Product quality prediction method for mass customization

A method for predicting quality of workpieces produced in mass customization, the method comprising:

• obtaining, by a data collection device, data of a plurality of sets of first process parameters respectively corresponding to a plurality of first workpieces in a one-to-one manner, wherein the data of the sets of first process parameters are collected or used when a production system processes the first workpieces respectively;
• obtaining, by the data collection device, a plurality of first actual measurement values respectively belonging to the first workpieces in a one-to-one manner, wherein the first actual measurement values are generated by using a metrology tool to measure the first workpieces that have been processed by the production system;
• calculating, by a virtual metrology server, an average value of actual measurement values of all workpieces listed in a DS (determinative samples) set, thereby obtaining a first average actual measurement value, wherein the workpieces listed in the DS set are processed by the production system after the first workpieces have been processed, and the actual measurement values of the workpieces listed in the DS set are obtained by using the metrology tool to measure the workpieces that have been processed by the production system;
• sorting, by the virtual metrology server, the first actual measurement values;
• calculating, by the virtual metrology server, an absolute value of a difference between every two adjacent first actual measurement values that have been sorted;
• searching for, by the virtual metrology server, a plurality of breakpoints of the first workpieces, wherein each of the breakpoints is defined at the workpiece corresponding to one of the every two adjacent first actual measurement values of which the absolute value of the difference is greater than or equal to a breakpoint threshold value;
• grouping, by the virtual metrology server, the first workpieces into a plurality of groups in accordance with the breakpoints;
• calculating, by the virtual metrology server, an average value of actual measurement values of all workpieces in each of the groups, thereby obtaining a plurality of second average actual measurement values;
• respectively adjusting, by the virtual metrology server, the first actual measurement values of the first workpieces to a plurality of target measurement values, wherein each of the target measurement values is the actual measurement value of its corresponding first workpiece minus the second average actual measurement value of its corresponding group and then plus the first average actual measurement value of the DS set;
• building or retraining, by the virtual metrology server, a prediction model by using the data of the sets of first process parameters and the target measurement values of the first workpieces in accordance with a prediction algorithm;
• and
• predicting, by the virtual metrology server, a virtual metrology value of a next workpiece by inputting data of a set of process parameters belonging to the next workpiece into the prediction model, wherein the production system uses the data of the set of process parameters to process the next workpiece after the first workpieces are processed.