Aditia Ginantaka
Description
Traceability is the ability to verify the history and location of a food product, thus providing information on each supply-chain actor, who the immediate supplier is, and to whom the product was sent. The information system approach has been used to manage and integrate all such information by collecting, storing, then retrieving data and information about the product from earlier stages of the production process. Besides documentation and information sharing, such traceability information systems can also support timely resolution of customer complaints. This paper presents modelling of routing and handling time prediction using a fuzzy associative memory (FAM) method. As a first response to customers, information about the time required to resolve an issue can be provided after the source of a product defect has been traced. With regards to handling, traceability can assist with several issues, e.g., product replacement, product recalls based on retrieval of the contact numbers of affected customers on a recall list , and inspections at each production unit to ensure food safety standards. Based on such activities, it is assumed that the handling time will be affected by the size of the product inventory that can be used to replace a defective product, the amount of product that must be recalled from the market, and the time required internally for the inspection process, which is set as the FAM input variable. A FAM is a set of fuzzy-set pairs (A, B) that maps an input vector fuzzy set A to an output vector fuzzy set B. Our experiments show that, from such a FAM formulation, one can obtain 27 rules. The FAM will encode a fuzzy-set pair (A, B) to obtain matrix memories, denoted by M. As the prediction result, the matrix B can be obtained from the computational matrix A and the matrix M; For instance, in case of a contamination incident in a fish product with inventory conditions of as much as 4 tonnes, the product recall amounts to 21 tonnes and the inspection will take 25 h, while the results of the computational experiment show that the total handling time for this case will be 66 h with low error rates.