [43]
Berkoune et al. [31] Tzeng and Chen [37] Liu et al. [44]
Yang et al. [38] Song et al. [45]
Saadatseresht et al. [46]
Ma et al. [47]
PSO Pang et al. [66] Uno et al. [64] Wang et al. [70] Gan et al. [68]
Bozorgi-Amiri et al. [67] Chai et al. [65] Zheng et al. [71] Tian et al. [69]
Hu et al. [102]
ACO Lim et al. [74] Yuan and Wang [75] Yi and Kumar [76]
Yan and Shih [77]
BBO Zheng et al. [82] Zheng et al. [81] Zheng et al. [80]
References [17] K. Hogan, C. Revelle, Concepts and applications of backup coverage, Manag.
Sci. 32 (1986) 1434–1444.
[1] M.R. Garey, D.S. Johnson, Computers and Intractability: A Guide to the Theory [18] B. Eksioglu, A.V. Vural, A. Reisman, The vehicle routing problem: a taxonomic
of NP-Completeness, W.H. Freeman, San Francisco, 1979. review, Comput. Ind. Eng. 57 (2009) 1472–1483.
[2] S. Yang, Y.-S. Ong, Y. Jin, Evolutionary Computation in Dynamic and Uncertain [19] M. Pinedo, Scheduling Theory, Algorithms, and Systems, 2nd ed., Prentice Hall,
Environments, Studies in Computational Intelligence, Springer-Verlag New New York, 2002.
York, Inc., Secaucus, NJ, USA, 2007. [20] S. Yan, Y.-L. Shih, A time-space network model for work team scheduling after
[3] W. Pedrycz, Fuzzy evolutionary computing, Soft Comput. 2 (1998) 61–72. a major disaster, J. Chin. Inst. Eng. 30 (2007) 63–75.
[4] C. Grosan, A. Abraham, H. Ishibuchi, Hybrid Evolutionary Algorithms, [21] S. Yan, Y.-L. Shih, Optimal scheduling of emergency roadway repair and sub-
Springer, Berlin, 2007. sequent relief distribution, Comput. Oper. Res. 36 (2009) 2049–2065.
[5] J.H. Holland, Adaptation in Natural and Artificial Systems: An Introductory [22] K. Miettinen, Nonlinear Multiobjective Optimization, Kluwer, Boston, 1999.
Analysis with Applications to Biology, Control and Artificial Intelligence, MIT [23] K. Deb, Multi-objective optimization using evolutionary algorithms, John
Press, Cambridge, MA, USA, 1975. Wiley & Sons, Chichester, 2001.
[6] J. Kennedy, R. Eberhart, Particle swarm optimization, in: IEEE Interna- [24] C.A.C. Coello, D.A. Van Veldhuizen, G.B. Lamont, Evolutionary Algorithms for
tional Conference on Neural Networks, vol. 4, 1995, pp. 1942–1948, Solving Multi-Objective Problems, Springer, New York, 2002.
http://dx.doi.org/10.1109/ICNN.1995.488968. [25] Q. Zhao, A general framework for cooperative co-evolutionary algo-
[7] M. Dorigo, G. Di Caro, Ant colony optimization: a new meta-heuristic, in: rithms: a society model, in: Evolutionary Computation Proceedings,
Proceedings of the 1999 Congress on Evolutionary Computation, IEEE, vol. 2, IEEE World Congress on Computational Intelligence, 1998, pp. 57–62,
1999, pp. 1470–1477. http://dx.doi.org/10.1109/ICEC.1998.699134.
[8] W. Fan, The consideration and suggestion of scientific problems on emergency [26] A. Tuson, R. Wheeler, P. Ross, An evolutionary/meta-heuristic approach to
management of national sudden public events, Sci. Found. China 21 (2007) emergency resource redistribution in the developing world, in: Artificial
71–76. Neural Nets and Genetic Algorithms, Springer Vienna, 1998, pp. 329–332,
[9] X. Zuo, Q. Ran, W. Gu, Research on emergency logistics: situations http://dx.doi.org/10.1007/978-3-7091-6492-1 72.
and development trends, in: IEEE International Conference on Grey Sys- [27] C.-M. Feng, C.-C. Wen, Traffic control management for earthquake-raided
tems and Intelligent Services, 2009, pp. 1638–1641, http://dx.doi.org/ area, J. East. Asia Soc. Transp. Stud. 5 (2003) 3261–3275.
10.1109/GSIS.2009.5408177. [28] L. Na, L. Zhi, Emergency relief goods multi-mode transportation based
[10] V. Srinivasan, G.L. Thompson, Benefit-cost analysis of coding techniques for on genetic algorithm, in: Second International Conference on Intelligent
the primal transportation algorithm, J. ACM 20 (1973) 194–213. Computation Technology and Automation, vol. 1, 2009, pp. 181–184,
[11] G.A. Vignaux, Z. Michalewicz, A genetic algorithm for the linear transportation http://dx.doi.org/10.1109/ICICTA.2009.52.
problem, IEEE Trans. Syst. Man Cybern. 21 (1991) 445–452. [29] P.C. Nolz, K.F. Doerner, W.J. Gutjahr, R.F. Hartl, A bi-objective metaheuristic for
[12] H. Lau, T. Chan, W. Tsui, F. Chan, G. Ho, K. Choy, A fuzzy guided multi-objective disaster relief operation planning, in: C. Coello Coello, C. Dhaenens, L. Jourdan
evolutionary algorithm model for solving transportation problem, Expert (Eds.), Advances in Multi-Objective Nature Inspired Computing, vol. 272 of
Syst. Appl. 36 (2009) 8255–8268. Studies in Computational Intelligence, Springer Berlin Heidelberg, 2010, pp.
[13] L. Özdamar, E. Ekinci, B. Küçükyazici, Emergency logistics planning in natural 167–187, http://dx.doi.org/10.1007/978-3-642-11218-8 8.
disasters, Ann. Oper. Res. 129 (2004) 217–245. [30] L. Yang, Z. Gao, K. Li, Railway freight transportation planning with mixed
[14] M. Melo, S. Nickel, F.S. da Gama, Facility location and supply chain manage- uncertainty of randomness and fuzziness, Appl. Soft Comput. 11 (2011)
ment - a review, Eur. J. Oper. Res. 196 (2009) 401–412. 778–792.
[15] A.B. Arabani, R.Z. Farahani, Facility location dynamics: an overview of classi- [31] D. Berkoune, J. Renaud, M. Rekik, A. Ruiz, Transportation in disaster response
fications and applications, Comput. Ind. Eng. 62 (2012) 408–420. operations, Socio-Econ. Plan. Sci. 46 (2012) 23–32.
[16] R. Church, C.R. VELLE, The maximal covering location problem, Pap. Reg. Sci. [32] D.-W. Wang, G.-X. Zhang, Model and algorithm to optimize location of cata-
32 (1974) 101–118. strophic rescue center, J. Northeast. Univ. (Nat. Sci.) 26 (2005) 953–956.
Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566 565
[33] S. Kongsomsaksakul, C. Yang, A. Chen, Shelter location-allocation model for [63] R. Eberhart, Y. Shi, Tracking and optimizing dynamic systems with particle
flood evacuation planning, J. East. Asia Soc. Transp. Stud. 6 (2005) 4237– swarms, in: Proceedings of the 2001 Congress on Evolutionary Computation,
4252. vol. 1, 2001, pp. 94–100, http://dx.doi.org/10.1109/CEC.2001.934376.
[34] H. Jia, F. Ordóǹez, M.M. Dessouky, Solution approaches for facility location [64] T. Uno, K. Kato, H. Katagiri, An application of interactive fuzzy satisficing
of medical supplies for large-scale emergencies, Comput. Ind. Eng. 52 (2007) approach with particle swarm optimization for multiobjective emergency
257–276. facility location problem with a-distance, in: IEEE Symposium on Compu-
[35] C.-f. Han, C. Zhang, Genetic algorithm for solving problems in emergency man- tational Intelligence in Multicriteria Decision Making, 2007, pp. 368–373,
agement, in: Fifth International Conference on Natural Computation, vol. 4, http://dx.doi.org/10.1109/MCDM.2007.369115.
2009, pp. 259–264, http://dx.doi.org/10.1109/ICNC.2009.333. [65] G. Chai, Y.-Y. Sun, C.-H. Zhu, Particle swarm optimization algorithm
[36] G. Chai, C. Fang, X. Gao, Q. Zhao, A cost-based study on highway traffic for emergency resource allocation on expressway, in: Proceedings of
emergency rescue sites location using heuristic genetic algorithm, J. Comput. the First ACM/SIGEVO Summit on Genetic and Evolutionary Compu-
Inform. Syst. 7 (2011) 507–514. tation, GEC 2009, ACM, New York, NY, USA, 2009, pp. 135–140,
[37] G.-H. Tzeng, Y.-W. Chen, The optimal location of airport fire stations: a http://dx.doi.org/10.1145/1543834.1543854.
fuzzy multi-objective programming and revised genetic algorithm approach, [66] H. Pang, N. Liu, Q. Wu, Decision-making model for transportation and distri-
Transp. Plan. Technol. 23 (1999) 37–55. bution of emergency materials and its modified particle swarm optimization
[38] L. Yang, B.F. Jones, S.-H. Yang, A fuzzy multi-objective programming for opti- algorithm, Cont. Decis. 27 (2011) 871–874.
mization of fire station locations through genetic algorithms, Eur. J. Oper. Res. [67] A. Bozorgi-Amiri, M. Jabalameli, M. Alinaghian, M. Heydari, A modified
181 (2007) 903–915. particle swarm optimization for disaster relief logistics under uncertain envi-
[39] Y. Ma, Y. Tian, W. Hou, Dynamic path optimization of emergency trans- ronment, Int. J. Adv. Manuf. Technol. 60 (2012) 357–371.
port based on hierarchical genetic algorithm, in: Y. Tan, Y. Shi, K. [68] X. Gan, Y. Wang, Y. Yu, B. Niu, An emergency vehicle scheduling problem with
Tan (Eds.), Advances in Swarm Intelligence, vol. 6145 of Lecture Notes time utility based on particle swarm optimization, in: D.-S. Huang, K.-H. Jo,
in Computer Science, Springer Berlin Heidelberg, 2010, pp. 445–452, Y.-Q. Zhou, K. Han (Eds.), Intelligent Computing Theories and Technology, vol.
http://dx.doi.org/10.1007/978-3-642-13495-1 55. 7996 of Lecture Notes in Computer Science, Springer Berlin Heidelberg, 2013,
[40] Y.-H. Lin, R. Batta, P.A. Rogerson, A. Blatt, M. Flanigan, A logistics model for pp. 614–623, http://dx.doi.org/10.1007/978-3-642-39482-9 71.
emergency supply of critical items in the aftermath of a disaster, Socio-Econ. [69] J. Tian, W.-Z. Ma, Y.-L. Wang, Emergency supplies distributing and vehicle
Plan. Sci. 45 (2011) 132–145. routes programming based on particle swarm optimization, Syst. Eng. – The-
[41] J. Zhang, J. Peng, Z. Xu, W. Xu, SDVRP model for emergency logistics ory Pract. 31 (2011) 898–906.
and evolutionary heuristic approach, in: International Conference on Auto- [70] C. Wang, B. Yang, L. Li, H. Huang, A modified particle swarm optimization-
matic Control and Artificial Intelligence (ACAI 2012), 2012, pp. 1809–1812, based human behavior modeling for emergency evacuation simulation
http://dx.doi.org/10.1049/cp.2012.1342. system, in: International Conference on Information and Automation, 2008.
[42] M. Hamedi, A. Haghani, S. Yang, Reliable transportation of humanitarian sup- ICIA 2008, 2008, pp. 23–28, http://dx.doi.org/10.1109/ICINFA.2008.4607962.
plies in disaster response: model and heuristic, Proc. – Social Behav. Sci. 54 [71] Y. Zheng, H. Ling, J. Xue, S. Chen, Population classification in fire evacuation: A
(2012) 1205–1219. multiobjective particle swarm optimization approach, IEEE Trans. Evol. Com-
[43] L.-Y. Chung, T.-Y. Chou, C.-C. Lee, Multiobjective dynamic length genetic put. 18 (2014) 70–81.
algorithm to solve the emergency logistic problem, in: 2012 IEEE/ASME [72] J.M. Mulvey, R.J. Vanderbei, S.A. Zenios, Robust optimization of large-scale
International Conference on Advanced Intelligent Mechatronics, 2012, pp. systems, Oper. Res. 43 (1995) 264–281.
1147–1152, http://dx.doi.org/10.1109/AIM.2012.6266026. [73] Q. Zhu, J. Hu, W. Cai, L. Henschen, A new robot navigation algorithm for
[44] Y. Liu, X. Lai, G. Chang, Two-level integrated optimization system for planning dynamic unknown environments based on dynamic path re-computation and
of emergency evacuation, J. Transp. Eng. 132 (2006) 800–807. an improved scout ant algorithm, Appl. Soft Comput. 11 (2011) 4667–4676.
[45] R. Song, S. He, L. Zhang, Optimum transit operations during the emergency [74] K.K. Lim, Y.-S. Ong, M.H. Lim, X. Chen, A. Agarwal, Hybrid ant colony algo-
evacuations, J. Transp. Syst. Eng. Inform. Technol. 9 (2009) 154–160. rithms for path planning in sparse graphs, Soft Comput. 12 (2008) 981–994.
[46] M. Saadatseresht, A. Mansourian, M. Taleai, Evacuation planning using mul- [75] Y. Yuan, D. Wang, Path selection model and algorithm for emergency logistics
tiobjective evolutionary optimization approach, Eur. J. Oper. Res. 198 (2009) management, Comput. Ind. Eng. 56 (2009) 1081–1094.
305–314. [76] W. Yi, A. Kumar, Ant colony optimization for disaster relief operations, Transp.
[47] C. Ma, Y. Li, R. He, F. Wu, L. Sun, B. Qi, Multi-objective fuzzy optimisation Res. E: Logist. Transp. Rev. 43 (2007) 660–672.
method for emergency evacuation based on hybrid intelligence algorithm [77] S. Yan, Y.-L. Shih, An ant colony system-based hybrid algorithm for an emer-
under fuzzy environment, Int. J. Comput. Appl. Technol. 46 (2013) 228– gency roadway repair time-space network flow problem, Transportmetrica 8
235. (2012) 361–386.
[48] T. Sato, K. Ichii, Optimization of post-earthquake restoration of lifeline [78] D. Simon, Biogeography-based optimization, IEEE Trans. Evol. Comput. 12
networks using genetic algorithms, Jpn. Soc. Civil Eng. 53 (1996) 245–256. (2008) 702–713.
[49] Y.-W. Chen, G.-H. Tzeng, A fuzzy multi-objective model for reconstructing the [79] K.P. Iyer, V. Shanthi, Location based emergency response management system
post-quake road-network by genetic algorithm, Int. J. Fuzzy Syst. 1 (1999) through satellites in road networks, UACEE Int. J. Comput. Sci. Appl. 3 (2012)
85–95. 1–5.
[50] T.Z. Zhang, Y.M. Lu, Study on simulation and optimization of the road rush- [80] Y.-J. Zheng, H.-F. Ling, H.-H. Shi, H.-S. Chen, S.-Y. Chen, Emergency rail-
repair model after disaster, Appl. Mech. Mater. 50 (2011) 298–303. way wagon scheduling by hybrid biogeography-based optimization, Comput.
[51] F.-S. Chang, J.-S. Wu, C.-N. Lee, H.-C. Shen, Greedy-search-based multi- Oper. Res. 43 (2014) 1–8.
objective genetic algorithm for emergency logistics scheduling, Expert Syst. [81] Y.-J. Zheng, H.-F. Ling, X.-L. Xu, S.-Y. Chen, Emergency engineering rescue
Appl. 41 (2014) 2947–2956. scheduling and its application in disaster relief operations in China, Int. Trans.
[52] Y.-J. Zheng, H.-F. Ling, Emergency transportation planning in disaster relief Oper. Res. (2014) (in press).
supply chain management: a cooperative fuzzy optimization approach, Soft [82] Y.-J. Zheng, H.-F. Ling, J.-Y. Xue, Ecogeography-based optimization: enhancing
Comput. 17 (2013) 1301–1314. biogeography-based optimization with ecogeographic barriers and differen-
[53] G.F. Mott, Optimising flowshop scheduling through adaptive genetic algo- tiations, Comput. Oper. Res. 50 (2014) 115–127.
rithms, Chemistry Part II Thesis, Ph.D. thesis, Oxford University, Oxford, 1990. [83] J.E. Hunt, D.E. Cooke, Learning using an artificial immune system, J. Netw.
[54] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multiobjective Comput. Appl. 19 (1996) 189–212.
genetic algorithm: NSGA-II, IEEE Trans. Evol. Comput. 6 (2002) 182–197. [84] Z.-H. Hu, A network for emergency logistics management inspired by immune
[55] B. Liu, Uncertainty Theory, 2nd ed., Springer-Verlag, Berlin, 2007. multi-affinity model, in: International Conference on Information Manage-
[56] I. Rechenberg, Evolutions strategies: Optimiering technischer systeme nach ment, Innovation Management and Industrial Engineering, 2009, vol. 4, 2009,
prinzipien der biologischen evolution, Frommberg-Holzboog, Stuttgart, 1973. pp. 22–25, http://dx.doi.org/10.1109/ICIII.2009.467.
[57] L.J. Fogel, A.J. Owens, M.J. Walsh, Artificial Intelligence Through Simulated [85] Z.-H. Hu, A container multimodal transportation scheduling approach based
Evolution, John Wiley & Sons, Chichester, 1967. on immune affinity model for emergency relief, Expert Syst. Appl. 38 (2011)
[58] H.-G. Beyer, H.-P. Schwefel, Evolution strategies – a comprehensive introduc- 2632–2639.
tion, Nat. Comput. 1 (2002) 3–52. [86] X. Chen, G.-z. Tan, B. Jiang, Real-time optimal path planning for mobile robots
[59] H. Xie, L. Zhang, Using dual-gene variation hybrid evolution strategy to based on immune genetic algorithm, J. Cent. South Univ. (Sci. Technol.) 39
construct fire emergency decision-making model, in: 2nd International Con- (2008) 577–583.
ference on Advanced Computer Control (ICACC), 2010, vol. 5, 2010, pp. [87] L.-F. Liu, S.-Q. Zhang, Dual aircrafts path planning for forest-fire rescue by
103–106, http://dx.doi.org/10.1109/ICACC.2010.5487295. artificial immune algorithm, Adv. Mater. Res. 1659 (2012) 532–533.
[60] J. Barrachina, P. Garrido, M. Fogue, F.J. Martinez, J.-C. Cano, C.T. Calafate, [88] X. Zhang, Z. Zhang, Y. Zhang, D. Wei, Y. Deng, Route selection for emergency
P. Manzoni, Reducing emergency services arrival time by using vehicu- logistics management: a bio-inspired algorithm, Safety Sci. 54 (2013) 87–91.
lar communications and evolution strategies, Expert Syst. Appl. 41 (2014) [89] S.D.G. Kirkpatrick Jr., M.P. Vecchi, Optimization by simulated annealing, Sci-
1206–1217. ence 220 (1983) 671–680.
[61] Y. Wang, F. Jin, Research on improved evolutionary programming algorithm [90] A. Jahangiri, S. Afandizadeh, N. Kalantari, The optimization of traffic signal
for solving the problem of emergency resources scheduling, Oper. Res. Manag. timing for emergency evacuation using the simulated annealing algorithm,
Sci. 21 (2012) 29–33. Transport 26 (2011) 133–140.
[62] R. Storn, K. Price, Differential evolution – a simple and efficient heuristic [91] C. Ma, Y. Li, R. He, B. Qi, A. Diao, Research on location problem of emergency
for global optimization over continuous spaces, J. Global Optim. 11 (1997) service facilities based on genetic-simulated annealing algorithm, Int. J. Wirel.
341–359. Mobile Comput. 5 (2012) 206–211.
566 Y.-J. Zheng et al. / Applied Soft Computing 27 (2015) 553–566
[92] F. Glover, Tabu search – part I, ORSA J. Comput. 1 (1989) 190–206. [107] M. Becker, F. Blatt, H. Szczerbicka, A multi-agent flooding algorithm for
[93] F. Glover, Tabu search – part II, ORSA J. Comput. 2 (1990) 4–32. search and rescue operations in unknown terrain, in: M. Klusch, M. Thimm,
[94] C. Xie, M.A. Turnquist, Lane-based evacuation network optimization: an inte- M. Paprzycki (Eds.), Multiagent System Technologies, vol. 8076 of Lecture
grated Lagrangian relaxation and tabu search approach, Transp. Res. C: Emerg. Notes in Computer Science, Springer Berlin Heidelberg, 2013, pp. 19–28,
Technol. 19 (2011) 40–63. http://dx.doi.org/10.1007/978-3-642-40776-5 5.
[95] Y. Zheng, S. Chen, H. Ling, Efficient multi-objective tabu search for emergency [108] Y.-J. Zheng, S.-Y. Chen, Cooperative particle swarm optimization for multiob-
equipment maintenance scheduling in disaster rescue, Optim. Lett. 7 (2013) jective transportation planning, Appl. Intell. 39 (2013) 202–216.
89–100. [109] M. Clerc, Standard PSO 2007, 2010 http://www.particleswarm.info
[96] Z.W. Geem, J.H. Kim, G. Loganathan, A new heuristic optimization algorithm: [110] A.K. Qin, V.L. Huang, P. Suganthan, Differential evolution algorithm with strat-
harmony search, Simulation 76 (2001) 60–68. egy adaptation for global numerical optimization, IEEE Trans. Evol. Comput.
[97] I. Landa-Torres, D. Manjarres, S. Salcedo-Sanz, J.D. Ser, S. Gil-Lopez, A multi- 13 (2009) 398–417.
objective grouping harmony search algorithm for the optimal distribution of [111] H. Ma, D. Simon, Blended biogeography-based optimization for constrained
24-hour medical emergency units, Expert Syst. Appl. 40 (2013) 2343–2349. optimization, Eng. Appl. Artif. Intell. 24 (2011) 517–525.
[98] L. Ozdamar, W. Yi, Greedy neighborhood search for disaster relief and evacu- [112] M.M. Noel, A new gradient based particle swarm optimization algorithm
ation logistics, IEEE Intell. Syst. 23 (2008) 14–23. for accurate computation of global minimum, Appl. Soft Comput. 12 (2012)
[99] P.M. Duque, K. Sörensen, A grasp metaheuristic to improve accessibility after 353–359.
a disaster, OR Spectr. 33 (2011) 525–542. [113] M. Fischetti, P. Toth, D. Vigo, A branch-and-bound algorithm for the capac-
[100] J. Peng, W. Xu, J. Yang, A hybrid heuristic algorithm for large scale itated vehicle routing problem on directed graphs, Oper. Res. 42 (1994)
emergency logistics, in: Second International Conference on Intelligent 846–859.
Computation Technology and Automation, vol. 3, 2009, pp. 899–902, [114] D. Wolpert, W. Macready, No free lunch theorems for optimization, IEEE
http://dx.doi.org/10.1109/ICICTA.2009.682. Trans. Evol. Comput. 1 (1997) 67–82.
[101] L.-Y. Zhang, T. Fei, X. Zhang, Y. Li, J. Yi, Research about immune ant colony opti- [115] X.-S. Yang, S. Deb, Cuckoo search via lévy flights, in: World Congress on Nature
mization in emergency logistics transportation route choice, in: M. Zhao, J. Sha Biologically Inspired Computing, 2009, pp. 210–214, http://dx.doi.org/10.
(Eds.), Communications and Information Processing, volume 288 of Commu- 1109/NABIC.2009.5393690.
nications in Computer and Information Science, Springer Berlin Heidelberg, [116] U. Premaratne, J. Samarabandu, T. Sidhu, A new biologically inspired
2012, pp. 430–437, http://dx.doi.org/10.1007/978-3-642-31965-5 50. optimization algorithm, in: International Conference on Industrial
[102] F. Hu, W. Xu, X. Li, A modified particle swarm optimization algorithm for and Information Systems, 2009, pp. 279–284, http://dx.doi.org/10.
optimal allocation of earthquake emergency shelters, Int. J. Geogr. Inform. 1109/ICIINFS.2009.5429852.
Sci. 26 (2012) 1643–1666. [117] X.-S. Yang, A new metaheuristic bat-inspired algorithm, in: J.R. González, D.A.
[103] J. Zhang, T. Fei, The weak economy emergency logistics path optimiza- Pelta, C. Cruz, G. Terrazas, N. Krasnogor (Eds.), Nature Inspired Cooperative
tion algorithm based on fish swarm ant colony algorithm, in: J. Lei, F. Strategies for Optimization, vol. 284 of Studies in Computational Intelligence,
Wang, H. Deng, D. Miao (Eds.), Emerging Research in Artificial Intel- Springer Berlin Heidelberg, 2010, pp. 65–74, http://dx.doi.org/10.1007/
ligence and Computational Intelligence, Communications in Computer 978-3-642-12538-6 6.
and Information Science, Springer Berlin Heidelberg, 2012, pp. 350–356, [118] Y. Tan, Y. Zhu, Fireworks algorithm for optimization, in: Y. Tan, Y.
http://dx.doi.org/10.1007/978-3-642-34240-0 46. Shi, K. Tan (Eds.), Advances in Swarm Intelligence, vol. 6145 of Lect.
[104] X.-L. Li, Z.-J. Shao, J.-X. Qian, An optimizing method based on autonomous Note. Comput. Sci., Springer, 2010, pp. 355–364, http://dx.doi.org/10.1007/
animats: fish–swarm algorithm, Syst. Eng. Theory Pract. 11 (2002) 32–38. 978-3-642-13495-1 44.
[105] T. Fei, L.-y. Zhang, J. Zhang, Y. Li, X. Liu, C. Zhu, Research of emergency [119] P. Fleming, R. Purshouse, Evolutionary algorithms in control systems engi-
logistics distribution routing optimization based on simulated annealing neering: a survey, Cont. Eng. Pract. 10 (2002) 1223–1241.
ant colony optimization, in: J. Lei, F. Wang, H. Deng, D. Miao (Eds.), Arti- [120] H.-S. Shih, W.-Y. Lin, E. Lee, Group decision making for topsis, in: The Joint
ficial Intelligence and Computational Intelligence, vol. 7530 of Lecture 9th IFSA World Congress and 20th NAFIPS International Conference, 2001,
Notes in Computer Science, Springer Berlin Heidelberg, 2012, pp. 177–183, pp. 2712–2717, http://dx.doi.org/10.1109/NAFIPS.2001.943653.
http://dx.doi.org/10.1007/978-3-642-33478-8 23. [121] E. Burke, E. Hart, G. Kendall, J. Newall, P. Ross, S. Schulenburg, Hyper-
[106] N.A. Kallioras, N.D. Lagaros, M.G. Karlaftis, An improved harmony search heuristics: an emerging direction in modern research technology, in: F.
algorithm for emergency inspection scheduling, Eng. Optim. (2013), Glover, G. Kochenberger (Eds.), Handbook of Metaheuristics, Kluwer, New
http://dx.doi.org/10.1080/0305215X.2013.854351 (online). York, 2003, pp. 457–474.