ABSTRACT
This chapter explains a generalized bad data identification method for the raw data of numerical weather prediction (NWP) aiming to improve the performance of wind power output forecast. It describes the NWP data adjustment to short-term wind power forecasting. The uncertainty and variability of weather is a significant factor leading to unpredictable renewable generation. The wind power forecast engine forms the basis of the whole forecasting system. Different from the wind power forecast engine, the input of the data adjustment engine is the features extracted from the abnormal detection engine. The used network structure of artificial neural network is identical with the one in the wind power forecast engine, an input layer with 3 neurons, a hidden layer with 4 neurons, and an output layer with one neuron. The chapter presents case study which is carried out using the data of one wind farm from ‘Global Energy Forecasting Competition 2012 - Wind Forecasting’ to validate the proposed method.
