中文摘要：

斜齿圆柱齿轮承载能力的设计过程中,涉及到大量的非线性约束条件分析计算,传统的优化方法很难得到全局最优解.利用BP神经网络非线性映射能力,建立了斜齿圆柱齿轮设计参数与承载能力之间的全局映射关系,利用训练后的BP神经网络模型对斜齿圆柱齿轮承载能力设计缺陷进行辨识.采用遗传算法对斜齿圆柱齿轮承载能力设计缺陷进行优化修正,利用BP神经网络得到的斜齿圆柱齿轮结构特性响应构造了遗传算法罚函数,提高了遗传迭代过程中的约束计算效率,优化结果表明该方法能够有效提高斜齿圆柱齿轮设计缺陷优化修正效率.

英文摘要：

There are large numbers of nonlinear constraint analyses in the design of helical gears load capacity. It is difficult to obtain global optimal solution by means of traditional optimization methods. Utilizing nonlinear mapping ability of BP neural network, global mapping relationships between helical gears parameters and load capacity are established. Design defects of helical gears load capacity are identified by trained BP neural network. Design defects of helical gears load capacity are corrected optimally using genetic algorithm. Taking structure characteristic response of helical gears obtained by BP neural network as penalty functions of genetic algorithm, computational efficiency of constraint analyses in genetic iterative process is improved. Optimization results demonstrate that above approach can enhance optimal correction efficiency for design defects of helical gears load capacity effectively.