Research On Water Temperature Prediction And Application Of Hairy Crab Farming Based On GA-BP Algorith

Intensive culture of hairy crabs has a broad development prospect in China,but the existing water temperature prediction models are still shallow in the field of aquaculture,and the prediction accuracy is not high,which is difficult to meet the strict requirements of intensive culture on the control of water temperature In this study,a water temperature prediction model for hairy crab culture based on GA-BP algorithm is designed for more accurate water temperature prediction;At the same time,the application of water temperature prediction model in practical production is given specific suggestions.The main research work includes:(1)Design of GA-BP algorithm: Based on existing research,in order to solve the shortcomings of traditional BP neural network,such as being easily trapped in local optimization,long training time,and difficult to determine the initial weights and thresholds of the network,this study uses the advantages of genetic algorithm(GA),such as global optimization and parallel computing,to design GA-BP algorithm to better meet the characteristics and requirements of hairy crab breeding water temperature prediction.This algorithm first uses the GA algorithm to generate the optimal weights and thresholds of the BP neural network,and then uses the error gradient descent method of the BP neural network to fine tune the weights and thresholds of the above network,so that the network quickly converges to the expected target.The simulation experiment shows that compared with the traditional BP neural network algorithm,the GA-BP algorithm has shorter training time,higher prediction accuracy and more stability,and the comprehensive performance of the network has been improved,which can provide better algorithm support for the prediction of water temperature in hairy crab breeding.(2)To build a prediction model of hairy crab culture water temperature: first,through the system dynamics analysis of the water temperature in the aquaculture pond,the influencing factors that can be used to predict the hairy crab culture water temperature are obtained,and the model input is determined;Then 600 training samples and 120 test samples were obtained from a pond in A hairy crab breeding base in Changsha.After outlier processing and normalization,they were input into a 5-5-1 three-layer network and trained using the above-mentioned GA-BP algorithm.At the same time,model tests were conducted after the training.Finally,a water temperature prediction model was developed that can output predicted water temperature values for hairy crab aquaculture ponds using five major meteorological data: temperature,solar radiation,rainfall,wind speed,and air humidity.(3)Carry out empirical analysis and simulation experiments on the prediction model of hairy crab culture water temperature.The empirical data were obtained from other ponds in A hairy crab breeding base in Changsha,and the water temperature prediction was carried out for three periods of time.It was found that the research model could complete the water temperature prediction with high accuracy,and the feasibility was good.At the same time,through horizontal and vertical comparative simulation experiments,it was found that the model in this study has better prediction accuracy than the water temperature prediction model based on Gaussian process regression and ARIMA in the task of predicting water temperature in hairy crab aquaculture;and it is stronger in convergence,accuracy,and stability than the water temperature prediction model based on BP neural network.(4)Propose application suggestions for water temperature prediction: In order to ensure that the validated model can be effectively applied in the production practice of intensive pond aquaculture of hairy crabs,and fully utilize the practical value of the model,this study provides relevant suggestions for the practical application of water temperature prediction models for hairy crabs,including using water temperature prediction models for intelligent water temperature warning,automatic control,and water temperature data analysis.This study considers the characteristics of intensive aquaculture of hairy crabs and the requirements for water temperature control,and in response to the limitations of existing water temperature prediction models,a new water temperature prediction model suitable for predicting water temperature in hairy crab aquaculture is proposed.Its application suggestions in practical production have strong practical significance for guiding water temperature control in hairy crab aquaculture and achieving scientific aquaculture.