摘要
针对传统PID控制器对非线性时变系统实时性差、抗干扰能力弱,且控制参数人工整定较为困难的问题,研究利用神经网络自学习、抗干扰能力强的特性以及任意逼近非线性函数的能力,在硬件层面实现神经网络控制。利用高性能嵌入式系统芯片作为神经网络计算核心,设计搭建了自整定PID逆变电源控制系统和电路。通过实验验证了该神经网络优化的PID控制系统在两种误差指标上分别降低了15.2%和67.8%,并实现了对期望输出更好的跟踪效果,提高了控制系统的自适应能力,实现了智能算法和功能。
In order to solve the problems of the traditional PID controller, such as poor real-time performance and weak anti-interference ability for the nonlinear time-varying system, and the difficulty of adjusting control parameters manually, this paper studies the use of the characteristics of neural network self-learning, strong anti-interference ability and the ability to approach the nonlinear function arbitrarily, realizes neural network control at hardware layer. The high-performance embedded chip is used as the neural network calculation core, and a self-tuning PID inverter power supply control system and circuit are designed and built. Experiments show that the PID control system optimized by the neural network reduces two error indexes by 15.2% and 67.8% respectively. It also achieves a better tracking effect on the expected output, improves the adaptive ability of the control system, and realizes the intelligent algorithm and function.
作者
马立新
季乐延
朱勇杰
MA Li-xin;JI Le-yan;ZHU yong-jie(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《控制工程》
CSCD
北大核心
2021年第3期579-584,共6页
Control Engineering of China
基金
上海市研究生创新基金项目(JWCXSL1302)。
关键词
神经网络
嵌入式系统
自整定PID
逆变电源
Neural network
embedded system
self-tuning PID
inverter power supply
