BP neural network PID control method for sparrow search algorithm optimization

Abstract

The invention discloses a BP (Back Propagation) neural network PID (Proportion Integration Differentiation) control method optimized by a sparrow search algorithm, which solves the problems of low learning efficiency, low convergence speed, easiness in falling into local optimum and the like of an existing neural network control system. According to the technical scheme, the method is characterized in that the number of initialized populations in a sparrow search algorithm is determined, a fitness function is set, an initial fitness value is calculated, a topological structure of a BP neural network is determined, the positions of a discoverer, a follower and an early warning person are updated, fitness is calculated, an optimal value is updated, and whether conditions are met or not is judged; if yes, endowing the obtained value to the BP neural network as an initial weight and a threshold value, and if not, performing iteration until the condition is met; bP neural network training is carried out by using the optimal weight, whether errors meet requirements is calculated and judged, and if yes, corresponding PID parameters are calculated; according to the method, the optimal PID parameters can be obtained, the convergence speed is high, the stability is good, and the control effect can be more obvious.

Description

technical field [0001] The invention relates to a PID controller parameter optimization algorithm, in particular to a BP neural network PID control method optimized by a sparrow search algorithm. Background technique [0002] The PID controller does not depend on the mathematical model of the controlled object, and has the advantages of good stability, easy implementation, and strong anti-interference ability. It is the most widely used and mature controller in the power conversion system. But the conventional PID controller has been unable to satisfy the control of the switching power supply that requires high performance. In this case, the intelligent optimization algorithm is introduced into the PID control system to tune the PID parameters. The intelligent optimization algorithm has the ability to fully and arbitrarily approximate any complex nonlinear relationship, has strong information synthesis ability, and can learn It is introduced into the PID controller design a...

AI Technical Summary

Problems solved by technology

However, in its practical application process, the BP neural network PID control system designed by using BP neural network has low control effect due to the low learning efficiency, slow convergence speed, weak global search ability and easy to fall into local optimum of the backpropagation learning algorithm. Not ideal, which limits the application of neural networks in PID controllers

Images