非对称损失函数的质量特性值优化选择

Designing the Most Cost-effective Process Target Using Asymmetric Loss Function

质量特性目标值的选定是一个重要的命题,它是影响质量成本的一个重要因素。由顾客定义的特性目标值并不一定能使得总体的质量损失最小化。针对此问题,讨论了非对称的质量损失函数,在此基础上构造出质量特性目标值的优选模型,得到最优目标值。在模型的应用中表明:质量损失系数不变、质量特性左偏比右偏更有害的情况下,质量特性最优值与期望总损失随分布标准差的增大而增大;分布标准差不变的情况下,质量特性最优值与期望总损失相对质量损失系数具有一定的稳健性。

The determination of the optimum process target （mean） is one of the most important decisionmaking problems. The customer-identified target may not be the most cost-effective setting. This article develops an optimization model for determining quadric loss function. The example shows that the optimum setting of process target by using asymmetric when quality loss coefficient is fixed and the loss under target is worse, the larger the standard deviation, the larger the optimum process target and the expected total loss; when the standard deviation is fixed, the optimum process target and the expected total loss have robustness to the quality loss coefficient.