摘要
鉴于传统BP神经网络在高铁桥沉降变形预报中随机性强、收敛速度慢、易陷入局部极值等缺点,本文引入了顾及邻域粒子群影响的改进粒子群算法(IPSO)优化BP神经网络,建立IPSO_BP的高铁桥台沉降变形预报模型,组合模型的预报结果与高铁沉降变形评估方法—Asaoka进行比较,结果表明:基于改进的粒子群优化BP神经网络模型较高铁桥传统BP预报模型收敛速度更快,预报精度更高;预报评估结果与Asaoka方法预报的结果相符,证明了IPSO_BP模型的可靠性和实用性。
As the shortcomings that traditional BP neural network has a strong randomness when Initialization,converges slowly and falls into local optimum easily,an improved particle swarm optimization algorithm,which takes into account the impact of neighborhood particle swarm,is introduced in this paper to optimize the BP neural network,finally,establishing a settlement deformation prediction model based on IPSO_BP of a High-speed railway bridge.Compared the prediction results of the combined model with the results of High-speed railway settlement deformation evaluation method—Asaoka,the results show that the IPSO_BP model has a faster convergence speed and higher forecast precision than the traditional BP neural network model;Consistent prediction and assessment result with the Asaoka method,which improves the reliability and practicability of IPS_BP model.
出处
《城市勘测》
2017年第6期135-138,共4页
Urban Geotechnical Investigation & Surveying