An Efficient and Effective Fuzzy Collaborative Intelligence Approach for Cycle Time Estimation in Wafer Fabrication

Abstract

A common characteristic of fuzzy collaborative forecasting methods is the establishment of an interval estimate that is guaranteed to include the actual value. Such a characteristic is crucial to various planning purposes because it reduces the risk of incorrect forecasting. A new fuzzy collaborative forecasting method is proposed in this study to estimate the cycle time of a job in a wafer fabrication factory; this is a critical task for managing the fabrication of wafers. The proposed fuzzy collaborative forecasting method uses a fuzzy back propagation network that effectively improves the precision of estimation. The accuracy is enhanced through expert collaboration. The performance of the proposed methodology was selectively compared with those of two existing fuzzy collaborative forecasting methods by applying them to a real case. The results confirmed that the proposed methodology substantially enhanced the performance in estimating the cycle time of a job.

Ubiquitous Hotel Recommendation Using a Fuzzy-Weighted-Average and Backpropagation-Network Approach

“…Subsequently, five linguistic terms−very unimportant (), unimportant (), moderate (), important (), and very important ()—were chosen by a traveler to express the importance (or weight) of a hotel attribute. Each of the linguistic terms is mapped to a triangular fuzzy number (TFN); for example, following the recommendation by Chen and Wang:…”

“…Subsequently, five linguistic terms−very unimportant ( V U), unimportant ( U ), moderate ( M), important ( I ), and very important ( V I )-were chosen by a traveler to express the importance (or weight) of a hotel attribute. Each of the linguistic terms is mapped to a triangular fuzzy number (TFN) 22,23 ; for example, following the recommendation by Chen and Wang 24 : Calculations based on TFNs are relatively simple. By contrast, applications of other complex types of fuzzy numbers may lengthen the computation process and do not necessarily lead to improved performance.…”

Industrial big data–based scheduling modeling framework for complex manufacturing system

“…Meidan et al 29 successively adopt maximal conditionally mutual exclusion method and selective naive Bayesian classifier for feature selection, and extract the most important 20 out of 182 factors which affect silicon wafer processing cycle time, and the prediction accuracy is effectively increased by nearly 40%. Chen and colleagues [30][31][32][33] then improve their prediction methods and further deploy them in a cloud computing environment.…”

An Approach for Multipath Cloud Manufacturing Services Dynamic Composition

“…ξ min = ξ max 20 (12) where Q is a constant value and CWS best−so−far is the optimal solution found so far.…”

“…Driven by market demands, the manufacturing industry is undergoing significant transformations. With the rapid development of information technology, such as web service, 1-4 cloud computing, [5][6][7][8] manufacturing technology, 9 manufacturing grid, 10 Internet of things, 11,12 and other networked manufacturing modes, cloud manufacturing (CMfg) is proposed and developed successfully. [13][14][15] The main purpose of CMfg is to provide user with on-demand, always-ready, high-quality and low-consumption service, which is available for product design, manufacturing, testing, simulation and maintenance, and other manufacturing lifecycle process.…”