| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To construct the highly precise relevant model of quality and operation by relatively small number of division by automatically performing proper space division according to the distributing status of data such as the deviation or roughness of operation data in the case of dividing an operation factor space for analyzing the relevance of an operation factor and quality. <P>SOLUTION: The grouping of data is executed based on a distance between data in an operation factor space, and normal distribution functions approximating the data group are derived. Membership functions are calculated from the normal distribution, and soft division on which the boundary neighborhood of a region is superimposed is prepared, and a local quality prediction model in each region is calculated so that an overall quality analytic model can be constructed. Thus, the relevance of the operation and quality of a manufacturing process is precisely and quickly analyzed where a plurality of generation factors are immanent by analyzing the content of the prepared overall model. <P>COPYRIGHT: (C)2007,JPO&INPIT</p> |
