Font Size: a A A

Application Of Neural Networks Based On Hybrid Particle Swarm Optimization In Short-Term Load Forecasting

Posted on:2019-06-09Degree:MasterType:Thesis
Country:ChinaCandidate:Y T JiangFull Text:PDF
GTID:2382330551456680Subject:Engineering
Abstract/Summary:PDF Full Text Request
The accuracy of short-term load forecasting directly affects the development of the power system and plays an important role in improving the economical efficiency and safety of the power system.The accuracy of the traditional load forecasting algorithm is not ideal.Based on the summarization of relevant studies at home and abroad,this paper combines the hybrid particle swarm optimization algorithm with neural network to apply to short-term load forecasting and obtains the expected results.Firstly,because BP neural network has advantages in nonlinear data processing,it is used as a basic model of short-term load forecasting model.For the problem of slow training speed and high requirement on initial value of BP neural network,this paper combines particle swarm optimization with BP neural network and uses it to perform predictive simulation.Compared with the traditional BP neural network model,the prediction effect has been greatly improved.What's more,in order to further improve the prediction accuracy of the prediction model,this paper uses the ant colony algorithm to optimize the particle swarm optimization algorithm for the problem of particle swarm algorithm falling into local minimum,improves the particle optimization process,improves the global convergence of the algorithm,and establishes GPSO-BP neural network prediction model and short-term load forecasting simulation.The three prediction models of BP neural network,PSO-BP neural network and GPSO-BP neural network were used to simulate.The results showed that the prediction accuracy of GPSO-BP neural network prediction model was the highest.Finally,a load forecasting model considering the influencing factors is proposed.A variety of external factors affecting the load change are analyzed and the GPSO-BP neural network model considering the influencing factors is proposed,and a variety of influencing factors and historical load data are used as the input of the neural network to perform predictive simulation.Forecast accuracy reaches 99.31%.For the complex network topology and long training time,a PCA-GPSO-BP model is proposed to improve the prediction efficiency.The principal component analysis of multiple inputs of the neural network reduces the data dimension,simplifying the topology structure of the network,thereby improving the training efficiency of the prediction model.Simulation results show that the prediction accuracy of PCA-GPSO-BP neural network model is 99.03%,which is 0.28%lower than that of GPS O-BP neural network model,but the operating efficiency is improved by 75.4%.Taking into account the accuracy and efficiency,the PCA-GPSO-BP neural network model is even better.
Keywords/Search Tags:Short-term load forecasting, BP neural network, Particle swarm optimization, Ant colony algorithm, Principal component analysis
PDF Full Text Request
Related items