Regional power generation prediction model and post-evaluation based on coal production and economy status

With the deepening of the reform of the electricity market and rapid development of power industry, advancement of economy and power generation is becoming more and more closely related to each other. Balance between supply and demand is always one of the core issues of regional electricity market research. This paper proposes an electricity generation prediction model in power plant combining statistical methods and neural network in order to improve predictive performance. On the basis of power generation influential factors analysis, first we adopt several statistical methods including ADF test, Granger Causality Model, Error Correction Model and Co-integration Test to investigate the relationship between several variables such as GDP, industrial structure, technological progress, cola production, even the temperature and so on. This step aims at the discovery of possible causes that lead to fluctuations of power generation. Subsequently, an ingenious predictive model using artificial intelligence method, a neural network, is created in numerical prediction of future amount of electricity generation. At last, efficiency results of both operation scale and technology are delivered by resorting to Data Envelopment Analysis (DEA) as post-evaluation. All the data are collected from website of National Bureau of Statistics, with time series in month up to 15 years (1998~2013). This study is of great significance to estimate the status of future electricity generation, and is of huge support to facilitate future intensification and optimization of resource allocation, which will further enhance operational efficiency.