Research On Model Predictive Control Method Based On Multi-energy Complementary Distributed Energy System

Today,the contradiction between economic development and environmental pollution is increasingly prominent.The distributed energy technology with clean energy and renewable energy as the core is an important development direction for the future world to study the energy problem,and is the key to solve the energy problem in China.In this paper,we build a multi energy complementary distributed energy system which combines wind energy,cogeneration and energy storage.Based on the model predictive control method,the energy in the distributed energy system is scheduled to configure an optimal energy supply scheme for the system.This paper focuses on the specific work of multi energy complementary distributed energy system as follows:(1)Firstly,the output characteristics of each module in the distributed energy system are analyzed,including wind power generation electronics,micro gas turbine and battery module,and the power output model of each module is established.(2)Then,according to the chaotic characteristics of the wind power system,a long short-term memory network prediction model based on phase space reconstruction is proposed.The mutual information method in the time series correlation method is used to select the delay time,so as to reduce the correlation of each element in the phase space.At the same time,the GP algorithm in the correlation dimension saturation method is used to calculate the embedding dimension.The validity of the model is verified by the analysis of the examples and the comparison of the results of different models.(3)After that analyze the reliability and economy of system operation,take the minimum power interaction with the power grid and the operating cost of the distributed energy system as the optimization objective function,and use the internal load balancing of the system and the output of each module as constraints.At the same time,the problems existing in the standard particle swarm optimization algorithm are analyzed,combined with the adaptive weighting method and the Bayesian optimization algorithm,an improved particle swarm optimization algorithm-Bayesian-AWPSO is proposed.Finally,an example analysis verifies that the improved particle swarm optimization algorithm has higher convergence and stability.(4)Finally,the MPC algorithm is introduced,and the MPC optimization strategy of distributed energy system is established by integrating the above two parts.This paper discusses the performance of MPC under two kinds of optimization indexes(minimum interaction with power grid and minimum operation cost),and analyzes the two cases.The data results show that the model predictive control is effective in the system.