Advanced predictive analytics for solar power generation

Solar power generation have been gaining ground as a result of improved generating efficiency, reduced installation cost as well as a global focus towards renewable energy. However solar power generation still faces a number of limitations that prevents it from being used on a larger scale. One solution to the problem is an accurate forecast of electricity load demand. With an accurate forecast, wastage of energy will be prevented and this is critical to the stability of the power system. In this final year project, the main objective is to study the viability of using various techniques to forecast electrical load to aid solar power generation in Singapore. First, literature review was conducted on the subject. Two techniques, the Auto Regressive Integrated Moving Average (ARIMA) and Multi-Layer Perceptron (MLP), were chosen to forecast half-hourly electrical load in Singapore. The two techniques were then developed and tested on real load data of Singapore’s electric utility. The test results displayed that the MLP technique is better suited for an electrical load forecasting application. The forecasting errors were smaller than with an ARIMA model as MLP takes into account weather factors and human’s energy consumption habits. The work suggests that an on-line testing of the model is required before an opinion on its applicability can be formed.Bachelor of Engineerin