Member selection optimization device, member selection optimization method and member selection optimization program

摘要

PROBLEM TO BE SOLVED: To efficiently suppress manufacturing variation of a secondary battery by a conventional method. SOLUTION: In the member selection optimization device of the present invention, from the quality confirmation value of two or more elements for N lots, the predicted value of the characteristics of the constituents for all the patterns of the combination of two or more elements for N lots. The predicted value generation unit 11 generated by the predictor using the machine learning model, and the convergent values that determine M convergent values within the convergent value candidate range set within the distributed range of the predicted values for all patterns. The combination of the candidate group generation unit 12 and the elements corresponding to the predicted values close to each convergence value is selected so that all lots of each element are included and there is no overlap of lots, and M combination candidates are converged. It has a combination candidate generation unit 13 that generates a combination candidate with respect to a value, and a combination determination unit 14 that determines a combination candidate having the smallest variation in predicted values among a plurality of combination candidates as an optimum combination to be output. [Selection diagram] Fig. 1