Research On Load Forecasting Model Based On Improved Deep Learning Algorithm

In power system,the application of prediction technology in time and place is an important means to ensure power quality.In the short term,it can provide a basis for power system scheduling.In the lon g run,it can provide the basis for the future planning and development of power system.However,with the increasing complexity of the power system,the traditional prediction algorithm has been difficult to achieve satisfactory prediction accuracy or spe ed.Therefore,it is more and more important to find an accurate and efficient prediction algorithm.Moreover,in recent years,with the increase of grid connected clean energy power generation,it increases the difficulty of power dispatching.Taking wind energy as an example,because the wind speed is difficult to control,the wind power output data has strong randomness.Accurately predict its output,and then formulate the corresponding scheduling strategy,can improve the controllability of wind power.Based on the above discussion,this paper takes short-term prediction as an example.Firstly,aiming at the problem of low accuracy of existing prediction algorithms,a short-term prediction method based on Adaptive Mutation Fruit Fly Optimization Algorithm(AMFOA)and improved Convolutional Neural Network(CNN)is proposed;Secondly,aiming at the problem of unnecessary resource waste caused by the uncertainty of wind farm output,a power system scheduling method with wind farm based on prediction algorit hm is proposed.This paper briefly introduces the basic theory and defects of CNN and FOA.This leads to the CNN and FOA improvement strategy.Then,the improved forecasting model is applied to the actual short-term load forecasting.In this process,the load factors and data preprocessing methods are introduced and analyzed.Finally,the paper introduces the defects of the power system dispatching model with wind power,and leads to the improvement strategy and the establishment of the final dispatching model.The improved scheduling model is verified by an example.