影响注塑模具成本的因素很多，而只能选择最主要的因素作为网络的输入。这是由于每一个因素对应神经网络的一个输入，输入增多会提高网络的复杂性；并且，输入越多，为了达到要求的精度，需要的训练样本就越多，而训练样本往往是十分短缺的，这就要求应该在满足估算精度的要求下尽量减少输入，选择最主要的因素作为网络输入

制件在分型面一侧成型的直线类，输入为0.1；制件在分型面两侧成型的直线类，输入为0.3；分型面的台阶类，输入为0.5；分型面的曲线类，输入为0.7；分型面复杂综合类，输入为0.9；

there are many factors that effects the cost of injection modling,but we should choose the most important factor as the network input.this is because each of the factors refers to a input of the neural network.at the same time,the more we input the more complex the network gets and the more precise it requirs as well as the more exercise sample.however,because of the exercise sample shortage,it requirs a choice of the most important factor of the network input to ruduce the input under a requaired estimating precision level.....
后面那段话真是专业阿，不会老哈