Professional Documents
Culture Documents
2010 - Active Disturbance Rejection Control For Non-Minimum Phase Systems
2010 - Active Disturbance Rejection Control For Non-Minimum Phase Systems
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/241172824
CITATIONS READS
5 112
2 AUTHORS:
Abstract: Active disturbance rejection control (ADRC) has been applied to solve various types of control problems across many
engineering disciplines, but largely within the confine of minimum phase systems. This paper, however, explores systematically
its applications to non-minimum phase (NMP) systems, particularly those with transfer functions that have right half plane zeros.
It is first shown that, a regular ADRC controller, if not tuned carefully, could easily yield an undesirable solution for NMP
systems. We then demonstrate the reasons behind the difficulty using transfer function analysis, leading to a solution in how we
tune the ADRC for NMP systems. In particular, a systematic design and tuning procedure is obtained based on the relationship
between the high frequency gain of the system and the controller bandwidth. Finally, the proposed method allows a user to easily
make a tradeoff between the undershoot, which is unique in NMP systems, and the response time, using the frequency response
method of gain and phase compensation.
Key Words: Active disturbance rejection control, Non-minimum phase system, Frequency response method
−2 Reference 1.2
Response
−3 1
−4
0.8
−5 Reference
0.6
Response
−6
0 1 2 3 4 5 6 7 8
0.4
Time (sec)
0.2
Fig. 1: Simulation results for high frequency gain formula-
tion 0
−0.2
As one can see, the transient part of the response is almost 0 20 40 60 80 100
Time (sec)
perfect without any undershoot. However, when the system
goes near to the steady state it becomes unstable. An intu- Fig. 2: Simulation results for low frequency gain formula-
itive explanation of this is that since b̂0 is negative, for the tion
high frequency part the system is a negative feedback which
explains why the transient response is very good. On the As shown above, the two design formulations of ADRC do
other hand, for the low frequency part the system is a pos- not fit the non-minimum phase problem quite well. More
itive feedback which leads to an unexpected result. More researches based on frequency domain analysis are carried
detailed analysis will be conducted in Section 3 to give more out in the next section to find the reasons behind it and the
insights to this problem. applicable approaches.
Fig. 3: Diagram of the ADRC structure with state feedback
30 1.5
25
1
o
20
ω
15 Reference
0.5 b0=100, ωo=50
10
b0=100, ωo=30
5
0
0
0 20 40 60 80 100 120 140 160 180 200
b0
−0.5