Study on performance evaluation and prediction of mixed-flow hydraulic turbine units under variable operating conditions

Abstract

Accurate performance evaluation and prediction is the basis for the safe operation and condition-based maintenance of the mixed-flow hydraulic turbine units. In process of evaluation, the monitoring data are highly relative to operating conditions and have a degree of randomness, making it difficult to construct healthy state models. Besides, in the process of prediction, the randomness margins of performance indicator prediction results are rarely quantized. To solve these problems, a new approach for performance evaluation and prediction of mixed-flow hydraulic turbine units is proposed. Firstly, to construct the healthy state model, the Gaussian mixture model is adopted to fit the three-dimensional probability density distribution of water head, power and lower bracket vibration. Then, the negative log-likelihood probability is calculated as the performance degradation indicator. Finally, the rolling prediction model based on Gaussian process regression is established to predict the performance degradation trend. Different kernel functions are compared and discussed. The engineering application verifies that the proposed healthy state model is less affected by the randomness of the data, and the prediction model effectively predicts the performance degradation indicators and the corresponding confidence intervals well.