Lyapunov-Based Dropout Deep Neural Network (Lb-DDNN) Controller
This is an incremental improvement for adaptive control systems in robotics or automation, addressing specific issues like overfitting with dropout regularization.
The paper tackles the problem of overfitting and co-adaptation in deep neural network-based adaptive controllers for nonlinear dynamic systems by developing a dropout DNN controller with Lyapunov-based weight adaptation, resulting in a 38.32% improvement in tracking error, 53.67% improvement in function approximation error, and 50.44% lower control effort compared to a baseline.
Deep neural network (DNN)-based adaptive controllers can be used to compensate for unstructured uncertainties in nonlinear dynamic systems. However, DNNs are also very susceptible to overfitting and co-adaptation. Dropout regularization is an approach where nodes are randomly dropped during training to alleviate issues such as overfitting and co-adaptation. In this paper, a dropout DNN-based adaptive controller is developed. The developed dropout technique allows the deactivation of weights that are stochastically selected for each individual layer within the DNN. Simultaneously, a Lyapunov-based real-time weight adaptation law is introduced to update the weights of all layers of the DNN for online unsupervised learning. A non-smooth Lyapunov-based stability analysis is performed to ensure asymptotic convergence of the tracking error. Simulation results of the developed dropout DNN-based adaptive controller indicate a 38.32% improvement in the tracking error, a 53.67% improvement in the function approximation error, and 50.44% lower control effort when compared to a baseline adaptive DNN-based controller without dropout regularization.