A knowledge diagnostic system for product defects
The need to fulfill customer satisfaction and increase product quality has motivated many manufacturing firms to investigate and diagnose their product failure. To gain a correct and accurate diagnostic, the entire processing root must be recorded and controlled in every step of the manufacturing process. In this research, a prototype system has been developed for a tile manufacturing company to diagnose tile defects and to recommend actions for improvement. This system consists of two main components, the knowledge base and inference engine. The knowledge base has been developed by capturing data and information that are related to tile defects, such as symptoms, probable causes, types of defects, processes, sub processes, tile classifications, etc. On the other hand, the inference engine has been built by implementing the forward chaining and depth first searching methods to search for the causes of defects. The analysis proves that this system can help the workers in the company to diagnose tile defects and solve the problems. Besides this, the system can also help to share and transfer knowledge among the knowledge workers in the company.
| Attachment | Size |
|---|---|
| id_93.wmv | 4.76 MB |
Thanks for your question.
the answer is "Yes". In System Development Life cycle, we have done some system testing phase. We applied balckbox testing, whitebox testing and usebility testing to justify the system.
In Blackbox testing step: we found that the system are running base on use case analysis process and producing the correct output regards to the user requirements.
In Whitebox testing step: we are focusing on the coding, functional and method for the overall class in the system, and fortunately it rusns properly.
Usebility testing step: we presents the system and make some questionnaire to White Horse Ceramics Company's staffs in Johor Bahru Malaysia. We got possitive feedback as specially from knowledge workers. The users felt that the new system was a considerable improvements on their existing methods and increased their knolwedge related to the tile defects in manufacturing firm.
Dear Wongky,
Thank you for replying, you have done a good work on teting your system. But my question was wether your system can reason out the decisions it make. In other words, when it makes a decision, can it display the details that led to make that decision?
Dear Charles,
Thank you. At the first i have been worried about my answer while i interpret your question again and again.
Oke. For that question, actually my system did not provide that such features. A prototype system just gives the solution for the defect found in diagnostic section. Such as:
1. The system gives information about which part of manufacturing process and sub process that has or causes a problem.
2. The system produces information of probable symptoms and causes of defect found.
3. The system produces the grade classification of defect tile.
4. The system suggests recommendation action to solve the problems.
The final conclusion are given based on the answers that have been key in by users.
But my system did not provide reason out for every decision given it make. It was good suggestion for my future work. Thank you.
Hi, an interesting application on tile manufacturing. Could your system will provide the justifications for the decisions it make?