The Accuracy of Renewable Energy Generation Forecasting in Energy Systems: Quality Improvement Practices

Russia is one of the countries that pay increasingly close attention to the development of renewable energy. It designs and implements special government programmes aiming to diversify its energy sources and meet international environmental standards. There is an active programme since 2013 aiming to further incentivise owners of renewable energy generators on the internal wholesale and retail markets, which stipulates additional payments to be received by participants using RES. In 2030, the total proportion of solar and wind power plants in the power system’s installed capacity structure should reach 4.2% (11.5 GW); by 2042, the total installed capacity of SPPs and WPPs may exceed 20 GW. [1]

According to key strategic planning documents for the development of the Russian power industry till 2035, special attention is given to forecasting on the basis of statistical data analysis and introduction of Artificial intelligence (AI)-based technology in the power sector. A target indicator is the establishment of a reliable (trusted) forecasting system for providing highly accurate generation forecasts and improving planning efficiency for power system operating parameters.

Successful integration of renewable energy generation forecasting systems in existing business processes requires high accuracy of generation forecast outputs for all planning horizons. Accurate forecasts are also necessary to ensure the reliable operation of the power system especially in Southern regions of Russia where the share of renewable energy exceeds level of tens of percent of total generation.

A key factor that substantially affects the accuracy of generation forecasts is telemetry data acquired from supervisory control and data acquisition (SCADA) systems and equipment status data. The quality of telemetry data depends on the validity, accuracy and reliability of information provided by sensors, software, machines and other sources at power facilities. All telemetry data received from objects is initially validated by automated SCADA tools.

Key challenges inherent in this problem include regular noise and errors in data arising from technical problems in equipment, failures of communications channels and false readings of sensors (e.g. due to their contamination or malfunction). As part of activities to improve the quality of RES forecasts, Russian specialists systematized the errors, developed approaches to data processing and introduced methodological foundations that consider the importance of interactions between process participants: from plant personnel to end users of forecasting systems. This allows for the use of renewable energy generation forecasts in light of the nature of solar and wind power facilities, as well as control, data acquisition and transmission systems.