Research And Application Of Artificial Fish Swarm Neural Network In Short Term Load Forecasting

The accuracy of power system short-term load forecasting values directly affects the power grid planning and electric power generation,transmission,distribution,use and other links.Therefore,finding an effective and practical high precision prediction method is of great significance for the study of short-term load forecasting.In recent years,methods of short-term load forecasting emerge in an endless stream,but the artificial neural network plays an important role.Elman neural network is a typical dynamic network,and undertaking layer can store the feedback information to make the system adapt to the time-varying characteristics,so it can meet the short-term load forecasting requirements.But the Elman neural network still has some algorithm inherent defects,so this paper uses artificial fish swarm algorithm to optimize Elman neural network.In this paper,the learning algorithm of Elman neural network is optimized by nonlinear damped least square method,and the parameters of the artificial fish swarm algorithm is dynamically adjusted,so the optimization stage can quickly obtain the global optimal solution domain,and the late local search improves the accuracy of the optimal solution.Combining the advantages and disadvantages of artificial fish swarm algorithm and Elman neural network,the improved artificial fish swarm algorithm is used to find a set of optimal initial weights and threshold of neural network so as to establish the artificial fish swarm neural network short-term load forecasting model.Selecting the appropriate sample data analyzes load characteristics and influencing factors,and then determine the model input.The simulation prediction of the model is implemented by Matlab software,and at the same time using prediction model of the traditional Elman neural network prediction model and artificial fish swarm neural network without considering the influencing factor compare the predicted load results on the same day.The result shows that the prediction model of artificial fish swarm neural network that considers influencing factor is better than the other two kinds of prediction model,because the prediction precision is improved.The model predicts respectively load of working days,non-working day and holiday,and forecasting results show that the model has reached the prediction precision for different day types and among them the best forecast is the working day.