Research And Application Of Intelligent Control And Assistant Decision Support System For Aquaculture

The integration of modern information technology and fishery is an inevitable trend to promote the development of fishery science.Collect historical data and process data generated by aquaculture for analysis,forecast,early warning and control of aquaculture water quality,and establish aquaculture decision-making knowledge base to provide scientific and effective auxiliary decision-support information to promote the informatization and intelligentization of aquaculture Development is of great significance.This thesis cooperates with the Suzhou Marouzu Agricultural Professional Cooperative Association Fishery Standardization Farming Project.The main research contents are as follows:(1)Water quality parameter collection The aquaculture monitoring system is implemented based on PLC.The sensor collects real-time data of the breeding process such as dissolved oxygen,pH,temperature,salinity,ammonia nitrogen,and light intensity through the wired network,Bluetooth,digital transmission,etc.to the on-site monitoring system.On-site monitoring system The water quality status is monitored in real time,and real-time data is transmitted to the data center,where the water quality status of each pond can be observed in real time,and data storage,data analysis,and abnormal alarm functions are provided.(2)Intelligent control of water quality parameters.In view of the current fuzzy rules in fuzzy PID controllers that rely on artificial prior knowledge for water quality parameter control,it is difficult to obtain optimal fuzzy rules that affect the control accuracy.According to the characteristics of fuzzy PID controllers that have correlations,The bat algorithm introduces a neighborhood search operator and a chaotic mutation operator to optimize fuzzy rules.The optimized fuzzy PID controller is applied to aquaculture water quality parameter control to improve control efficiency.(3)Water quality prediction and early warning.For the prediction of water quality,most of them are predicted by single factors,without considering the correlation between the parameters.Principal component analysis and PCA-LSTM network water quality parameter prediction model are proposed.First,the main component analysis method is used to screen the multivariate time series such as pH,dissolved oxygen and salinity in aquaculture to reduce the input data dimension of the prediction model.Secondly,the sliding window is used to divide and train the historical data set,and finally the LSTM network is used to predict the aquaculture water quality parameters and provide early warning of abnormal water quality.(4)Auxiliary decision support for aquaculture.Based on aquaculture water quality parameter collection,water quality parameter intelligent control and water quality prediction and early warning,data analysis is performed to provide aquaculture personnel with auxiliary breeding decision support information,through human-computer interaction to determine breeding decisions,implement decisions,and apply this decision to improve and optimize The knowledge base for breeding decision-making provides more scientific methods to assist breeding and reduce breeding risks.