将人工神经网络应用于机床刚度建模,针对基本BP算法收敛速度慢、易陷入局部极小点的不足,从步长和搜索方向两方面对基本BP算法进行了改进,并引入具有全局寻优能力的粒子群算法。通过统计误差计算的次数、设定多组初始权值及方差分析等方法,对几种优化算法在机床刚度建模中的应用效果进行了比较,最后以输出误差最小时的连接权值建立了机床刚度神经网络模型。 The artificial neural network is applied to the modeling of machine tool stiffness. In order to overcome the shortcomings of the basic BP algorithm, such as slow convergence rate and easy falling into local minimum point, the basic BP algorithm is improved from the step size and search direction, Capability Particle Swarm Optimization. Through the calculation of the number of statistical errors, set of multiple initial weights and analysis of variance and other methods, several optimization algorithms in the machine tool stiffness modeling results were compared, and finally the output error when the connection weights established the minimum Machine stiffness neural network model.