On using Pareto optimality to tune a linear model predictive controller for wind turbines

Abstract

Optimal operation of wind turbines is important in order to minimize cost of energy, which is one of the major focus areas of the wind industry. Model predictive control (MPC) is a candidate for a control solution which effectively balances the different and potentially conflicting objectives, e.g. generated power and structural loads. This article presents a method on how to tune multi-objective MPC problems using Pareto curves. The approach is applied to a realistic wind turbine MPC problem, in which a joint power and tower fore-aft fatigue load optimization is performed. The controller is evaluated on a high fidelity model using a Vestas wind turbine simulator. In addition to the multiple control objectives, a number of constraints are considered as well. The evaluation shows a good potential of using model predictive control for this problem compared with an industrial baseline controller as, it approximately obtains the same mean generated power, while lowering the tower fore-aft fatigue loads. The computed Pareto curves of the trade-off between tower fore-aft fatigue load and mean generated power for a number of different weight matrices, demonstrate a potential tool for tuning MPC solutions for a wind turbine.