Next-Generation Intelligent Electric Drive Control Using Deep Neural Network-Based Adaptive Learning

Abstract

The electric drive system is used extensively in industrial automation, electric car schemes, robotics and renewable applications because they are highly efficient and can accurately be controlled. But the traditional control techniques including proportional-integral (PI) controllers tend to be ineffective when the system is non-linear, when there is a change in parameters and external shocks. To address these shortcomings, this paper suggests a 2 nd -generation adaptive learning-driven intelligent control architecture of electronic drives, using the framework of deep neural network (DNN)-driven adaptive learning. The proposed strategy will combine a multilayer deep neural network controller with an adaptive learning control mechanism which will keep on updating control parameters based on continual feedbacks of the drive system. The neural network acquires the nonlinear relationship between the motor state and the control inputs, making it capable of correctly tracking the speed and a better disturbance rejection. An electric drive system simulation model was formulated in detail to determine the efficiency of the suggested control strategy. The results of the DNN-based adaptive controller were compared to those of a traditional PI controller in different load and operating environments. According to simulation findings, the given method has a pronounced positive effect on the dynamic response features such as lower rise time, lower overshoot, decreased settling time, and increased steadiness. Moreover, the adaptive learning property enhances the robustness of controller to the uncertainty in parameter and load disturbances. The suggested intelligent control system is a credible and effective option in enhancing the next generation electric drive systems to work in dynamic and highly challenging environments.