Development Of Fish Pond Water Quality Detection Platform Based On BP Neural Network

China has always been a large traditional aquaculture country,with its output exceeding one third of the world.China’s traditional fisheries mainly focus on small-scale family farming,but in family farming,farmers often judge the water quality through experience,and are prone to deterioration of water quality caused by improper personal operation.This paper aims to design a water quality detection platform that can help farmers realize the fine management of fish ponds and detect the water quality environment of fish ponds.This paper mainly carries out the following research work.(1)The main parameters of fish pond culture are analyzed.The platform selects dissolved oxygen,p H,ammonia nitrogen and phosphate to evaluate the overall water quality of the fish pond in terms of water quality detection.In terms of water quality evaluation standard,according to the characteristics of complex water environment and various influencing factors of the fish pond,the environmental quality standard for surface water is selected as the water quality evaluation standard.(2)The structure and basic principle of artificial neural network are deeply discussed.Using the advantages of BP(back propagation)neural network and fuzzy mathematics,BP fuzzy neural network is selected as the algorithm of output water quality evaluation.A 4-9-1 three-layer fuzzy BP network model is constructed in MATLAB.Because there are few standard samples,the random interpolation method is used to expand the standard samples as the training and test data set.Because of the slow convergence of BP neural network,LM(Levenberg-Marquardt)algorithm is analyzed and selected to improve the network model.The membership degree is calculated by triangular membership function and Gaussian membership function respectively,which makes the output of water quality evaluation of the network model more objective and accurate.(3)Sj-twc 3101 water quality analyzer is selected for water quality data acquisition.The monitoring parameters of the water quality analyzer include the four parameters required in this paper.It has the advantages of dust prevention,light avoidance,stability and anti-interference.It can be widely used in freshwater aquaculture,mariculture,swimming pool water,fish pond aquaculture and agricultural irrigation.It can communicate with the host computer through RS485.In terms of data preprocessing,the emergence and solutions of missing values and abnormal values are analyzed.For missing values,Lagrange interpolation method is used;For outliers,outliers are treated as missing values.(4)In the design of the web platform,Alibaba cloud server,layui,spring,springboot,mybatis and other frameworks are analyzed and selected to complete the functions of fish pond management,food and drug administration,water quality parameter visualization and upload them to Alibaba cloud server.