By defining this resource quality objectives, the combination evaluation model based on principal component analysis determining objective weight and analytic network process determining subjective weight was presented to solve the problems of combinatorial optimization arising from the dimensions' differences between manufacturing grid and resource allocation and distributed manufacturing resource selective preference. In the final model analysis of grinding resources comprehensive evaluation, grinding resources comprehensive evaluation index system is simplified by principal component analysis, and ordering vector of candidate resource is formed by analytic network process, then objects will be comprehensive evaluated through the comprehensive evaluation model we've got, from which the result avoids the limitation of single evaluation method and shows the practicability and effectiveness of this method.