Design And Implementation Of Intelligent Monitoring System For Water Quality Monitoring Industry

In recent years,China's economy has developed rapidly.However,due to the weak awareness of environmental protection in the early stage of development,the contradiction between the current socio-economic development and environmental protection has been caused.With the improvement of people's material demand and overall quality,the environmental quality,especially the quality of water resources,is becoming more and more high,which makes various policies from the central government to the government to promote the protection of water resources,but also makes the water quality monitoring industry get rapid development.Although the whole industry is developing rapidly,it has the characteristics of weak development ability,backward technology and lack of water quality knowledge platform.In the development of water quality monitoring system,at present,many water quality monitoring systems use threshold method for early warning,without considering the overall water quality,resulting in a large number of false alarms and false alarms.When sending early warning information to the relevant personnel,the information provided is generally only the name of the abnormal site and the index exceeding the standard.A small amount of information increases the difficulty of the relevant personnel to solve and identify the anomaly.After the occurrence of abnormal water quality,the system can not provide a water quality knowledge platform for the relevant personnel to reference and learn,which increases the cost of solving the problem.Machine learning and knowledge mapping are the research hotspots in recent years.Machine learning can mine the potential patterns of data,while knowledge map can detect the potential causal relationship between different entities.Therefore,a water quality monitoring system with machine learning and water quality knowledge map as the core is constructed,and the related algorithms and theories are studied,and the design and implementation of water quality monitoring system are also carried out.This paper mainly completed the following work:(1)This paper introduces the source and acquisition method of water quality data,and analyzes the water quality data.It is found that the water quality data has some characteristics,such as low characteristic dimension,few abnormal data and sparse distribution,more local abnormal points,and the characteristics of water quality category are not dependent on a single index.A variety of anomaly detection algorithms are proposed and analyzed,including lof and one class Support vector machine(SVM),isolated forest and deep isolated forest are used to parallelize the isolated forest,and the parallel isolated forest is constructed.Through experiments and analysis,it is found that the accuracy of identifying local abnormal points and complex data model is low due to the overlapping and coverage effect of parallel partition data along the axis of isolated forest Analysis shows that the performance of deep isolated forest is not only better than that of isolated forest and other two algorithms in water quality anomaly data,but also performs well in anomaly identification of local outliers and complex data model.Because of the algorithm of early stop,deep isolated forest can save some time cost.(2)Through the research of water quality early warning strategy,the warning level of water quality is divided and the process of water quality early warning is formulated.Because this paper hopes to provide more useful information for relevant staff at the same time of water quality early warning,and can provide a query platform of water quality knowledge,the concept of knowledge map is introduced,and the related task entity identification,relationship extraction,knowledge question and answer are also known This paper introduces the theory of map recognition tasks,and carries out simulation and analysis of various algorithms,and finds that multi head in entity recognition The performance of attention is better,and the accuracy rate of training set and verification set can reach more than 99%;in the aspect of relation extraction,the performance of PCNN considering the relative position of entities is far better than the other three algorithms in both data set 1 and data set 2;the method of matching based on template rules is used in knowledge question answering and the effect is demonstrated.(3)Design and implementation of water quality monitoring system.The system architecture,technical route and database are designed;seven modules including authority module,data display module,message module,management module,exception handling module and water quality knowledge map module are realized,and the function display and details are introduced.