The Research Of Dissolved Oxygen Prediction Algorithm And Its Application In Aquaculture Monitoring Platform

The total amount of aquatic products in China has been rising year after year,and the scale of aquaculture is also expanding.Strict management and control of the water quality in the aquaculture area is a very important link in the process of aquaculture,which is directly related to the economic results of aquaculture.The current conventional water quality monitoring system has disadvantages such as high cost,slow control response,low data collection accuracy,and inability to predict data.In order to solve the problems existing in the conventional water quality monitoring system,improve the ability of accurate water quality management,and enhance the ability of risk prevention in the process of aquaculture,the following studies are carried out in this paper:First,A water quality data acquisition system with high precision is designed.The water quality data acquisition system designed based on ultra-low power consumption chips and high-precision sensors can collect and upload p H,temperature,dissolved oxygen,conductivity,and ammonia nitrogen data in real time.This paper employs the principle of fluorescence quenching to design the dissolved oxygen sensor,the acquisition precision is higher than the accuracy of general dissolved oxygen sensor.Second,A revised model for correcting water quality data with noise is proposed.The water quality data collected by the system mainly consist of popcorn noise,environmental noise and electrical noise.In this paper,median filter,notch filter and wavelet transform are respectively used to correct the data successively.In this paper,six kinds of wavelet functions in wavelet transform are selected to amend the electrical noise in the data respectively,and the six corrected results are compared and analyzed to elect the most appropriate wavelet functions for different types of parameter data.Third,A prediction model of dissolved oxygen in water quality parameters was established.After correcting the five kinds of data,this paper realized the prediction function of dissolved oxygen based on three methods: RNN,LSTM and GRU.Set up the network model,set the corresponding parameters in the neural network,process and annotate the data set.For different data sets,they are put into different neural network models for training and prediction.Evaluate and analyze the prediction results of various models,and select the model with the best prediction performance finally.The model can predict the content of dissolved oxygen within 1 h accurately,compared with other models,the model has the best performance in four evaluation indexes.Fourth,An early warning system for water quality monitoring and dissolved oxygen with low delay control is designed.This paper designed a set of water quality monitoring system,mainly designed several parts of data collection,data processing and analysis,data visualization and system control.In this paper,a pre-warning mechanism of dissolved oxygen is added to the system,and the function of system control and prediction of dissolved oxygen is implemented in the upper computer control center,which reduces the response delay of the system control and enhances the ability of preventing hypoxia.The aquaculture monitoring system established in this paper realizes the dynamic monitoring of water quality data and the prediction of dissolved oxygen within 1 h.The system has low cost,high data acquisition accuracy,low control delay,and a more accurate dissolved oxygen prediction function.Applying this system to water quality aquaculture can improve the ability to monitor water quality and prevent the occurrence of hypoxic or oxygen-enriched conditions in advance,thus enhancing the ability to accurately control water quality in the aquaculture process.