Prediction model for the number of crucian carp hypoxia based on the fusion of fish behavior and water environment factors

Abstract Fish hypoxia is one of the main tasks of aquaculture monitoring. Abnormal fish hypoxia easily leads to ponding or even mass death of fish, causing serious economic losses to farmers and aquaculture enterprises. In this paper, we propose a model for predicting the number of abnormal crucian carp hypoxia based on the fusion of fish behavior and water environment factors. The cubic convolution interpolation method is used to fuse, synchronize and enhance the calculated number of crucian carp hypoxia with the water environment factors such as dissolved oxygen, pH and temperature. An improved particle swarm optimization (IPSO) algorithm with the dynamically adjusted inertia weight optimizes the initial weights and thresholds of the BP neural network. We adopt an IPSO-BPNN algorithm to construct a non-linear prediction model to predict the number of crucian carp hypoxia. The experimental results show that the model can accurately predict the number within 3 h, and sum of squares due to error (SSE), mean squared error (MSE), mean absolute percentage error (MAPE) of the model are 1.11, 0.015, 0.083, respectively. It shows that the proposed model has good generalization ability and predictive performance.