A system for real-time delivery of wind farm production forecasts

Intermittency Management Systems (IMS) are expected to improve the quality of power produced by wind farms and thereby facilitate their integration into the power grids. Wind farm production forecasting is seen as a key enabling technology for Intermittency Management. Such a Wind farm Production Forecasting (WPF) system needs to interface and collaborate with multiple external systems for its operation. It requires inputs from meteorological systems, wind being the primary driver for power production on wind farms. It also needs real-time operational SCADA (Supervisory Control and Data Acquisition) data from wind turbines to monitor their states. The output of the WPF system can in-turn be potentially used by a diverse set of secondary systems, of which the IMS is one. Hence a forecasting system is expected (a) to implement the forecasting model, (b) to establish standards for interfacing with external systems, and (c) to establish mechanisms for making the production forecasts available to other systems or users. This paper presents a pattern, which we call the Input-Process-Output pattern, for designing a WPF system. The design was implemented in the form of the GE Wind Farm Production Forecasting (GE WPF) system which provides a site-specific short-term, one hour ahead to sixty hours ahead, forecasts of aggregate wind farm power. It comprises of a set of loosely-coupled, independently-executing modules. The External Interface Module.(Input) handles all communications with the external systems such as the Meteorological Forecasting agents and the SCADA systems. The Forecasting Engine (Process) implements the forecasting model and hence translates the meteorological forecasts to the wind production forecasts with the aid of SCADA data. The User Interface Module (Output) makes the output forecasts available to users in the form of an web interface. The Model-View-Controller paradigm was suitably leveraged for its implementation.