In modern industrial societies, exceptional standards of quality will be the crucial key which will determine the degree of success or prosperity of any business. Manufacturers always demand quality improvement. According to American Society for Quality, quality improvement is the reduction of variability. To achieve that, Statistical Process Control (SPC) is proposed for detecting possible causes. SPC involves three parts: discovery the process, eliminating assignable sources of variation and monitoring the ongoing process. Among them, statistical monitoring methods serve as the useful tool to improve the quality.

Advanced manufacturing refers to the use of new technology to improve processes. For example, 3D printing is a new technology of fabricating products directly from th...[ Read more ]

In modern industrial societies, exceptional standards of quality will be the crucial key which will determine the degree of success or prosperity of any business. Manufacturers always demand quality improvement. According to American Society for Quality, quality improvement is the reduction of variability. To achieve that, Statistical Process Control (SPC) is proposed for detecting possible causes. SPC involves three parts: discovery the process, eliminating assignable sources of variation and monitoring the ongoing process. Among them, statistical monitoring methods serve as the useful tool to improve the quality.

Advanced manufacturing refers to the use of new technology to improve processes. For example, 3D printing is a new technology of fabricating products directly from three-dimensional CAD models. The efficiency is independent of the complexity of product geometry. Another example could be Web-based manufacturing systems, which adopts Web technologies in various product development phases. It enables manufacturers to integrate the information collected during all development phrases to improve the product quality.

In this thesis, a statistical strategy is proposed for error compensation in fused deposition modeling (FDM), which is widely used in consumer-level 3D printing. The products fabricated using FDM suffers from the problem of low dimensional accuracy. First, this strategy attributes the dimensional inaccuracy into two sources: the positioning error and the shape deformation. Second, we transform the positioning error into design input error. Third, a compensation plan is computed for the overall shape deviation.

To integrate customer comments in electronic commerce platform into the Web-based manufacturing process for quality improvement. A quality-related aspect monitoring scheme is proposed. Initially, all comments are analyzed using feature level sentiment analysis techniques. Then a Random Forest model is adapted to select quality related aspects. Finally, a monitoring scheme is developed based on selected quality related aspect.

With rapid developments in advanced manufacturing, sensors are widely implemented in all industries. As a result, profile data is increasingly common in many industrial applications. Before we adopt SPC methods, discovery analysis is necessary to find the natural variations in collected profile dataset. In this thesis, a novel methodology is proposed to discover the natural variations in both spatial and temporal dimensions. A visualization mechanism is also proposed primarily assist engineers with graphically animating any variation.

All three topics involve real data examples, and the implemented methods are provided using the proposed methods. The three methods facilitate engineers in advanced manufacturing areas to both effectively monitor the processes as well as improve the quality of products.