Intelligent prediction model for rice pests and its application

Prediction of rice pests is greatly influenced by uncertain factors, for example time and weather change. To improve prediction efficiency and precision, a new intelligent prediction model for rice pests, which uses artificial neural network, genetic algorithm and simulated annealing algorithm is proposed. First the network model structure was built based on artificial neural network technology, and available multi-dimensional data as well as pest occurrence levels history data were used as network input and output variables. Then genetic algorithm was set in network exterior circle, and simulated annealing algorithm was applied in network interior circle to train network nodes connections weights and thresholds until model output results approximate target vector. The model was applied to Chongqing Yongchuan rice chilo suppressalis predictions. Result shows that model can predict future chilo suppressalis occurrence level more exactly and compared with traditional BP neuron network, model prediction accuracy and operation time were improved greatly. So for rice pests prediction, computational intelligence technology is of good practical utility value.