College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China.
National Innovation Center for Digital Fishery, China Agricultural University, Beijing 100083, China.
Math Biosci Eng. 2020 Dec 4;18(1):386-399. doi: 10.3934/mbe.2021021.
The present research envisaged the analysis of the dissolved oxygen fault of the water quality monitoring system using the genetic algorithm-support vector machine (GA-SVM). The real-time data collected by the dissolved oxygen sensor was classified into the fault types. The fault types were divided into complete failure fault, impact fault, and constant output fault. Based on the fault classification of the dissolved oxygen parameters, SVM fault diagnosis experiments were conducted. Experimental results show that the accuracy of dissolved oxygen was 98.53%. On comparison with the experimental results of the back propagation (BP) neural network, it was found that the diagnosis results of the dissolved oxygen parameters using SVM were better than those of the BP neural network. The genetic algorithm (GA) was used to optimize the parameters. After iteration, the optimal parameters such as C and g were selected (C is the penalty coefficient, which adjusts the weight of the two index preferences in the optimization direction, i.e., the tolerance for errors, and g is a parameter that comes with the function that implicitly determines the distribution of the data after mapping to the new feature space.). By using GA, after iteration, the optimized values of C and g was found to be 2.1649 and 5.3312, respectively. The experimental results showed that the method exhibited a good accuracy.
本研究旨在利用遗传算法-支持向量机（GA-SVM）分析水质监测系统中溶解氧的故障。利用溶解氧传感器采集的实时数据对故障类型进行分类，将故障类型分为完全失效故障、冲击故障和恒输出故障。基于溶解氧参数的故障分类，进行了 SVM 故障诊断实验。实验结果表明，溶解氧的准确率为 98.53%。与反向传播（BP）神经网络的实验结果进行比较，发现使用 SVM 对溶解氧参数的诊断结果优于 BP 神经网络。利用遗传算法（GA）对参数进行优化，经过迭代，选择出最优的参数 C 和 g（C 是惩罚系数，它调整了优化方向上两个指标偏好的权重，即对误差的容忍度，g 是一个与隐含地确定映射到新特征空间后数据分布的函数相关的参数。）。通过使用 GA，经过迭代，发现优化后的 C 和 g 的值分别为 2.1649 和 5.3312。实验结果表明，该方法具有较好的准确性。
