模糊神经网络的真空木材碳化设备控制系统设计与仿真

Design and Simulation of Fuzzy Neural Network Controller with BP Network

针对目前真空木材碳化设备的控制系统具有大滞后、强耦合、时变性以及难以建立精确数学模型等特点,提出了一种模糊神经网络算法的真空木材碳化设备控制系统。通过对输入输出变量、论域及隶属函数的选择,设计出真空木材碳化设备控制器;再将神经网络与模糊控制系统相结合,得到模糊神经控制网络。对模糊神经网络控制器的算法进行了分析;在Matlab环境下编写控制器的程序,用Simulink进行仿真实验。结果表明:模糊神经网络控制器的真空木材碳化设备输出的温度曲线,稳态误差为0、最大偏移量为1℃、调节时间约为8 s、超调量为2%;湿度曲线在6 s时即可达到稳定,稳态误差为0、最大偏差为1%、超调量为4%;加入扰动后,误差能被快速消除,温湿度的波动幅度相对减小,系统的稳定性更强。模糊神经网络控制器,可减小调节时间、消除误差、提高控制精度,具有很好的鲁棒性。将二者结合设计出的模糊神经网络控制器,具有自适应、学习、识别和模糊信息处理等功能,在处理大规模复杂的模糊应用问题方面具有更好的控制效果。

In view of the fact that the control system of vacuum wood carbonization equipment has the characteristics of large lag, strong coupling, time variability and difficuh to establish accurate mathematical model, a control system of vacuum wood carbonization equipment based on fuzzy neural network algorithm was proposed. By selecting input and output variables, the choice of domain and membership function, the vacuum solid wood carbonization equipment controller was designed. Then, the neural network and the fuzzy control system were combined to obtain the fuzzy neural control network. The algo- rithm of fuzzy neural network controller was analyzed. The program of controller was written in Matlab, and Simulink was used to simulate the experiment. The results show that the temperature curve of the vacuum wood carbonization equipment of the fuzzy neural network controller is 0, the maximum offset is 1 ℃, the adjustment time is 8 s, the overshoot is 2%, and the humidity curve is 6 s. The steady-state error is 0, the maximum deviation is 1%, and the overshoot is 4%. After adding the disturbance, the error can be eliminated quickly, the fluctuation range of temperature and humidity is relatively reduced, and the stability of the system is stronger. Fuzzy neural network controller can reduce the adjustment time, eliminate the error, and improve the control accuracy with good robustness. The fuzzy neural network controller, which combines above two, has the functions of adaptive, learning, recognition and fuzzy information processing, and has better control effect in dealing with large and complex fuzzy application problems.