Approach to controlling the temperature of drying kiln using feedback-error-learning based adaptive inverse control

In this paper, an approach was presented for controling the temperature of the wood drying kiln. In this approach, the Feedback-Error-Learning(FEL) based adaptive inverse control was adopted by integrating the traditional feedback controller PID with a Nonlinear Auto-Regressive Exogenous Input (NARX) neural network. The system was stabilized by the PID, while NARX was used as a dynamic inverse feed-forward controller, which involves the output of the PID to realize adaptive online. Consequently, the adjustment of the control parameters was adaptively conducted by the input and output online information, which was applied to learn the parameter transformation and unmodeled dynamics of the system. Furthermore, the pre-training of the neural network was not needed when using this method. Experiments were carried out on homemade JBGZ-1.8 experimental drying kiln. The experimental results show that control precision was improved to 0.5℃ compared with the control error of 3℃ by only PID method, and the rising time and overshoot distortion were also reduced. This illustrates the theoretical and practical merit of this approach.