基于MPSO-RBF神经网络的切向刚度研究

为了能更快速、准确地计算在多影响因素下的机械结合面切向刚度,采用改进的粒子群算法优化径向基神经网络参数,实现了两个算法的有机结合。考虑结合面的材质、表面加工方法、表面粗糙度、结合面面压、介质等影响结合面切向刚度的因素,以实验参数作为样本,利用建立的模型进行了结合面切向刚度仿真,并对仿真结果与实验结果进行了对比分析。分析结果表明,模型预测精度可达92%以上。

Research on Tangential Stiffness Modeling Based on MPSO-RBF Neural Network Algorithm

In order to calculate the mechanical joint surface tangential stiffness under the influence of multi-factors quickly and accurately,the modified particle swarm algorithm is adopted to train and to optimize the radial basis function neural network parameters,whereby realizing the organic combination of two algorithms.In considering the material quality on joint surface,surface machining method,surface roughness,surface pressure o joint surface,medium and other factors affecting tangential stiffness on joint surface and with experiment parameters as the model,the established model in used to simulate the tangential stiffness on joint surface.A contrast analysis is made of the simulation results and experimental results.The analytical results indicate that the prediction accuracy by the model can reach over 92%.