|
Nowadays, most faculties pre-estimate the types and total amount of material they need for manufacturing by executing Material Requirement Planning (MRP) system. If relevant input data is not well prepared before the system executes, the final suggestion about material purchasing must be wrong. For general manufacturers, the method to solve the above problem is to ask a senior worker to look out and take in charge of the illegitimate part of the reports. However, it is difficult to groom an administrator to control each detail, since checking the reports is a heavy job. Therefore, it is quite necessary to computerize the mission. This research is taking advantage of Fuzzy Logic and Artificial Neural Networks to construct a prototype of a Job-shop inventory management warning system. After taking the MRP reports as input, it can detect the illegal part in the reports automatically, and provide warning messages immediately. In addition, the capability of explanation can provide the possible causal reasons of the detected problem. Thus, the warning system can make managers more easier to correct the found problems and decrease the burden of the business. This prototyping system trained and checked with the 200 data pairs provided by the Formosa Tai Rank Industrial Corp. primarily. As the result of the above experiments, the system can demonstrate more than 90% debugging capability successfully with such a small amount of data. And it can also illustrate the capability of explanation which matches the intuitive interpretation of human beings. In a conclusion, the prototyping system could be regarded as a useful material management warning system.
|