Research On MID-Long Term Power Load Forecasting Considering Economic Rebalancing

Since China entered the “13th Five-Year”,China's economic development shows three characteristics: speed change,structure optimization and power conversion.the speed of economic growth has changing from high to mid-high.Economic development structure is transformed from production increasing and energy consumption to storage adjustment and increment optimization,driving power from resource consumption and low-cost labour to innovation driven.These changes indicate that China's economic development model has begun to rebalancing.With the development of smart grid,the load growth trend in China has changed in context of economic rebalancing,the amount of data in power load increases significantly.Predicting the power load accurately by using many related factors and dealing with a large amount of load data effectively have become a challenge currently.Starting from economic and meteorological factors,the relationship between them and load is studied,and the power load in a practical power grid are predicted base on the relationship.Firstly,the current research state of mid-long term power load forecasting and big data technology in power load forecasting are introduced.Then,according to the development of economy and power load in this practical power grid,a gray-cointegration analysis method is proposed considering economic rebalancing,and the main economic indexes are selected.Finally,combined forecasting model based on multi attribute decision making algorithm is proposed,and power load is forecasted by main economic indexes.The case study result showed the effectiveness and feasibility of the method.The concept and technology of big data are introduced to cope with large amounts of power load data.A multi-dimensional sample set is established with big data feature.According to meteorological effect on the load fluctuation,meteorological factors are measured and scored by the variable important measure based on out of bag data error rate.Random forest regression algorithm is submitted to big data platform built by SJTU,and power load is forecasted by distributed structure of big data platform.The results prove this method can be used to quantitatively analyze the importance of a large amount of meteorological factors to power load,and the effectiveness of analyzing and predicting the massive load data,compared with the support vector machine and relevance vector machine.