Sensor Fault Tolerant Control For Multi-agent Systems Based On Deterministic Learning

With the development of social economy and the progress of science and technology,the internal structure of industrial system has become more and more complex,and the traditional control methods have been unable to effectively meet the expected requirements.Therefore,the concept of "multi-agent system" has been put forward and widely used in the field of control.Multi-agent system can realize message interaction and cooperate to complete tasks through network information channel,so as to improve work efficiency.However,due to its special extension and complement structure,the local fault of a single component may threaten the stability of the whole system,so fault tolerant control for multi-agent system has strong practical significance.In this paper,a sensor fault-tolerant control scheme based on deterministic learning is proposed for multi-agent systems with sensor faults.Firstly,by combining the high gain observer and the linear adaptive observer into the adaptive observer,the consistent boundness of fault signal and system state estimation is guaranteed.Then,a Radial Basis Function(RBF)neural network is used to identify system state information and fault parameters by combining the radial basis observer constructed above with the determinative learning theory.Moreover,the RBF neural network satisfies the Persistence of Excitation(PE)condition.The corresponding weight vectors of neural network nodes located near the system locus converge to an ideal state.Thus,local accurate identification of sensor faults can be realized.Finally,a pattern library is established for different sensor faults,and a group of estimators based on constant RBF neural network is constructed by using known fault information.Rapid fault diagnosis and data recovery can be realized without repeated online learning.The fault tolerant controller is constructed on the results of the estimator.Since the fault signal has been repaired,the standard controller can directly control the state of the system to track the reference trajectory stably,thus realizing the stable control of the fault tolerant controller to the system,and the design of this controller is simpler than the existing method.Finally,in order to verify the practical effect of the above fault-tolerant control scheme,this paper uses the interconnected inverted pendulum system as a model for simulation.The experimental results show that the system exhibits good tracking characteristics in the case of sensor failures,which further confirms the effectiveness of the fault-tolerant control scheme.