Enhancing the performance of wind turbine's O&M by employing multi-agent-systems

The consideration of several conditions e.g. weather conditions, wind power forecasts, electricity tariff, stock keeping etc. are crucial for wind farm operators to make optimal maintenance decisions. However, due to this enormous amount of information sophisticated tools are needed. The contribution will present the possible application of high-performance computing methodologies, which may help wind farm operators examining optimal maintenance strategies. The so called Multi-Agent-System (MAS) is a relatively new discipline in the world of Artificial Intelligence and Data Mining technology. Data Mining is a high-performance computing methodology used to observe and deduce hidden knowledge and logical dependencies of a great amount of data using several appropriate algorithms. This paper proposes a hybrid methodology combining MAS and Data Mining to help the wind farm operators optimizing his maintenance policies. Optimizing the maintenance policies surround many aspec ts dealing with cost reduction, time saving and profit maximizing. This could be achieved by e.g. suggesting the best time to proceed a preventive, condition-based maintenance etc., how to better manage the stock keeping by determining the ideal component number to be hold on. Other aspects to be considered within this contribution are the amelioration of personal scheduling on the sites according to their skills and experiences and deriving optimal bids under uncertainty in realized wind power and market prices.