Fractional data driven controller based on adaptive neural network optimizer

Given the nonlinearity and complexity inherent in the dynamics of industrial systems, model-based control methods are deemed inadequate for addressing the evolving needs of industries, often resulting in high levels of complexity in control system design. Presently, in response to industrial requirements, data-driven control methods are undergoing significant development and progress. This paper introduces a novel method, namely Fractional Order Data-Driven Sliding Mode Control (FODDSMC), based on a newly proposed Fractional Order Sliding Mode Observer (FOSMO), specifically designed for nonlinear systems. All control parameters in the proposed controller and the novel FOSMO are optimized using Artificial Neural Networks (ANN). By virtue of harnessing the advantages of fractional calculus, the proposed method is capable of enhancing system robustness against external disturbances. This paper aims to propose a novel control method in the design of nonlinear controllers capable of addressing the challenges and enhancing the performance of the system. In this context, the performance of the proposed control system is evaluated under normal conditions and in the presence of internal uncertainty, noise and external disturbances applied to the nonlinear system to demonstrate the benefits of the proposed approach. Furthermore, the performance of the proposed method is compared against that of sliding mode controllers and optimal sliding mode controllers to underscore its superiority. The proposed method has significant performance improvement compared to the discussed sliding mode controllers. To evaluate the performance of the proposed method and validate its superiority, the performance of the proposed method has been examined under two different case studies. The proposed method has demonstrated suitable performance under ideal conditions, external disturbances, and system uncertainties, with the proposed observer providing accurate estimates. Also the proposed adaptive neural networks ensure the optimality of controller’s coefficient. In both examples, the superiority of the proposed method compared to the other methods considered for comparison is clearly observable.