Research on Energy Management System in Electric Vehicles Based on Model Predictive Control Strategy

With the development and progress of science and technology, microgrids have attracted extensive attention from industry players because of their broad development prospects, and also the energy management system in microgrids has been rapidly developed. Due to the intermittent and fluctuating characteristics of distributed power in microgrids, it will lead to the instability of system power supply and distribution. In order to better solve this problem, a model predictive control (MPC) strategy based on parameter sharing mechanism is proposed. The traditional MPC produces large errors when dealing with sudden disturbances, which affects the stability and robustness of the whole system; and the traditional MPC is difficult to obtain real-time control results because of the long optimisation intervals, so it is difficult to design the feedback link, which results in the predicted values being far away from the actual values. To address the shortcomings of the traditional MPC, the preferred depth deterministic gradient algorithm with parameter sharing is proposed to be applied to the improved model predictive control strategy, which improves the accuracy of the calculation results and accelerates the calculation efficiency. Finally, MATLAB is applied for simulation to prove the accuracy of the proposed strategy.

I. Introduction

In the domain of microgrid energy management, contemporary research is actively exploring both facets of energy optimisation, encompassing not only deterministic optimisation techniques but also those tailored for managing uncertainty in the system dynamics. This ensures a comprehensive examination that spans across precise and predictable scenarios as well as stochastic and variable conditions inherent to microgrid operations. For deterministic optimization, at this stage, the main research in the literature [1]–[2] based on multi-timescale coordination optimization is an effective technical means of consuming intermittent distributed energy, which can solve the traditional based on the trend section information prone to unit regulation response is not timely, the plan tracking error is large and other problems, but there are also many drawbacks, for example, with the subdivision of the time scale, the dimensions of the problem and the number of variables to consider will increase significantly; there is also a strong dependence on information, resulting in a substantial increasein the computational resources required to solve the optimization problem; there is also a strong dependence on information, and uncertainty optimization is currently studied.For example, as the time scale is subdivided, the dimension of the problem and the number of variables to be considered will increase significantly, leading to a substantial increase in the computational resources required to solve the optimisation problem; there is also a strong dependence on the information, and the difficulty of model coupling. For uncertainty optimisation, the main studies at this stage include literature [3]–[5], which use robust optimisation theoretical framework to construct a robust optimisation model for the economic operation of microgrids, but the solutions of such models may tend to be conservative in coping with uncertainty, and the computational complexity is high, and there is also the problem of greater difficulty in convergence. On the other hand, literature [6]–[7] addresses the characteristics of the prediction error of renewable energy sources by treating them as random variables and generating a series of prediction error scenarios, and then builds a stochastic optimisation-based scheduling scheme. Furthermore, literature [8] proposes a multi-timescale energy optimisation dispatch model for microgrids, especially for the stochastic volatility of renewable energy sources such as wind power.The stochastic optimisation approach has a strong dependence on the probability distributions of the stochastic variables, and dealing with a large number of scenarios for selection and construction undoubtedly significantly increases the workload and challenge of the study.