基于混沌时间序列的水电机组状态短期预测

Short-term Prediction of Hydroturbine Generating Unit Condition Based on Chaotic Time Series

基于混沌时间序列短期可以预测的特点,构建水电机组状态短期预测。用采样周期确定相空间时延τ,G-P算法确定关联维数从而确定相空间的嵌入维数m,小数据量法证明水电机组振动状态的混沌特性。在重构相空间中,运用加权一阶局域法构建水电机组状态短期预测模型。结果表明:混沌特性指数λ=0.2605的水电机组振动状态具有混沌特性,可以在最佳嵌入维数m=4的情况下进行预测,实例结果表明采用混沌理论进行水电机组状态短期预测是可行的。

Based on the characteristic of chaotic time series,a model was built to predict hydroturbine generating unit condition.The time delayτ was determined by sampling period,and the embedding dimension m was chosen according to correlation dimension,which was calculated by G-P algorithm.Chaotic characteristic of vibration signal series of hydroturbine generating unit was proved by small data sets arithmetic.The prediction model of hydroturbine generating unit condition was constructed by an adding-weight one-rank local-region method after the phase space was reconstructed.The results show that vibration signal series has a chaotic characteristic while the chaotic property exponent λ=0.2605.Therefore,a prediction model can be carried out while the best embedding dimension m is 4.The results demonstrate that the prediction method is feasible.