自构建小波神经网络的内模控制在深度脱硅中的应用

摘要

To stabilise thermal regulation of alumina deep desilication process and reduce energy consumption,we use internal model control method identified by wavelet neural network for process control of alumina deep desilication process.The hidden layer neurons in wavelet neural network can be added or deleted according to the excitation intensity and attenuation degree of wavelet basis function,so that the structure of hidden layer in wavelet neural network is optimised.Then the self-constructing wavelet neural network is employed to identify the forward model and inverse model of internal model control system,thereby the technology of internal model control is improved.Experimental results show that the proposed control method has better performance than the traditional control methods in aspects of robustness and immunity,every index of the algorithm is better than that of the traditional control methods,hence the optimisation of deep desilication process is achieved.%为了稳定氧化铝深度脱硅过程的热工制度和降低能耗，采用小波神经网络辨识的内模控制方法进行氧化铝深度脱硅工艺过程控制。根据小波基函数的激励强度和衰减程度可以添加或者删除小波神经网络隐含层神经元，从而优化小波神经网络隐含层结构。再用自构建小波神经网络辨识内模控制系统的正模型和逆模型，从而改进神经网络内模控制技术。实验结果表明，所提出的控制方法比传统方法在鲁棒性和抗扰性方面具有更好的性能表现，各项指标均优于传统控制方法，实现了氧化铝深度脱硅工艺优化。