Multi-objective Optimization Operationof Micro Grid Based On Wavelet Transform And PSO-SBL Prediction

The technology of micro-grid is an effective approach to promote the development and utilization of renewable energy. Wind generation’s randomness has a great influence on operation characteristic of a micro-grid. The key method to reduce the unfavorable influence for a microgrid with wind turbines due to the randomness of wind energy lies in the fast and accurate prediction of wind power. This paper studied the wind power prediction, load forecasting based on the wavelet transform and particle swarm optimizing sparse bayesian learning regression model and the multi-objective optimal operation problems of micro grid system.The first step, wind turbines, photovoltaic cells, micro gas turbine and the fuel cell of micro power grid with the working principle and operation characteristics are analyzed, on the basis of above, the modeling of each unit are set up. Then the working principle of all kinds of energy storage system is analyzed and hybrid energy storage model are set up.The second step, at a huge advantage in data mining based on bayesian theory, this paper proposing a particle swarm optimized sparse bayesian learning model based on the wavelet transform. With reference to the historical data by using time series analysis method revealing the dynamic structure of system and law of Mallat algorithm through the discrete wavelet transform to signal decomposition. In the training phase, This paper puts the straton sequence modeling and uses particle swarm optimization algorithm to find the optimal parameter set, then forecasts each layer of the decomposed time subsequence and reconstructs the result.The third step, particle swarm optimized sparse bayesian learning model based on the wavelet transform to forecast the wind speed and load are used. This paper uses the least squares polynomial fitting power curve to get its power. Aiming at the deficiency of deterministic prediction method, the prediction is introduced into the monte carlo method based on Latin hypercube sampling, giving under a certain confidence level of confidence interval. Through the measured data and analysis results show that using this method can improve the prediction precision, which has certain practical.The fourth step, both micro grid running economy and environmental protection of the multi-objective optimization model are established. The improved multi-objective bacterial colony chemotaxis optimization algorithm is applied to solve the micro grid multi-objective optimal operation model solving. The TOPSIS method based on grey correlation calculating decision solution are used to avoid the blindness of decision makers to choose optimal solution. The optimization of the different schemes is compared with the results of the analysis show that the proposed optimization model is effective and the algorithm is reasonable.