Intelligent Regression Load Forecasting for Non-Uniform Urban Load

A well-studied load forecast is usually perceived as an essential parameter for the cost control of any utility. Not only, but also, load forecasting has a great impact on creating an energy profile for utilities. The energy profile helps to assume the amount of renewable energy interference needed to sustain the electricity consumption from conventional generation. Maintaining a balanced and economic profile of an urban consuming utility can be very challenging when considering any external influencing factors. This study finds the optimum forecast using artificial intelligence for a commercial mall load data in the state of Arizona among different regression models within linear regression models, regression trees, Gaussian process regression models, support vector machines and ensembles of regression trees models to train the data supplied. According to the model used in this paper, the study provides forecasts along with their root mean squared errors in medium-term load forecasting, considering temperature, humidity, time, and seasonal data. This study compares forecasting errors after clustering a yearly hourly load profile into a summer season with three parts of the day. The results states that fine regression trees, quadratic support vector machines and rational gaussian process regression models resulted in the least root mean square errors for load forecasting.