乌鲁瓦提大坝监测数据去噪统计模型研究

Study on the Statistical Model for Denoising the Monitoring Data of Wuluwati Dam

针对大坝监测数据中含有高斯白噪声,单一去噪方法存在精度低、结合传统统计模型效果差等问题,为了实现高斯白噪声的滤除,甄选出去噪精度更高的方法,提出小波和卡尔曼滤波结合的方法对含高斯白噪声数据进行预处理,经过小波去噪处理后再结合卡尔曼滤波的组合模型能够有效滤除原始信号中的高斯白噪声和粗差。用去噪后的数据建立统计回归模型进行预测,用后10期数据进行实测值和预测值对比,结果显示:去噪处理后建立的统计回归模型预测的绝对误差平均(MAE)和均方根误差(RMSE)明显降低,小波和卡尔曼结合的去噪模型能够有效滤除高斯白噪声,结合统计模型,提高了预测的精度和可靠性。

Aiming at the problems of Gaussian white noise contained in dam monitoring data,low precision of single denoising method and poor effect in combining with traditional statistical model,in order to filter out Gaussian white noise and select a method with higher denoising accuracy,a method combining wavelet and Kalman filter is proposed to preprocess the data containing Gaussian white noise.The results show that after wavelet denoising,the combined model with Kalman filter can effectively filter the Gaussian white noise and gross errors in the original signal.Use the denoised data to establish a statistical regression model for prediction.The last 10 periods of data is applied to compare the measured and predicted values.The results show the average absolute error(MAE)and root mean square error(RMSE)predicted by the statistical regression model after denoising is significantly reduced.It is concluded that the denoising model combined with wavelet and Kalman filter can effectively filter out Gaussian white noise,and the statistical model improves the accuracy and reliability of prediction.