Study On Prediction Model And Method Of Aquaculture Water Environment

Aquaculture water environment is an important living place for aquaculture organisms.The quality of the water environment will greatly affect the normal growth of aquatic organisms,and the water environment is affected by many complex factors.In order to ensure the safe production of aquaculture water environment,this paper selects the key parameters of aquaculture water environment,dissolved oxygen and p H value as the research object,constructs its prediction model by using data analysis method,machine learning and neural network combined with optimization algorithm,and designs aquaculture water environment monitoring and early warning system.The details are as follows:(1)A nice data source is the basis for ensuring the accuracy of prediction.As the aquaculture water environment is vulnerable to the influence of weather,physics,chemistry and human activities,the attachment of aquatic plants limits the monitoring of sensing equipment,and the collected data are often missing or have abnormal values.In order to improve the data quality,this paper first preprocesses the original data and adopts different repair methods for different missing or abnormal problems.Due to the complex interaction mechanism between aquaculture water environment influencing factors,The key factor is selected and the data dimension is reduced by principal component analysis,which can avoid the redundancy of model input data to a certain extent.(2)An improved k-means clustering and particle swarm optimization(IPSO)algorithm combined with long short-term memory(LSTM)neural network dissolved oxygen prediction model is proposed.The improved k-means clustering algorithm is applied to divide the environmental data into several categories.On this basis,The improved aquaculture dissolved oxygen prediction model is established through LSTM neural network algorithm.At the same time,the improved particle swarm optimization algorithm is introduced to optimize the model parameters in order to reduce the unscientific selection of empirical parameters.Under different weather conditions,The model is used to predict dissolved oxygen.The experimental results show that under good weather conditions,the prediction error curve of the model fluctuates less and the prediction accuracy is higher.When the weather changes suddenly,the average absolute percentage error,root mean square error,average absolute error and Nash coefficient of the evaluation indexes of the dissolved oxygen prediction model are0.1295,0.6453,0.4613 and 0.9022 respectively,which have reached a good level.The model improves the problems of data loss and poor robustness under the condition of weather changes to a certain extent.(3)The p H value prediction model of back propagation(BP)neural network optimized by orthogonal Signal Correction for Partial Least Squares Method is proposed.Principal component analysis(PCA)is used to reduce the dimension of the data,Leaky Re Lu functions are used as the activation function of BP neural network,OPLS is used to establish the linear relationship between the output matrix and the hidden layer,and genetic algorithm is used to optimize the initial weight and threshold of neural network.GA-OPLS-LRBP model is established.The experimental research shows that compared with OPLS-LRBP,LRBP and BP,the prediction accuracy of p H value by the model is the highest,reaching 0.9437.It shows that the proposed GAOPLS-LRBP model can improve the prediction accuracy of p H value of aquaculture water environment to a certain extent.(4)Through the organic combination of sensor data collection,data transmission method and prediction model,the aquaculture water environment monitoring and early warning system is designed.The data is monitored online and queried through the remote terminal,and the water environment parameters are predicted and the data are visualized effectively.The system improves the backward mode and poor reliability of aquaculture water environment monitoring,and improves the farmers’ mastery of aquaculture environment.