A Belief-Desire-Intention Modeling Approach for Decision Support in Inspection-Oriented Quality Management

Description

Product quality inspection, carried out to determine whether or not a product conforms to quality requirements, is an important instrument to assure product quality and is widely adopted in industry practice. However, errors can happen when a supplier deceives in the inspection to conceal the real quality through producing a “conforming quality” that satisfies the requirement. An inspection error may lead to serious consequences. For example, in 2008, some suppliers for Sanlu, a well-known Chinese dairy manufacture, diluted milk for profits and added melamine to dupe an inspection for determining protein content, affecting some 294,000 infants and killing six. This tragic scandal has revealed a serious problem: the quality inspection processes are lack of effective methods to avoid inspection errors when suppliers deceive in order to maximize their profit.The solution to this problem relies on incorporating domain knowledge into the inspection policies that guide an inspection process. However, domain knowledge cannot be captured in the traditional mathematical or statistical models used in quality management. To fill this void, we propose a belief-desire-intention modeling approach to factor the domain knowledge the inspection environment and the inspection capability of various measurements, and to estimate the production behavior of supplier through logic reasoning when analyzing the risks for inspection errors. The proposed model can be used as a foundation to design intelligent systems to monitor and detect quality problems timely and effectively. To validate the approach, a prototype system is to be developed based on the proposed belief-desire- -intention model to support flexible quality inspection policy decision making and laboratory experiments will be conducted to examine if such an intelligent system can enhance the decision outcome of quality inspection.This research will have significant contributions for both theory and practice. First, this research will provide a novel methodology to facilitate decision making on quality inspection policies. Second, by taking account of the domain knowledge regarding inspection methods and processes, our modeling approach can significantly reduce inspection errors in quality management and thus our research will have important economic implications for companies worldwide. Third, as a measure of technology transfer, we will make our prototype system available over the Internet and accessible to quality control managers, quality inspection decision makers, and other professional consultants.