基于遗传算法的双圆弧齿轮传动的神经网络优化设计

Optimal design of transmission system of double circular arc gears by means of genetic algorithm and neural network

针对常规设计中双圆弧齿轮体积较大的问题,在先选定螺旋角的情况下,以圆弧齿轮传动的小齿轮齿数、法面模数及齿宽为设计变量,以两圆弧齿轮的体积之和最小为目标函数,应用遗传算法设计理论,建立双圆弧齿轮传动的遗传算法优化设计的数学模型,采用神经网络对其进行优化.结果表明,应用遗传算法和神经网络相结合的方法对双圆弧齿轮传动进行优化设计,可以克服单纯用遗传算法的早熟现象和神经网络发生局部收敛的缺点,体积减少30%以上.

Aimed at the problem of bulkier dimensions of the double circular arc gears in the conventional design of transmission system, an optimization of the system design was performed. Firstly, the number of teeth of the minor gear in the system, its normal module, and its tooth width were taken as design variables in the condition of definitely selected helical angle, and the sum of volumes of the double circular arc gears was taken as objective function. Then, the design principle of genetic algorithm was employed to build up a mathematic model for optimization of the design of aforesaid transmission System. Finally, the optimization was performed by using the neural network procedure. The result of optimization showed that, when the genetic algorithm in combination with the neural network procedure was used in the system optimization design, the shortage such as the prematurity phenomenon with the use of the former alone or the local convergence with the latter alone could be overcome.