A Data-driven Approach for Wind Turbine Performance Bench-marking

摘要

Wind turbines consist of several components, sub-assemblies which are prone to fail. Failures in the wind turbines can be broadly categorized into: (1) soft failure, i.e. sensor malfunctioning and (2) hard failure, i.e. actual component failure. Large scale wind farms may contain over hundreds of such fault prone wind turbines. With limited resources, prioritizing the maintenance operations based on the severity and extent of failure is the key to ensure minimum power loss. The research developed here presents a data driven approach for benchmarking performance of utility scale wind farm. Performance of wind turbines is assessed based on turbine power output, and input wind speed. The statistical models are extracted from high frequency operational data which is further utilized to identify performance benchmarks for whole wind farm. Numerous fault and normal operational scenarios are assessed to ensure the applicability of the developed models. The data-driven models developed herein can ensure a continuous performance monitoring to be used for turbine fault prognosis, and maintenance management.