论文部分内容阅读
针对不恰当地选取RBF神经网络的网络结构和参数会使网络收敛慢的问题,采用粒子群优化算法对RBF神经网络参数进行优化,建立了基于粒子群优化算法的RBF神经网络模型(PSO-RBF模型),对泾惠渠灌区地下水位埋深进行了模拟和预测。结果表明,与单一的RBF神经网络相比,PSO-RBF模型具有较高的预测精度。再根据时间序列预测法预测的降水量、径流量、蒸发量、渠灌引水量、地下水开采量、气温等模型的输入变量,用训练好的PSO-RBF模型预测了泾惠渠灌区2009~2020年地下水位埋深,发现该灌区地下水位埋深呈下降趋势。
Aiming at the problem that the network structure and parameters of RBF neural network are inappropriately selected, the problem of network convergence is slow. Particle swarm optimization algorithm is used to optimize the parameters of RBF neural network, and RBF neural network model based on Particle Swarm Optimization (PSO-RBF) is established Model) to simulate and predict the groundwater table depth in Jinghuiqu irrigation area. The results show that compared with a single RBF neural network, the PSO-RBF model has higher prediction accuracy. Based on the input variables of precipitation, runoff, evaporation, canal water diversion, groundwater exploitation and temperature predicted by the time series forecasting method, the PSO-RBF model was used to predict the economic benefits of the Jinghuiqu irrigation district from 2009 to 2020 The depth of groundwater table, found that the irrigation groundwater depth is declining.