Data mining for improvement of product quality

Abstract

The assemble-to-order strategy delays the final assembly operations of a product until a customer order is received. The modules used in the final assembly operation result in a large product diversity. This production strategy reduces the customer waiting time for the product. As the lead-time is short, any product rework may violate the delivery time. Since quality tests can be performed on the stocked modules without impacting the assembly schedule, the quality of the final assembly operations should be the focus. The data-mining approach presented in this paper uses the production data to determine the sequence of assemblies that minimizes the risk of producing faulty products. The extracted knowledge plays an important role in sequencing modules and forming product families that minimize the cost of production faults. The concepts introduced in the paper are illustrated with numerical examples.

Assembly system synthesis using association rule discovery

“…Knowledge discovery models and particularly association rules discovery techniques were used successfully in many industrial applications [13][14][15][16][17][18][19][20][21][22][23]. For instance, in quality control, Da Cunha et al [13] studied the association between the assembly sequence and the likelihood of having defective products and utilized it in sequencing of modules and forming product families which minimizes the cost of production faults.…”

A Knowledge Dashboard for Manufacturing Industries

“…Traditional machine learning (ML) approaches for knowledge acquisition in manufacturing started to gain much attention only in recent years [3][4][5][6][7][8][9][10], mostly because the majority of the ML algorithms and tools require skilled individuals to understand the output of ML process [3]. However there has been some work on using traditional ML techniques for specific areas (such as fault detection, quality control, maintenance, engineering design, etc.)…”

“…Association rule mining algorithms were used to identify relationships among the attributes describing the data. [9] used association rule mining for detecting the source of assembly faults, thus improving the quality of assembly operations. [10] extracted association rules from historical product data to identify the limitations of the manufacturing processes.…”

“…A dashboard interaction paradigm has been chosen as it provides large amounts of information in one interface, without compromising on clarity [23] and it is an increasingly common visualisation paradigm thanks to its adoption by several wellknown websites like igoogle 9 and BBC 10 . Such an approach offers the possibility of dynamically choosing the best visualisation tool for the task in hand, as differently represented data can reveal different insights.…”

“…However there has been some work on using traditional ML techniques for specific areas (such as fault detection, quality control, maintenance, engineering design, etc.) employing classification [6,7], clustering [8] and association rule mining [9,10] algorithms [3][4][5]. Classification algorithms were used for categorising data into different classes, for example classifying defects in the semi-conductor industry [5].…”