Abstract: For a class of linear constant systems, the problem of continuous gain type fault diagnosis and reliable control of a single component of an actuator is investigated based on the trapezoidal region. Firstly, in order to solve the problem that the pole information of closed-loop system is difficult to observe, a design scheme of full-dimensional state observer is given to realize the real-time observation of pole information and form a pole classification database for system failure. Secondly, according to the characteristics that the poles are located in different areas when different channels have faults, support vector machine is applied to design a pole classifier to diagnose faults in the system and
achieve accurate and reliable control of the system based on the fault diagnosis results. Then, in order to solve the problem of difficulty in selecting parameters for support vector machines, the bat algorithm (BA) is proposed to achieve automatic parameter optimization, which has the advantages of strong robustness and easy to combine with
other methods.

Key words: Pole classifier, Support vector machine, Bat algorithm, Regional pole assignment, Fault diagnosis