Model Design Of Power Demand Forecast Based On System Dynamics

In the area of forecasts of electricity demand there are many ways, can be divided into traditional forecasting methods and modern forecasting methods, but they have some limitations。For example, traditional forecasting methods only model and forecast in math on historical data or the development trend ,which theoretically have a lot of supportive, but did not consider impact on electricity demand factors, also have no effectiveness . Although the modern method of prediction theory takes into account the factors that affect demand for electricity, but the reflection in the model is fixed and can not make adjustments and change by the surrounding circumstances, that is to say the factors influencing electricity demand is not known. For these reasons, in this paper the model of electricity demand is method which used is the system dynamics. In considering the impact of demand factors, by studying a large number of related articles, we find some impact on electricity demand factors, including economic growth, industrial restructuring, energy policy adjustment, demographic factors, climatic factors influence prices, the impact of scientific and technological progress, etc. These factors can be classified by the deep analysis of them, for example, can be divided into the impact of government and socio-economic environment, natural factors and the non-natural factors. After the set of classification and hierarchical relationships then can be applied AHP to weight the various factors affecting the determination of ordinary people can rely on the experience of various factors to determine the importance of order, then the mathematical derivation to science calculate the weight of different factors.Each will have its influence factors influence, some degree of influence factors is determined based on people's subjective sense, for example, the impact of science and technology, energy policy adjustments, etc.; and the influence factors is very large, for example, GDP, growth in the number of population, climate temperature, etc., so the data need to be standardized, mapping them in the values between 0 and 1.In the application of system dynamics theory to model, the time of the community or region is divided into the first industrial electricity consumption, electricity consumption of secondary industry, tertiary industry and living with power consumption. Among the three industrial electricity consumption, electricity consumption for each industry is equal to the value of electricity intensity of the industry and the product of the industry's GDP, while GDP in the industry by a state or local government intended target, are government behavior, so only need to build the three industries of electricity intensity value model; living in the forecast when the power consumption, considered approach is living with population number and per capita electricity consumption of the product to predict living with power at the same time, population is the population growth rate by the population base and determined, these data are readily available, so we only need to build living per capita electricity consumption modelSimulated in the model run with the validation phase, first use Jilin Province in 1998 -2008's real data as validation data, combined with Jilin Province, the specific circumstances of the past few years to determine the impact of policy factors, such as science technology development, the government's investment, etc., by applying data normalization method on the number of additional data processing, mapping them in the values between 0 and 1, and then made them into graph, numerical impact the vertical axis of the curve, this is because the characteristics of system dynamics, a factor only requires a change in trend can be. At the same time, AHP to determine weights of different factors; in classification after the influencing factors, such as natural and unnatural factors affecting factors, governmental factors and socio-economic environmental factors, etc., can be determined the specific weight by an expert make points system. This model will use the variables which have been identified, the model can simulate the operation of the established system dynamics model, weight values were entered into the computer, applications Stella simulation software, while the post-processed Jilin data from the model input into the converter, obtained by the software running after the operation results, and Jilin real information comparing the predicted results can be found in error within 8%, due to long-term electricity demand forecast accuracy less than 10% is reasonable and feasible, so it can be said that the model is reasonable and feasible.Finally, this paper established the model for sensitivity analysis, which mainly has two objectives, first, what wants to know the established model has the ordinary characteristics, second, it wants to sensitivity analysis, this found that some factors in the effect of individual then put forward some policy recommendations. Specific approach is to make the same number of factors to adjust the ratio, the ratio of the number used in this paper is 5%, 10%, 15%, 20% of these four levels, since the purpose of this process sensitivity analysis, thus it Jilin Province in 2007 the historical data, a normal simulation, the initial prediction by the model, then choose one of the factors as a percentage of adjusted target, no adjustment of other factors, after 5%, 10%, 15 %, 20%, which is four times the simulation is running, get a different operating results, and then some results of this comparative analysis reveals that factors affecting the demand for electricity situation, applying the same approach, these factors for the same treatment, so that various factors will be the impact on electricity demand situation. Through sensitivity analysis, the impact of different factors that can make relevant policy recommendations, such as improved energy-saving technology in the accumulation of a certain amount by 10%, the order to play its role in the future will be played and the value of the role played by electricity intensity will bi reduced.The model in this paper on the accuracy and application are of practical value, in the future what we need to do is to model the impact of these factors gradually refined so that it will improve the prediction accuracy.