基于动态均值感知器的拍门污水泄漏精准检测系统  

Precise Detection System for Sewage Leakage in Pneumatic Flap Doors Based on Dynamic Mean Perceptron Neural Network

作　　者：姜楠 封莉[1] 林子茵 毕海涵 JIANG Nan;FENG Li;LIN Zi-yin;BI Hai-han(School of Environmental Science and Engineering,Beijing Forestry University,Beijing 100083,China;School of Electric Power,South China University of Technology,Guangzhou 510641,China;School of Information,Beijing Forestry University,Beijing 100083,China)

机构地区：[1]北京林业大学环境科学与工程学院,北京100083 [2]华南理工大学电力学院,广东广州510641 [3]北京林业大学信息学院,北京100083

出　　处：《中国给水排水》2024年第17期118-122,共5页China Water & Wastewater

摘　　要：采用均值检测原理与感知器神经网络分类原理,基于动态均值感知器神经网络分类模型,研发了排水管网拍门污水泄漏精准检测系统。首先,以固有拍门临界工作气压为中心逐次对称增压/减压20 kPa,并连续9次检测拍门位置状态,组成9维动态特征向量;然后与9维均值权重向量加权求和,提取动态均值特征量;最后,通过阶跃激活函数,采用输出阈值分类检测拍门污水泄漏状态,并训练输出阈值为3.6。在广州市猎德涌进行了现场测试,拍门污水泄漏检测精度为96%。该方法有效解决了误检和漏检问题,从而减少了排水管网的拍门污水泄漏。Based on the principles of mean detection and perceptron neural network classification,a dynamic mean perceptron neural network classification model was used to develop a precise detection system for sewage leakage in pneumatic flap doors of drainage network.Firstly,pressure was symmetrically increased/decreased by 20 kPa with the inherent critical working pressure P as the center,and the position and state of the flap door 9 times was continuously detected to form a 9-dimensional dynamic feature vector.Then,by weighting and summing with the 9-dimensional mean weight vector,the dynamic mean feature quantity was extracted.Finally,by using a step activation function,the output threshold classification was used to detect the leakage status of the sewage in the flap door,and the output threshold was trained to be 3.6.The on-site application system test of Liede Chong in Guangzhou City showed that the accuracy of the sewage leakage detection of the flap door was 96%.This method effectively solves the problems of the random and systematic errors of detection results,thereby reducing the leakage of sewage in the drainage network.

关 键 词：排水管网 拍门 污水泄漏 随机误差 均值检测 神经网络 感知器 

分 类 号：TU992[建筑科学—市政工程]