Study On Agricultural Irrigation Leakage Monitoring System Based On STM32

In the process of modern agricultural irrigation,leakage of water pipelines will cause serious waste of water resources,and even secondary pollution.Therefore,the leakage of water pipelines must be reported to users in real time,so as to play the role of real time monitoring.Not only that,there are certain requirements for the accuracy of pipeline leakage location judgment.To satisfy the above requirements,this paper provides a solution for agricultural irrigation leakage monitoring system based on STM32.The system obtains pipeline vibration information through low-power Iot sensing equipment,and analyzes the data by using support vector machine algorithm,so as to obtain pipeline leakage status and leakage location in time.Refer to the literature on water pipeline leakage monitoring,through research and analysis,make clear the method of water pipeline leakage detection.There are currently many leak detection techniques for water pipelines,and classical leak detection methods are generally divided into acoustic leak detection and vibration leak detection.In the agricultural irrigation system,it is hard to calculate the acoustic sensor position in the water pipe,so this paper adopts the method of vibration leakage detection to detect whether the water pipe leakage occurs in the agricultural irrigation system.In the leak monitoring experiment,the data collected by the data acquisition module is analyzed in time domain and frequency domain,and the data is compressed based on time domain and frequency domain,namely feature extraction.Using chi-square test and model training,libsvm toolbox was used in MATLAB software to realize classification and in-depth analysis of extracted features.Finally,five features with top Chi-square score values on multiple axes were used to classify so as to more accurately judge the leakage situation of pipelines.In addition,the location of the leakage was predicted.The experimental results show that the average prediction error of the prediction results under mixed characteristics is 25.3 millimeter.However,the average prediction error for a single Y-axis feature is large,reaching 76.5 millimeter,it shows that it is more accurate to predict the location of pipeline leakage by using mixed features.Finally,the data upload time before and after feature compression is compared.The results show that it takes 694 seconds to send 512 bytes and 170 seconds to send 5 bytes.It indicates that the power consumption of data acquisition module in the system can be greatly reduced by compressing the extracted features.