PSO优化RBF神经网络在变形监测中的应用

Application of PSO-optimized RBF Neural Network in Deformation Monitoring

变形监测数据非平稳、非线性、时变性和不确定性特征明显,传统方法难以取得理想的分析结果。针对该问题,提出一种基于粒子群(Particle Swarm Optimization,PSO)算法优化的径向基函数(Radical Basis Function,RBF)神经网络变形监测数据拟合和预测方法,利用PSO实时对RBF神经网络初始参数进行自动寻优,从而解决RBF神经网络初始化参数选择困难、易陷入局部极值的问题。最后采用实测数据构建典型试验对所提方法进行验证,结果表明PSO-RBF网络模型能够实现高精度的变形监测数据拟合和预测,结果优于传统数据分析方法。

The data of deformation monitoring are non-stationary,non-linear,time-varying and uncertain.It is difficult to get ideal analysis results by traditional numerical method.To solve this problem,a RBF neural network deformation monitoring data fitting and prediction method based on particle swarm optimization(PSO)algorithm optimization is proposed,which uses PSO to automatically optimize the initial parameters of RBF neural network in real time,so as to solve the problem of slow convergence speed of RBF neural network and easy to fall into local extremum.Finally,a typical test is built to verify the proposed method.The results show that the PSO-RBF network model can achieve high-precision deformation monitoring data fitting and prediction,and the results are better than the traditional numerical calculation method.