The Research And Application Of Exponential Smoothing Methods Based On The Artificial Intelligence Optimization

Exponential smoothing describes a class of forecasting methods which are still the most popular forecasting methods used in business and industry, and state space model can underlie all of the exponential smoothing methods. Common features of this class of methods is that the point forecast represents a weighted moving average of all past observations with the weights decreasing exponentially-the older the observations are,the lower the weights are. Hence we name them "exponential smoothing".There are some gaps and deficiencies of the study of exponential smoothing methods in China, such as the lack of integrated study and study for their statistical frameworks. According to these shortcomings, theories of exponential smoothing methods are studied systematically, including categories, point forecasts and the state space models of them. Then, some other methods are introduced into the models to improve the prediction accuracy. They are the time series aggregation theory, combination forecasting theory and artificial intelligence algorithms such as particle swarm optimization, cuckoo algorithms and artificial fish swarm algorithm. Finally, the actual power load data in New South Wales, Australia are used to test the effectiveness of the proposed combination forecasting model based on exponential smoothing methods.Results of case study show that for short-term power load forecasting, combination forecasting model proposed in this paper is superior to the traditional forecasting methods, especially after the optimization by particle swarm optimization algorithm or cuckoo algorithm. Therefore, the proposed combination forecasting model in this paper is effective and our study is practical and significant to a large extent.