Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process
Manufacturing cost for machining components is affected by the available machining parameters which include the selection of appropriate cutting material, cutting tools, and machining data of cutting speed, feed, and depth of cut. Computerized machining data systems have been classified into two...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2004
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/269/1/549481_t_fk_2004_72.pdf |
| _version_ | 1825943177342746624 |
|---|---|
| author | Ali Al-Assadi, Hayder M. A. |
| author_facet | Ali Al-Assadi, Hayder M. A. |
| author_sort | Ali Al-Assadi, Hayder M. A. |
| collection | UPM |
| description | Manufacturing cost for machining components is affected by the available machining
parameters which include the selection of appropriate cutting material, cutting tools, and
machining data of cutting speed, feed, and depth of cut. Computerized machining data
systems have been classified into two general types, the mathematical model and the
database model. The database model is based on the collection and storage of a large
quantity of data from laboratory experiments and workshop experience, which can then
simply retrieve recommended cutting speeds and feed. The most widely used source of
such data is the Machining Data Handbook (MDH) published by Metcut Research
Association, (1980). Although the handbook approach is often a logical and effective
solution to the requirement of machining data, but it has limitations.
The applications of computational intelligence in manufacturing, in particular, play a
leading role in the technological development of intelligent manufacturing systems. In
this study an intelligent learning system was developed to automate the collection of the
machining data used by the skilled machinist. The Machine Learning Method is utilized for this task, which gives the computer the ability to learn. Artificial Neural Network
(ANN) was selected from Machine Learning Algorithms to be the learning algorithm.
ANN is a computer-based simulation of the living nervous system which works quite
differently from conventional programming. The design network is trained by presenting
several target machining data that the network must learn according to a learning rule
(algorithm). In designing the network, a combination of back propagation or generalized
delta learning rule with sigmoid transfer function has been used.
The machining data available in MDH was used to train the designed network. One
cutting material (medium carbide steel) with its complete set of cutting tools (High Speed
Steel, Brazed Uncoated Carbide, Indexable Uncoated Carbide, and Coated Carbide)
discretized into 243 data sets was used in one training session for the designed network.
Building knowledge within the network was measured by calculating the total percentage
of error between target machining data and the outputs from the network during the
training process.
The process of building the machining data knowledge (training) was successfully
achieved. A Comparison between the learned target machining data and data from MDH
shows a low percentage of error. An Intelligent Learning System for the turning process
was developed. Visual C++ object-oriented programming language was used to build the
Intelligent Learning System for Turning. Live data can be fed into the system from
indirect way (Keyboard, Internet) or directly from machine to computer. The developed
system may open the door for automating the collection of machining data for all
manufacturing processes |
| first_indexed | 2024-03-06T06:52:45Z |
| format | Thesis |
| id | upm.eprints-269 |
| institution | Universiti Putra Malaysia |
| language | English |
| last_indexed | 2024-03-06T06:52:45Z |
| publishDate | 2004 |
| record_format | dspace |
| spelling | upm.eprints-2692015-08-06T03:36:35Z http://psasir.upm.edu.my/id/eprint/269/ Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process Ali Al-Assadi, Hayder M. A. Manufacturing cost for machining components is affected by the available machining parameters which include the selection of appropriate cutting material, cutting tools, and machining data of cutting speed, feed, and depth of cut. Computerized machining data systems have been classified into two general types, the mathematical model and the database model. The database model is based on the collection and storage of a large quantity of data from laboratory experiments and workshop experience, which can then simply retrieve recommended cutting speeds and feed. The most widely used source of such data is the Machining Data Handbook (MDH) published by Metcut Research Association, (1980). Although the handbook approach is often a logical and effective solution to the requirement of machining data, but it has limitations. The applications of computational intelligence in manufacturing, in particular, play a leading role in the technological development of intelligent manufacturing systems. In this study an intelligent learning system was developed to automate the collection of the machining data used by the skilled machinist. The Machine Learning Method is utilized for this task, which gives the computer the ability to learn. Artificial Neural Network (ANN) was selected from Machine Learning Algorithms to be the learning algorithm. ANN is a computer-based simulation of the living nervous system which works quite differently from conventional programming. The design network is trained by presenting several target machining data that the network must learn according to a learning rule (algorithm). In designing the network, a combination of back propagation or generalized delta learning rule with sigmoid transfer function has been used. The machining data available in MDH was used to train the designed network. One cutting material (medium carbide steel) with its complete set of cutting tools (High Speed Steel, Brazed Uncoated Carbide, Indexable Uncoated Carbide, and Coated Carbide) discretized into 243 data sets was used in one training session for the designed network. Building knowledge within the network was measured by calculating the total percentage of error between target machining data and the outputs from the network during the training process. The process of building the machining data knowledge (training) was successfully achieved. A Comparison between the learned target machining data and data from MDH shows a low percentage of error. An Intelligent Learning System for the turning process was developed. Visual C++ object-oriented programming language was used to build the Intelligent Learning System for Turning. Live data can be fed into the system from indirect way (Keyboard, Internet) or directly from machine to computer. The developed system may open the door for automating the collection of machining data for all manufacturing processes 2004-09 Thesis NonPeerReviewed application/pdf en http://psasir.upm.edu.my/id/eprint/269/1/549481_t_fk_2004_72.pdf Ali Al-Assadi, Hayder M. A. (2004) Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process. Masters thesis, Universiti Putra Malaysia. Machine learning Turning |
| spellingShingle | Machine learning Turning Ali Al-Assadi, Hayder M. A. Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title | Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title_full | Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title_fullStr | Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title_full_unstemmed | Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title_short | Development of Machine Learning Algorithm for Acquiring Machining Data in Turning Process |
| title_sort | development of machine learning algorithm for acquiring machining data in turning process |
| topic | Machine learning Turning |
| url | http://psasir.upm.edu.my/id/eprint/269/1/549481_t_fk_2004_72.pdf |
| work_keys_str_mv | AT alialassadihayderma developmentofmachinelearningalgorithmforacquiringmachiningdatainturningprocess |