data-driven control

This paper proposes an Enhanced Data-Driven Quantized Model-Free Adaptive Control (EDD-QMFAC) structure for a class of unknown nonlinear systems based on the Group Method of Data Handling (GMDH) neural network. In this study, the output quantized data is given to the GMDH block to overcome the data quantization challenges in Data-Driven Control methods. In the proposed control loop the GMDH derives a model to estimate the actual output of the system from the quantized output data based on the predictive feature of this network. The controller then generates the input control signal based on the system’s estimated actual output data. The Lyapunov theory is used to prove the stability of the suggested structure. The simulation results demonstrate the advantages of the proposed control structure over the conventional QMFAC.

In this paper, an extension for Data-Driven Model Predictive Control for Linear Parameter Varying systems is presented. Model-based controllers are highly dependent on model precision. On the other hand, data-driven methods are either replaced with the model or import its dynamic data into the design. Throughout this paper, direct data-driven approaches, which have gained considerable attention in recent years, are used in designing different parts of the controller, including future predictions. In addition, the stability and recursive feasibility guarantees are presented as the first novelty of this research with respect to a prior platform for data-driven approach. Furthermore, the base platform of direct DD-MPC for LPV systems is extended. The new developed form with the goal of robustness against measurement noise is defined as the next novelty of this paper. In order to check the performance of the proposed method, simulations on DC motor are applied. The results show effectiveness of the proposed approach as compared with similar approaches reported in the literature.

Over the past two decades, an increasing number of engineers and researchers in the field of control engineering have shifted their focus towards data-driven methods for system analysis and design. The defining characteristic of these data-driven approaches is their departure from conventional models and their associated assumptions. Instead, these methods harness the wealth of readily available, cost-effective, and reliable data derived from real, complex, and complex adaptive systems. Their primary objective is to facilitate system analysis and control solely through measured data, without relying on explicit or implicit model utilization. In this article, we commence by presenting an overview of the design principles embedded in model-based control systems, with a specific emphasis on adaptive and robust control system design methods. Subsequently, we delve into the fundamental principles of data-driven control system design. To comprehensively examine these data-driven methods, we categorize them into two groups: those grounded in machine learning and soft computing, and those based on traditional control systems analysis and design methods, often referred to as classical methods. Within this article, we initiate with a concise review of machine learning and soft computing-based methods before delving into a more comprehensive exploration of classical methods in this field.

The model-based controllers need a mathematical model of the system, whereas the data-driven controllers operate based on measuring the input-output data. Nowadays, considering complexity of the industrial systems and unavailability of an accurate mathematical model of the system, scientists try to reduce dependency of the controllers on the mathematical model. In this paper, an adaptive sliding-mode data-driven controller for a class of unknown multi-input multi-output nonlinear discrete-time systems is proposed. Because the chattering phenomenon is the main challenge of the sliding-mode controllers, an adaptive sliding-mode controller is used to solve this problem. In addition, to solve the dependencies of the controller on the mathematical model, the proposed adaptive sliding-mode controller is combined with a data-driven controller. Next, the new adaptive laws for the switching gain and the Pseudo Jacobian Matrix (PJM) are calculated. In addition, the closed-loop stability based on the Lyapunov theory is investigated. To show performance of the controller, the proposed method is applied to a three-tank system. The proposed controller has some advantages in comparison with similar methods in references, such as reducing the conservatism and complexity in the controller design and simplifying the closed-loop stability analysis. The simulated results show that the proposed method better tracks the reference signals and improves rejection of the external disturbances. Furthermore, the chattering phenomenon is considerably reduced.

Unfalsified adaptive control is a new approach in supervisory control that ensures the selection of a stabilizing controller from a control set based on the system input-output data. A prerequisite for ensuring stability is the existence of a pre-designed controller set that contains a stabilizing controller. The supervisor selects the controller based on the cost function calculated with the system input-output data. In this method, the control system performance is restricted to the controllers of the control set. In this paper, the controller set update is performed by introducing the concept of performance falsification along with the stability falsification of the active controller. To falsify the performance of the controller set, the structure of the model reference is proposed to evaluate the performance of the control system. In case of performance falsification, a new controller is designed and added to the controller set based on system data and without using any model. To design the controller, a linear matrix inequality problem is solved. In this paper, no system model is used, and the presented method is completely model-free and data-oriented. The simulation results show the performance improvement of the proposed method compared to other methods in a standard robust adaptive benchmark system.

Nowadays, industrial systems deal with a wide range of constraints. Output saturation and lack of system model are two types of these constraints. In this paper, a Data-Driven Adaptive Predictive Control (DDAPC) is propounded for a family of unknown non-linear systems featuring output saturation. The design of the control signal only dependent on the input and output data of the system. In this regard, a new adaptive predictive control scheme is suggested using the new developed dynamic linearization model. The stability analysis of the proposed method is provided by proving the boundedness of the tracking error for both time varying and constant desired reference signal and considering the output saturation data, which is a common physical constraint in industrial systems. Furthermore, the proposed method is more robust against the model uncertainties and nonlinearities, in comparison with the common model-based adaptive methods, since its controller design procedures as well as the stability analysis are independent of plant model. To verify the efficiency and applicability of the suggested approach some applicational and numerical simulation examples are provided.

A model free terminal sliding mode controller has been proposed. The proposed controller is data driven i.e. based only on input and output data (thus, model free). This method employs a recursive nonlinear sliding surface. This leads to higher tracking precision and to finite time convergence of the response. The method uses “Model Free Adaptive Controller” approach, combined with “Discrete time Sliding Mode Controller” method. Boundedness of tracking error is proved analytically. Theoretical analysis also shows the superiority of the proposed method over Model Free Linear Sliding Mode Controller. Analysis results are confirmed via simulation