小波去噪结合ARMA模型在电力设备故障率预测中的应用

Failure Rate Prediction Method Based on Wavelet Correlation Denoising Combined with ARMA Model

针对电力设备故障率具有周期性、随机性和多变性等特点,提出小波相关性去噪算法与时间序列自回归滑动平均(ARMA)模型的电力设备故障率预测方法.将电力设备故障率数据进行小波相关性去噪,最大限度保留有效序列,把重构后的序列进行ARMA建模及预测,预测值与实际值进行比较.仿真结果表明,小波相关性去噪后的ARMA模型预测结果有较高的精度,实际故障率预测效果较好.

To solve the problem of periodicity,randomness and variability of electric power equipment failure rate,a wavelet correlation denoising algorithm and a time series autoregressive moving average (ARMA) model for power equipment failure rate prediction are proposed.The power equipment failure rate data were denoised by wavelet correlation,and the effective sequence was retained to the maximum.The reconstructed sequence was modeled and predicted by ARMA,and the predicted value was compared with the actual value.The simulation results show that the ARMA model with wavelet correlation denoising has higher accuracy and the actual failure rate prediction is better.