Estimation of Confidence-Interval for Yearly Electricity Load Consumption Based on Fuzzy Random Auto-Regression Model

Many models have been implemented in the energy sectors, especially in the electricity load consumption ranging from the statistical to the artificial intelligence models. However, most of these models do not consider the factors of uncertainty, the randomness and the probability of the time series data into the forecasting model. These factors give impact to the estimated model’s coefficients and also the forecasting accuracy. In this paper, the fuzzy random auto-regression model is suggested to solve three conditions above. The best confidence interval estimation and the forecasting accuracy are improved through adjusting of the left-right spreads of triangular fuzzy numbers. The yearly electricity load consumption of North-Taiwan from 1981 to 2000 are examined in evaluating the performance of three different left-right spreads of fuzzy random auto-regression models and some existing models, respectively. The result indicates that the smaller left-right spread of triangular fuzzy number provides the better forecast values if compared with based line models.