Intelligent Model Building

for Generalized Intelligent Grinding Advisory System

 

Factors Affecting Grinding Process Conditions

 The process output variables of grinding processes are governed by a number of operation parameters, material conditions as well as the machine parameters.    The following shows the schematic of the input-output relationship.    Therefore, it is very difficult to have one model describing all the relationships.

grinding_input_output.jpg (45238 bytes)



Autonomous Hierarchical Modeling of the Process

  In order to capture the input-output relationships effectively, the following hierarchical structure is used via the autonomous learning scheme developed.  This structure allows for model building based on a few designed experiments.

   

 
Results of the Generalized Modeling

    An illustrative example of the modeling result by the developed modeling scheme is shown for surface roughness conditions of a surface grinding process.   Overall, excellent predictions are achieved.

surface_roughness model results.jpg (27140 bytes)


Output surface of the Developed Fuzzy Basis Function Network Model

    An illustrative example of the maximum residual stress model developed using a fuzzy basis function network is shown below.  This illustrates that the model can represent the nonlinear relationships of unknown and uncertain processes.

surf_vwa_100.jpg (196839 bytes)

 

[ Welcome ] [ Research ] [ Education ] [ Laboratories ] [ Collaboration ] [ People ]

© 2000 Purdue ME Manufacturing Laboratories
Last modified by Web-master, 1 February 2001.