基于改进差分进化算法-径向基神经网络的电热水浴串级控制系统研究

Electric heated water bath cascade control system research based on improved differential evolution algorithm-radial basis function neural network

针对电热水浴装置温度控制中被控对象存在的大惯性、非线性、大延迟等特点,设计了一种基于改进差分进化(improved differential evolution,IDE)算法的径向基(radial basis function,RBF)神经网络串级控制系统。采用IDE算法对RBF神经网络的初始参数进行优化,采用优化后的RBF神经网络辨识主控制回路被控对象的Jacobian信息,进而实现主控制回路PID(proportional integration differentiation)控制器参数的在线调整。针对主控制回路控制器包含输出噪声,导致控制性能下降的问题,引入Kalman滤波器对串级控制的主回路进行重新设计,控制对象的输出值经过Kalman滤波算法处理后再返回闭环控制系统。以微化工领域常用电热水浴装置为对象,对IDE-RBF-PID-PI串级控制系统进行仿真实验,结果表明,IDE-RBF-PID-PI串级控制系统相较于常规串级控制,大大提高了控制性能,主控制回路引入的Kalman滤波算法有效消减控制系统的输出噪声,控制效果接近于无噪声的理想状态。

Aiming at the large inertia,nonlinearity and large delay of the controlled object in the temperature control of the electric heated water bath device,RBF(radial basis function)neural network cascade control system based on the IDE(improved differential evolution)algorithm is designed.The IDE algorithm is used to optimize the initial parameters of the RBF neural network.The optimized RBF neural network is used to identify the Jacobian information of the controlled object of the main control loop,and then the online adjustment of the parameters of the main control loop PID(proportional integration differentiation)controller is realized.Aiming at the problem that the main control loop controller contains output noise which leads to the decline of control performance,the Kalman filter is introduced to redesign the main loop of the cascade control.The output value of the control object is processed by the Kalman filter algorithm and then returned to the closed loop control system.The simulation test of the IDE-RBF-PID-PI cascade control system is carried out for the common electric heated water bath device in the micro chemical industry.The results show that the IDE-RBF-PID-PI cascade control system greatly improves control performance compared to conventional cascade control.The Kalman filtering algorithm introduced in the main control loop effectively reduces the output noise of the control system,and the control effect is close to the ideal state without noise.