关键词: 麻雀搜索算法, 极限学习机, 土壤供肥量预测, 智能灌溉, 智能农业

Abstract: Aiming at the shortcomings of low accuracy and poor efficiency of soil fertilizer prediction in traditional agricultural irrigation system, an agricultural soil fertilizer prediction model MHISSAELM based on improved sparrow search algorithm and optimized limit learning machine is proposed. Firstly, the blind flight, the population diversity, the balance between global search and local development, and the global search ability to jump from local optimization of the traditional sparrow search algorithm are improved by introducing the flight guidance, the piecewise weighted sine cosine optimization, the discoverer position update, the vigilant step factor nonlinear update and the mutation opposition learning mechanism, so as to improve the optimization performance of the algorithm and realize the multistrategy hybrid improved sparrow search algorithm MHISSA. Then, in order to improve the prediction accuracy and generalization ability of limit learning machine elm, MHISSA algorithm is used to iteratively optimize the connection weight and hidden layer deviation of ELM network. Aiming at the prediction of agricultural soil fertilizer, a soil fertilizer prediction model MHISSAELM based on MHISSA optimized limit learning machine is constructed. The analysis of experimental results shows that compared with the four similar comparison models, the prediction curve of MHISSAELM is closest to the actual curve. The prediction error can be controlled between ［-10, 15］kg/hm2, the maximum relative error is 4.8%, the absolute percentage error is 1.7% in MHISSAELM. And the prediction accuracy is the highest among all comparison models, and is practical in the field of agricultural intelligent irrigation.

Key words: sparrow search algorithm, extreme learning machine, soil fertilizer prediction, intelligent irrigation, intelligent agriculture