Nonlinear system self-adaptive neural fault-tolerant control method

Abstract

The invention discloses a nonlinear system self-adaptive neural fault-tolerant control method. A multi-input multi-output nonlinear system is established, and an intermittent fault model is adopted toobtain an actuator state; a multi-input single-output extreme learning machine neural network is adopted to approach an uncertainty function in the multi-input multi-output system; then an inequalityis introduced to avoid the singularity problem of the controller; and finally the bounded performance of the estimated parameters is guaranteed by adopting a projection operator. According to the method, the uncertain function of the system is approached by adopting the extreme learning machine, so that the dependence on system prior knowledge is reduced; a continuous inequation is introduced, sothat the singularity of the controller is avoided, and meanwhile, the bounded performance of the estimated parameters is guaranteed by adopting the projection operator; the boundary of jump parameters caused by the fault of the intermittent actuator is determined clearly, so that the influence of an unknown actuator intermittent fault and an unknown correlation item on the system can be effectively compensated.

Description

technical field [0001] The invention belongs to the technical field of fault-tolerant control of control systems, and in particular relates to a nonlinear system adaptive neural fault-tolerant control method. Background technique [0002] During the operation of the control system, the actuator, as its core component, often suffers from various unpredictable failures. If these faults cannot be compensated effectively in time, it may lead to system performance degradation or instability. Additionally, as modern control system systems become more complex, the likelihood of system actuator failures increases. How to improve the safety and reliability of the system has become the most important subject in the field of control engineering. Therefore, inventing an effective fault-tolerant control method to compensate the impact of faults on the system has important engineering application value. [0003] Actuator failures are intractable due to the complete unknown of timing, s...

AI Technical Summary

Problems solved by technology

However, the hidden layer parameter selection of traditional neural networks often depends on the prior knowledge of the objective function, which becomes very complicated in actual engineering and may weaken the fault-tolerant control method’s ability to compensate for faults.

Images