Computational Intelligence & Design Informatics Laboratory

The goal of this project is to advance a virtually guided Resistance Spot Welding (RSW) weldability qualification environment. VRWP will allow an OEM and suppliers to rapidly converge on the feasibility of weldment designs during the early development stages of new product designs. Currently, system integrators/OEMs working with products having metallic structures rely on material suppliers and testing service companies to conduct the actual physical testing of materials for new weldment designs. However, without more timely delivery of test results, the optimal selection of new materials, processes, and related designs is being hindered. When new (or new combinations of) materials are considered for an assembly, industry often requires new physical tests or numerical simulations (e.g., finite element analysis). Data-driven weldability prediction could improve product design efficiency but is being underutilized because of existing data inconsistences. RSW processes and parameters are complex (e.g., coating condition, surface roughness) and give rise to significant data inconsistences, a well-known reliability issue. In this project, the Wayne State team realizes a reliable RSW weldability prediction tool by conducting the following tasks performed in an iterative manner: 1) physical testing for emerging Advanced High Strength Steel (AHSS) materials, 2) construction of a sharable weldability knowledge representation and databases, 3) implementation of weldability prediction methods capable of reducing the effects of data inconsistency, and 4) provision of a web-based weldability prediction platform.

Funded by DMDII (Digital Manufacturing and Design Innovation Institute) & Ford Motor Company

Wayne State University, Penn State University, and Oregon State University are collaborating to develop a web-based sustainable lifecycle engineering design education portal called CooL:SLiCE (Constructionism in Learning: Sustainable Life-Cycle Engineering). CooL:SLiCE is a cyber-learning platform that takes a constructionist approach to provide an interactive learning environment and enable engagement in sustainable design and analysis. The portal consists of 3D online CAD and visualization tools, a design architecture analysis and supplier selection tool to examine product module selection impacts, and a manufacturing analysis tool to compare the environmental impacts of current and new manufacturing process selections. The tools seek to cover as much of the product and process design as possible, ranging from product architecture design, to supply chain analysis, to sustainability analysis. CooL:SLiCE is being developed by three universities, and each of these universities are making contributions in different fields of expertise.

Funded by the National Science Foundation

Proper design considering the Customer Service Life Cycle (CSLC) has been emphasized. The CSLC describes services that a customer receives throughout a product lifecycle. The lifecycle stages include Requirements, Acquisition, Ownership, and Retirement. However, the current design analytics systems often ignore the costs associated with the Ownership and Retirement stages. For example, maintenance cost is pointed as one of the main cost factors of medical devices; however, it is often not captured in the current practices. It is usually hidden to the activities of processing and reprocessing the devices. Extracting hidden cost information scattered in the unstructured data repositories such as legacy databases and data warehouses can reduce significantly the access of the cost information related to the devices. One of the main challenges to mechanical systems including reusable medical equipment is difficulty in identifying the relevance of cost factors of maintenance activities for specific devices that are recorded into the database. Extracting the related cost factors from unstructured databases for a device and forecasting the further costs for system acquisition using design and process characteristics requires significant research efforts. Another challenge is the difficulty to identify their implicit relationships between the factors. To extract proper cost factors from unstructured data repositories, this research aims to identify systematically the relationship between design and process characteristics and cost factors. It is achieved by first, defining the taxonomy of design and process characteristics, and second, finding the explicit and implicit cost factor from historical data (which is potentially linked with design and process characteristics). After conducting aforementioned processes, proper learning algorithms should be developed to predict the device cost by learning the relations between the factors. The figure included in this article illustrates the overall framework of the design analytics system. The web-based design analytics system for sharing medical equipment’s total cost assessment information can potentially support purchase and retirement decisions and other comparative analyses. Trends from recorded traces of device maintenance, replacement, failure, and other incurred costs can be analyzed and displayed in the developed design analytics system. This system provides two modules: 1) a module for activity based costing from text analysis and data mining that extracts cost information from legacy data warehouses, and 2) a visualization module to provide users with a conceptual understanding of the attributes associated with the decision information. This information is useful for medical device tactical and strategic planning.

Funded by the Department of Veterans Affairs & the National Science Foundation

As product design and development becomes more knowledge-intensive and collaborative, research in the areas of knowledge capture, retrieval, accessibility, and reusability becomes increasingly important. Heterogeneous tools and multiple designers are frequently involved in modern product development, and designers often use their own terms and definitions to represent a product design. Thus, to efficiently share design information among multiple designers, a designer's intentions should be persistently captured and the semantics of the designer's terms and intents should be interpreted in a consistent manner. The semantic assembly design system, Assembly Design Browser, and the semantic assembly model provide an understanding of assembly geometry and its physical effects. This system is unlike current solid modelers and simulation software that provide incomplete product definitions and are not able to act according to the semantic content of the models. In this new design system, ontologies for assembly serve as a formal, explicit specification of a shared conceptualization of assembly design. Mereotopological formal ontology and standard ontology technologies differentiate and share morphological characteristics of products.

Funded by the National Science Foundation Center for e-Design & the Ministry of Knowledge and Economics (MKE), Korea

This research highlights the complexity of semantic design decision rules. A product design can be semantically represented in ontology; however, the computational complexity of semantic reasoning is a very sophisticated and time-consuming task. By enhancing the design model with assembly rules, computer-aided systems will be more easily able to understand and to discern joining types. Current design models have been restricted in terms of any systematic interpretation of joining types. In order to reduce the computational complexity of design rule reasoning, the disparate attributes algorithm addresses a new concept of inapplicable information, which is significantly different from the traditional missing information or existing unavailable information. Missing information considers the values of missing attributes as a subset of existing attribute values. Unavailable information defines the values as unknown. Inapplicable information, however, means that some attributes are relevant to a certain rule and are not to the other rules simultaneously. This inapplicable information is often indicated in semantic assembly design rules.

Funded by the Ministry of Knowledge and Economics (MKE), Korea

Current product development processes still include unintended feedback due to insufficient product design knowledge. Current design knowledge support systems focus on the explicit search technique (e.g., matching keywords and file names, specific indices) and require manual input to incorporate the designer's knowledge. Such drawbacks should be managed to implement and utilize design knowledge support systems. To systematize the knowledge management process for design systems, a prerequisite is to capture and evaluate ever-evolving causal design knowledge. A causal network based design knowledge management system provides capabilities of knowledge capture from domain experts, systematic knowledge acquisition of current working engineering knowledge, knowledge transforming to better knowledge representation and knowledge network evaluation. The Knowledge Network Object Evaluation System (KNOES) is able to evaluate the causal knowledge network. For the network evaluation, a degree of causal representation (DCR)-based knowledge network evaluation method is developed. In this method, causality and network connectivity are used for the causal knowledge network with weighted vertices, and weighted network connectivity for a network with weighted edges.

Funded by the Ministry of Knowledge and Economics (MKE), Korea

The research for RME (Reusable Medical Equipment) design evaluation develops an evaluation method for assessing the level of reusability based on Design for Reusability (DFR) principles and index by considering critical design factors of RME and its sub-components. The current research efforts that address medical device design issues do not provide practical solutions to reduce the contamination risk of RME. Furthermore, the current sterilization methods have been developed without considering critical factors with respect to design characteristics. For this research, three tasks are conducting that are to design an evaluation method to evaluate reusable medical equipment (RME) samples such as several types of endoscopes, collected from the John Dingell VA Hospital in Detroit and other VA Hospitals in Midwest region, to develop index and scoring scheme can help VHA to identify types of RME equipment with high risk of infections due to human errors in reprocessing processes, and to develop DFR principles and index for RME design assessment and RME evaluation framework. This research impacts to the application for cleaning technicians, to the SPD operation reconfirmation considering design issue, to the decision support system for purchasing device decision, and the SPD operation automation.

Funded by the U.S. Department of Veterans Affairs, VERC

In a globalized product development environment, heterogeneous departments should interact, and vehicle information should be available continuously and immediately to identify new vehicle systems' needs, although both of those statements are rarely true in practice. Today's vehicle development environment sees very tight pressure both from customers' needs and governmental regulations. In this research, a new web-based system-Weight Efficiency Metrics Application (WEMA)-to support product engineers' vehicle weight targeting and cascading has been developed. WEMA was implemented using web scripting technology, which allows collaborative information collection and sharing within the corporation. For testing and benchmarking the application, WEMA was installed on an OEM company's internal network and linked to its corporate database.

Funded by Ford Motor Company

The research examines the effectiveness and accuracy of procuring wheeled mobility and seating devices for individuals with mobility impairments through the use of a telerehabilitation consultation model. The current selection and evaluation process, which is based on in-person assessment, is often unavailable to patients in rural areas. It is also time-consuming due to lack of expertise in wheeled mobility and seating interventions. Additional limitations on current wheelchair selection and prescription practices include clinicians' knowledge limitations, inconsistencies in decisions, and practitioners' information overload. This research was very practical and clinician/practitioner oriented. The developed system, RWS-A, was well adapted by expert clinicians and inexperienced clinicians, and it was also utilized to educate clinical students in rehabilitation science and engineering. It was recognized as an innovative clinical processes for OT/ATP practitioners by the University of Pittsburgh Press.

Funded by the National Institute of Disability and Rehabilitation Research (NIDRR) Department of Education

Welding task sequencing is a prerequisite in the off-line programming of robot arc welding. The purpose of welding task sequencing is to decide the welding order of different weld seams. The welding sequences obtained are utilized for robot and positioner motion planning and welding operation planning. Even though several criteria have been proposed, welding task sequencing has traditionally been performed using the experience and knowledge of welding experts in the robot welding industry. Therefore, a systematic approach to determine efficient welding task sequencing with considerations of productivity and quality of welding is a must. The goal of this project is to optimize welding task sequencing (WTS) while considering the deformation behavior of welded constructions and welding time. The main factor to be considered is the quality of the assembly after welding, which is measured by the deformation behavior at pre-defined critical locations. Other factors include the levels of productivity to optimize welding time and robot and resource utilization. Also, this project investigates the use of alternative joint design and joining methods to reduce harmful effects of welding and to increase joining quality.

Funded by the National Science Foundation