基于数据分析的污水监测采样点数据研究

Research on that Data of Sewage Monitor Sampling Point Based on Data Analysis

本文针对性地设计了一个常规的三层感知机作为神经网络的预测模型,并采用与感知机相同输入、相同隐层的受限玻尔兹曼机作为特征提取器,对数据深层次特征进行提取,并将特征提取后得到的权值矩阵移植到神经网络中进行初始化,最后利用反向传播算法(BP)进行第一层与第二层之间的微调和第二层和第三层之间的训练,随机取指标的4/5作为训练集。最终,模型拟合精度达到0.90以上,MSE为0.0849,并且建立基于灰色理论的预测模型进行对比分析。采用这两种预警模型对未来污水监测采样点30天的变化进行预测,并根据预测结果分析两种模型的优缺点及推广可行性。

In this paper,a conventional three-layer perceptron is designed as the neural network prediction model,and the restricted Boltzmann machine with the same input and the same hidden layer as the perceptron is used as the feature extractor to extract the deep-seated features of the data,and the weight matrix obtained after the feature extraction is transplanted to the neural network for initialization.Finally,the back propagation algorithm(BP)is used to fi ne tune the fi rst layer and the second layer and train the second layer and the third layer,randomly take 4/5 of the index as the training set.Finally,the fi tting accuracy of the model reached above 0.90,and the MSE was 0.0849.And the prediction model based on grey theory is established for comparative analysis.The two epidemic warning models are used to predict the 30 day change of sewage monitoring sampling points in the future,according to the prediction results,the advantages and disadvantages of the two models are analyzed.