Active fault diagnosis method for space manipulator

Abstract

The invention discloses an active fault diagnosis method for a space manipulator. An active fault diagnosis framework is constructed through active input based on constraint reinforcement learning (CRL) and a fault isolation method based on a maximum mean difference (MMD) index; a constraint reinforcement learning algorithm is introduced into the input of the modeled space manipulator system, so that the input can maximize the output difference degree of each fault mode and realize fault separation on the premise of ensuring the fault-tolerant control (FTC) performance; and under the excitation of the active input, through a fault isolation method based on a maximum mean value difference index under a data driving framework, according to input and output signals of an analysis system, error modes are eliminated so as to determine which fault mode the current system belongs to. According to the method, two targets of active fault diagnosis and fault-tolerant control performance can be taken into consideration at the same time, fault diagnosis is carried out on the system in a safe mode, and the problems of modeling difficulty, model uncertainty influence, high calculation complexity and the like can be effectively solved.

Description

technical field [0001] The invention relates to the field of fault diagnosis, in particular to an active fault diagnosis method for a space manipulator. Background technique [0002] With the development of social economy and the advancement of science and technology, modern control systems show a trend of intelligence and complexity. During the operation of the system, the occurrence of failure is inevitable, if not dealt with in time, it will lead to property damage and casualties. Therefore, fault diagnosis technology is the key technology to improve system safety, reliability and reduce accident risk. [0003] The vast majority of existing fault diagnosis methods are carried out under the "passive" framework, that is, only the input and output data of the system are used to generate residual signals used to indicate faults, and no additional input signals are applied to the system, so that the evolution of the system is not affected. rules have an impact. Such fault d...

AI Technical Summary

Problems solved by technology

[0006] 1. In the existing model-based AFD methods, there are difficulties in modeling accurate models and uncertain factors;

[0007] 2. Most of the existing methods solve the optimal input through optimization, and the computational complexity is too high, so the method does not meet the real-time performance, it is difficult to apply and the computational cost is too high;

[0008] 3. Existing AFD methods lack consideration of fault tolerance performance during diagnosis, active input may damage the system, or cannot effectively balance the proportion of the two optimization objectives of AFD and FTC in the optimization problem

Images