PSO-BP模型在水库大坝变形预测分析中的应用研究

Application of PSO-BP Model in Deformation Prediction Analysis of Reservoir Dams

针对BP神经网络预测模型收敛速度慢及容易出现局部极值等弊端,难以对水库大坝变形趋势进行准确预测的实际情况,本文采用粒子群算法对其进行优化研究,构建PSO-BP神经网络模型。以某水库大坝沉降连续30期监测数据为数据源,将前24期数据作为训练基础,对后6期沉降数据进行预测。为对PSO-BP神经网络模型预测成果的可靠性进行分析研究,分别采用GM(1,1)模型、BP神经网络模型、PSO-BP神经网络模型对水库大坝变形趋势进行预测分析。实验结果表明,3种预测模型的均方根误差分别为0.3574.0.2550.0.1783mm,优化后的预测模型准确性相对更高,故PSO-BP预测模型能够对水库大坝变形趋势进行更为准确的反映。

In view of the disadvantages of BP neural network prediction model,such as slow convergence speed and local extreme val-ue,it is difficult to accurately predict the deformation trend of reservoir dams.Particle swarm optimization algorithm is used to opti-mize it and build a PSO-BP neural network model.Taking the 30 periods of consecutive monitoring data of dam settlement of a reser-voir as the data source and the first 24 periods of data as the training basis,the subsequent 6 periods of settlement data are predicted.In order to analyze and study the reliability of the prediction results of the PSO-BP neural network model,GM(1,1)model,BP neu-ral network model and PSO-BP neural network model were respectively used to predict and analyze the deformation trend of reservoir dam.The experimental results showed that the root-mean-square errors of the three prediction models were 0.3574 mm,0.2550 mm and 0.1783 mm,respectively.The optimized prediction model was relatively more accurate.Therefore,PS0-BP pre-diction model can reflect the deformation trend of reservoir dam more accurately.