Improvement and automation of tools for short term wind power forecasting

We present the results from an on-going project financed by the Danish PSO-fund where a number of subjects relevant for further automation and improvement of short term wind power forecasts methods are studied. The technological basis of the project is adaptive forecast methods as the methods forming the basis of WPPT (Wind Power Prediction Tool) – a well proven system for wind power forecasting. In the project we investigate (i) initialization of the self-calibrating models, (ii) automatic selection of tuning parameters, (iii) robust estimation, and (iv) combination of several meteorological forecasts. In the paper we present why these aspects are considered important for further development of forecast systems. Following this we present the investigations performed and outline the solutions which follow from these investigations.