Abstract
The industrial process is usually a lag inertial system. Predictive control is an effective control algorithm for this kind of system, but a more accurate object model is needed. In this paper, DMC predictive control algorithm is used, which does not need the specific form of the model, only needs the step excitation response data of the model. In this paper, the deep learning algorithm is applied to the modeling of industrial process system. After obtaining the more accurate step excitation response data, the predictive control can be carried out, and the ideal control quality can be obtained. First, the input and output data of the closed-loop system are obtained by adding pseudo-random sequence of appropriate period and amplitude into the control instruction of the closed-loop system. The first-order inertia and delay object are used to fit the characteristics of the object, and the first-order inertia time constant T is obtained by using genetic optimization algorithm. Secondly, a third-order inertial link and DNN deep learning network are embedded in the discrete structure of the third-order inertial model to build the intelligent model structure; In order to ensure that the third-order inertial link is close to the inertia time of the object, the inertia time constant of each link is set to t / 3, the input and output data are sent to the intelligent model for training, and the dnn1 model of the object can be obtained; After adding delay $\tau$ to dnn1 model, the genetic algorithm is used to fit the characteristics of the object, and the delay time $\tau$ is obtained; According to the input and output data, the DNN model with delay $\tau$ is trained for the second time to obtain a more accurate identification model dnn2. Thirdly, step excitation is applied to dnn2 model to obtain excitation response data, which is put into predictive controller to obtain excellent control quality. Finally, the first-order object model identified by the least square method is put into the predictive controller, and the control effect is compared with that of this paper. This method has great practical significance for the design and application of predictive control based on deep learning modeling.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.