To solve the problem of predicting water level in front of check gate under different time scales, a different time scale prediction model with a long short term memory (LSTM) neural network based on adaptive inertia weight comprehensive learning particle swarm optimization (AIW-CLPSO) is proposed. The AIW and CLPSO are adopted to improve the global optimization ability and convergence velocity of particle swarm optimization in the proposed model. The model was applied to the water level prediction in front of the Chaohu Lake check gate. The example of the water level prediction in front of the Chaohu Lake check gate shows that the proposed model predicts the trend of water level fluctuation better than LSTM with high accuracy of Nash coefficient up to 0.9851 and root mean square error up to 0.0273 m. The optimized algorithm can obtain the optimal parameters of the LSTM neural network, overcome the limitations of the traditional LSTM neural network in parameter selection and inaccurate prediction, and maintain good prediction results in the predicting water level in front of the check gate at different time scales.This study can provide important reference for water level prediction, scheduling warning, water resources scheduling decision and intelligent gate control in long distance water transfer projects.

针对不同时间尺度下闸门前水位预测问题，提出了基于自适应惯性权重综合学习粒子群优化的长短期记忆（LSTM）神经网络不同时间尺度预测模型（AIW-CLPSO）被提议。该模型采用AIW和CLPSO来提高粒子群优化的全局优化能力和收敛速度。该模型应用于巢湖闸门前水位预测。以巢湖闸门前水位预测为例表明，该模型对水位波动趋势的预测优于LSTM，纳什系数高达0.9851，均方根误差高达0.0273 m。优化后的算法能够获得LSTM神经网络的最优参数，克服了传统LSTM神经网络在参数选择和预测不准确方面的局限性，在不同时间尺度的闸门前水位预测中均保持了良好的预测结果该研究可为长距离调水工程中的水位预测、调度预警、水资源调度决策和智能闸门控制提供重要参考。