The Research And Application Of The Combined Forecasting Model Based On Data Preprocessing And Artificial Intelligence

The development of information technology and the popularization of the Internet make information data increase rapidly.How to deal with the huge amount of data effectively and dig more and more useful information has attracted global extensive attention.So nowadays,the most effective way to conduct the increasing data is to forecast data.Forecasting is an effective method to reveal the nature and rule of things in certain scientific and reasonable way.It will clearly describe the future development trends of the forecasted things by the way of data analyzing.By means of the forecasting method,enterprises and other people can formulate reasonable projects which will help avoid a loss effectively.Especially in the electric system,an accurate electric forecasting method can not only save production cost,but also bring a lot of economic benefits to society.In the meanwhile,it will help environment protection and build a green,low carbon production and living place.Electricity,as a kind of clean energy,has been widely used in people's production and life.It's a key lifeline in the development of national economy.However,affected by the climate,environment,population and other factors,electric energy is hard to store.Due to this matter,electric power departments are difficult to estimate the electricity production in advance.Thus may give rise to the shortage of energy or energy dissipation.Unstable fluctuation energy production at the same time will affect the electricity price fluctuations.What's more,it will influence the residents and enterprises to a certain degree.With the wide application of electric power,an accurate,precise and strong applicability forecasting method is of great importance in the energy system.Based on data pre-processing and artificial intelligence optimization algorithm,a combined forecasting method was introduced in this paper.The proposed method contains five individual methods,namely wavelet transform(WT),extreme learning machine(ELM),phase space reconstruction(PSR),least squares support vector machine(LSSVM)and particle swarm optimization(PSO).Each individual method has its own usage.Before the experiment started,the threshold processing of WT is adopted to reduce noise from the original data.Because no single method is suit for all kinds of data,three individual methods(ELM,PSR and LSSVM)are applied to obtain the intermediate forecasting results.In the end,PSO is employed to attain the best proportion of each single method of the combined model.By multiplying the corresponding optimal weights by all the three forecasting results and then adding them up,the final result of the combined method can be got.In order to verify the forecasting performance of the proposed combined method,in this paper,two kinds of simulation experiments were carried out: electric load forecasting and electricity price forecasting.Austrian Energy Market(AME)is an electric market which operates electricity of five regions.The electric data of New South Wales(NSW)and Victoria(VIC)are chosen as case studies.The data used in the state of NSW are aimed at validating the effectiveness of the proposed combined model,while the state of VIC using the same time period predominantly focuses of feasibility and extensive applicability.In the meanwhile,the proposed combined method was compared with some of the individual methods and the already existing combined methods.The experiment results turn out that the combined method with high forecasting accuracy outperforms the individual methods and is appropriate for different kinds of data.