Intelligent modeling and optimization of grinding processes

Cheol Won Lee, Purdue University

Abstract

The objective of this study is to develop and implement a methodology for modeling complex grinding processes and finding optimal process conditions to meet a general class of process requirements. In order to achieve these goals, novel modeling schemes and optimization methods based on fuzzy logic, neural networks, and evolutionary algorithms (EA) are developed. A hierarchical structure which consists of radial basis function networks (RBFN), fuzzy basis function networks (FBFN), and analytical equations is proposed to construct a comprehensive model of the grinding process. Two new algorithms, orthogonal least-squares learning using genetic algorithm (OLSGA) and adaptive least-squares (ALS) algorithm, based on the least-squares method and genetic algorithm (GA), are proposed for autonomous learning and construction of RBFN's and FBFN's, respectively. The first approach taken in this study for optimization of a grinding processes is based on the improved fuzzy rule-based optimization scheme in the generalized intelligent grinding advisory system (GIGAS). Next, a new approach based on model-based optimization scheme is also proposed in order to deal with a more general class of grinding optimization problems. Since grinding optimization can be considered as constrained nonlinear optimization problems with mixed-integer variables and time-varying characteristics, a novel algorithm based on the evolution strategies (ES) is proposed as a solution-finding method. The proposed modelling scheme is first applied to process control of a disc grinding process in the computer industry. A model for the material removal rate is constructed using the process data collected from the actual operation, and an adaptive controller based on the model is designed and implemented to achieve the required dimensional accuracy of ground parts. The potential improvement of process capability via adaptive control is demonstrated. Next, to implement the model-based optimization scheme for surface grinding processes, process models for grinding force, power, surface roughness, and residual stress are developed through designed experiments. Case studies are performed to demonstrate the efficacy and versatility of the proposed optimization method with various optimization objectives including minimization of grinding cost, minimization of cycle time, and process control. Finally, the optimal process conditions determined by the optimization scheme are validated by experimental results.

Degree

Ph.D.

Advisors

Shin, Purdue University.

Subject Area

Mechanical engineering|Artificial intelligence

COinS