Industrial process minimum-maximum optimization fault-tolerant control method based on reinforcement learning

Abstract

The invention relates to the technical field of industrial control, in particular to an industrial process minimum-maximum optimization fault-tolerant control method based on reinforcement learning. Comprising the following steps: (1) establishing an augmented state space model containing tracking errors and state increments on the basis of an original system state space model with actuator faults and external disturbance, and proposing a performance index function according to the augmented state space model; (2) proposing a value function and a Q function according to the performance index function, and constructing corresponding expressions of optimal control input, worst external disturbance, optimal control gain and worst external disturbance gain; (3) giving initial control gain and external disturbance gain capable of stabilizing the system to collect data theta j (k) and rho kj, wherein theta j (k) and rho kj are data which are generated by jth iteration and contain system production information; (4) updating the control gain K1F and the external disturbance gain K2F through reinforcement learning; and (5) if an iteration ending condition is met, ending iteration, otherwise, returning to the step (4) to continue iteration. The method is wide in application range, good in tracking performance and good in control effect.

Description

technical field [0001] The invention relates to the technical field of industrial control, in particular to a fault-tolerant control method based on reinforcement learning for minimum-maximum optimization of industrial processes. Background technique [0002] The modern industrial process has undergone many changes with the improvement of the level of science and technology. More intelligent and efficient production processes, larger production scales, and more sophisticated and complex production equipment are gradually emerging. This also means that during production, industrial processes are more susceptible to failures or external disturbances, which weakens the ideal situation. The control effect of the designed control method even makes the control effect completely zombie. In this context, people no longer only focus on designing control methods for ideal conditions, which makes it possible to design robust controllers for the purpose of weakening the negative impact...

AI Technical Summary

Problems solved by technology

However, looking back at their development results, it can be found that most of the past control methods are model-based control methods, which have a great dependence on the system model, so once they leave the model, they will fall into trouble and cannot achieve the control goal. People began to search for new control methods, especially some new fault-tolerant control methods for industrial processes with both external disturbances and actuator failures

Images