Multi-sensor integration using neural networks for predicting quality characteristics of end-milled parts. Part 2: Interaction effects in training parameters
- Univ. of Missouri, Rolla, MO (United States). Dept. of Mechanical, Aerospace Engineering and Engineering Mechanics
Artificial neural networks have been shown to have a lot of potential as a means of integrating multi-sensor signals for in-process real time monitoring of machining processes. However a lot of questions still remain to be answered on how to optimize the training parameters of neural networks during the training phase in order to optimize their subsequent performance, especially in view of the fact that the few published literature have made conflicting recommendations. This paper presents a systematic evaluation of the interaction effects of the training parameters: learning rate, momentum rate and number of hidden layer nodes on the performance of back propagation networks in predicting quality characteristics of end milled parts. Multi-sensor signatures (acoustic emission, spindle vibration, and components of the horizontal cutting force) acquired during end milling of 4140 steel and the corresponding measured quality characteristics (surface roughness and bore tolerance) were used to train the networks. The network performances were evaluated using four different criteria: maximum error, RMS error, mean error and number of training cycles. The results indicate that there are significant interaction effects between the training parameters that can adversely affect the networks performance. Optimum combinations of training parameters have been observed. The effects of various combinations of training parameters are presented.
- OSTI ID:
- 395220
- Report Number(s):
- CONF-940691-; ISBN 1-56676-171-9; TRN: IM9648%%343
- Resource Relation:
- Conference: 2. international conference on intelligent materials, Williamsburg, VA (United States), 5-8 Jun 1994; Other Information: PBD: 1994; Related Information: Is Part Of Second international conference on intelligent materials: Proceedings; Rogers, C.A.; Wallace, G.G. [eds.]; PB: 1410 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Final Technical Report - Applications of Machine Learning Techniques to Geothermal Play Fairway Analysis in the Great Basin Region, Nevada
Adaptive hyperparameter updating for training restricted Boltzmann machines on quantum annealers
Related Subjects
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATION
99 MATHEMATICS
COMPUTERS
INFORMATION SCIENCE
MANAGEMENT
LAW
MISCELLANEOUS
ON-LINE MEASUREMENT SYSTEMS
NEURAL NETWORKS
INDUSTRIAL PLANTS
STEELS
MILLING
AUTOMATION
MANUFACTURING
QUALITY ASSURANCE
COMPUTER NETWORKS
EXPERIMENTAL DATA
SURFACE PROPERTIES
PERFORMANCE
MILLING MACHINES
DATA PROCESSING
PROCESS CONTROL