Pond dissolved oxygen prediction method based on data restoration

Abstract

The invention discloses a pond dissolved oxygen prediction method based on data restoration. The prediction method comprises the steps of determining the prediction object, defining the meteorological composite index, data restoration, constructing a GRNN neural network model, initializing the GRNN neural network and training the neural network. Eleven index parameters of meteorological factors and aquaculture environment factors related to pond dissolved oxygen prediction are involved. The eleven index parameters act as the input quantity, and the lost and abnormal data in sensor acquisition are restored by using the meteorological composite index and linear interpolation based on short period of continuity of the meteorological environment data and the water quality data. The dissolved oxygen concentration acts as the output quantity, and a GRNN neural network prediction model is determined. Compared with the dissolved oxygen prediction training effect of the conventional BP model, the GRNN algorithm of the GRNN network model has higher accuracy in comparison with the BP model and can better reflect the basic trend of the change of the pond dissolved oxygen within a period of time.

Description

technical field [0001] The invention relates to a method for predicting dissolved oxygen in ponds, in particular to a method for predicting dissolved oxygen in ponds based on data restoration. Background technique [0002] Intensive aquaculture will play an important role in future aquaculture. Dissolved oxygen prediction is a very important task in aquaculture management. How to obtain dissolved oxygen concentration information and take measures to increase oxygen before low-concentration dissolved oxygen occurs is an important issue in the process of intensive industrial aquaculture. Dissolved oxygen prediction through modern information technology can provide an important reference for water quality management and control for intensive industrial aquaculture, so as to reduce the risk of aquaculture and improve economic benefits. [0003] At present, in the prior art, genetic algorithm and BP neural network are used to establish models from meteorological indicators such ...

AI Technical Summary

Problems solved by technology

[0003] At present, in the prior art, genetic algorithm and BP neural network are used to establish models from meteorological indicators such as air pressure, temperature, rainfall, light intensity, wind speed, wind direction and air humidity to predict the dissolved oxygen concentration in ponds; The regression machine obtains the predicted value of dissolved oxygen concentration from the water quality index and related meteorological factor data within a predetermined period; in addition, it also uses the ant colony algorithm to optimize the penalty factor and kernel function width parameters of the least squares support vector machine to establish the dissolved oxygen concentration. Prediction model, various prediction methods still have certain errors in the actual dissolved oxygen prediction

Images