Research And Application On Automated Analysis Based On Live Fish Aquaculture Water Quality

Our country is a aquaculture superpower. But for a long time, it was under theinfluence of traditional concept of aquaculture and the low level technology,aquaculture present on a smaller scale, scattered distribution range, in the majoritywith mechanized farming feature. In reality，it often gives us the impression of lowdegree of automation and lacking of management way. By the way，its water qualityearly warning ability is so poor. All these have brought big problem to the pondaquaculture. According to this problem，the paper makes a research and implement anautomated analysis system based on live fish aquaculture water quality. The papermainly includes the following:(1) In the pond water quality comprehensive evaluation，Because in the real life,it only aims at a single water quality factor monitoring and evaluation, lacking of fullyconsidering the potential relationship between each quality factor. So the paper usescompound gaussian trapezoidal membership function to build the live fish waterquality evaluation model to each water factor based on six water parameters whichhave stored in the system. Then model include two different parts which is constitutedby gaussian function. The system use the rank numbers and rank exchanging fuzzyclustering method to calculate out the final comprehensive grade evaluation. Thisfunction overcomes the problem which is great variety in water quality partition anduse rank numbers and rank exchanging chart to reduce the difficulty of water qualityhierarchy. All these will realize the grading and rating of intelligence.("a gradeevaluation method of live fish aquaculture water quality" has been applied fornational patent， application number:201310088205.5)(2) In the comprehensive prediction of pond water quality, the monitoring andevaluation equipment often present the single monitoring function, can’t predict thefuture water quality and give the warming tips. So the paper uses radial basis functionneural network as the basis of forecasting and recursive rolling prediction method forembodying the characteristics of pond water quality environment dynamic change. To some key parameters, it uses system dynamics model to find the potential relationsand the feedback mechanism between different parameters for building the neuralnetwork prediction model of key parameters. Using the model which have been builtto analysis and calculate out the key parameter value. lastly，it can get the final resultthrough readjusting the comprehensive forecasting value and key parameterforecasting value.("a intelligent integrated prediction method of live fish aquaculturewater quality" has been applied for national patent， application number:201310088194.0)(3) The paper follows the software engineering designing thought， using B/Sarchitecture which is based on Struts， Spring and Hibernate integration framework todesign and implement the intelligent live fish aquaculture water quality monitoringand controlling system. Through the analysis and calculation, the system return thewater quality and warming information to pond managers, it realizes the automationof pond water quality management and greatly improves the efficiency of pondfarming.