Janis Terpenny

Knowledge interaction with genetic programming in mechatronic systems design using bond graphs

This paper describes a unified network synthesis approach for the conceptual stage of mechatronic systems design using bond graphs. It facilitates knowledge interaction with evolutionary computation significantly by encoding the structure of a bond graph in a genetic programming tree representation. On the one hand, since bond graphs provide a succinct set of basic design primitives for mechatronic systems modeling, it is possible to extract useful modular design knowledge discovered during the evolutionary process for design creativity and reusability. On the other hand, design knowledge gained from experience can be incorporated into the evolutionary process to improve the topologically open-ended search capability of genetic programming for enhanced search efficiency and design feasibility. This integrated knowledge-based design approach is demonstrated in a quarter-car suspension control system synthesis and a MEMS bandpass filter design application.

Interactive evolutionary solution synthesis in fuzzy set-based preliminary engineering design

This paper describes an interactive evolutionary approach to synthesize component-based preliminary engineering design problems. This approach is intended to address preliminary engineering design as an evolutionary synthesis process, with the needs for human-computer interaction in a changing environment caused by uncertainty and imprecision inherent in the early design stages. It combines an agent-based hierarchical design representation, set-based design generation, fuzzy design trade-off strategy and interactive design adaptation into evolutionary synthesis to gradually refine and reduce the search space while maintaining solution diversity to accommodate future changes. The fitness function of solutions employed is not fixed but adapted according to elicited human value judgment and constraint change. It incorporates multi-criteria evaluation as well as constraint satisfaction. This new approach takes advantage of the different roles of computers and humans play in design and optimization. The methodology will be applicable to general multi-domain applications, with emphasis on physical modeling of dynamic systems. An automotive speedometer design case study is included to demonstrate the methodology.

Toward a multi-agent information management infrastructure for product family planning and mass customisation

Development of complex new products requires numerous decisions by many individuals and groups, which are often geographically and temporally distributed. There is a need to share and coordinate distributed resources and synchronise decisions. Recent advances in Information Technology (IT) pose an untapped potential in assisting the capture, storage, retrieval, and facilitated use of product development information. By sharing assets such as components, processes, and knowledge across a family of products, companies can efficiently develop differentiated products and increase the flexibility and responsiveness of their product realisation process. This paper describes a recent effort in realising an information management infrastructure for product family planning and platform customisation. Particular focus is on current research thrusts:

HARNESSING PRODUCT COMPLEXITY: STEP 1—ESTABLISHING PRODUCT COMPLEXITY DIMENSIONS AND INDICATORS

With todays level of market competition and demand for diverse product offerings, more companies are now pressured to increase product variety as a strategy to maintain and increase market share. In addition to its associated direct costs, variety is considered the main source of product complexity, which has been proven to negatively impact product development time, productivity, costs, and customer satisfaction. Although several researchers have tackled the issue, product complexity continues to be a theoretical concept with different definitions and measurements established based on research area, scope, and objective. This lack of a unified approach has made it difficult for companies to take full advantage of existing research to manage the impact of product complexity. This article introduces five main dimensions of product complexity based on identifying different complexity sources in product design, development, manufacturing, assembly, and supply chain, and on understanding the impact of these sources on different direct and indirect costs. Establishing the dimensions of product complexity is an essential first step in developing a unified product complexity metric to be used as a support tool to improve product design and systematically manage product complexity.

Development and Utilization of Ontologies in Design for Manufacturing

There are many challenges associated with the design and realization of fast changing highly customized products. One promising approach is to implement design for manufacturing (DFM) strategies aimed at reducing production costs without compromising product quality. For manufacturers doing business in a globally distributed market place, effective reuse and sharing of the DFM knowledge in a collaborative environment is essential. In recent years, ontologies are increasingly used for knowledge management in engineering. Here, ontology is defined as a formal specification of domain knowledge that can be used to define a set of data and structure that enables experts to share information in a domain of interest, to aid information reasoning, and to manage and reuse data. The primary goal of this paper is to put forward the process of ontology development and utilization for DFM and to study the most important phases in the process, including: the concept categorization and class hierarchy development, slot categorization and development, identification and realization of relations among slots, and methods to support knowledge capture and reuse. Four cases are presented to illustrate the promising use of a DFM ontology. These cases prove that the DFM ontology and the process of ontology development and utilization for the DFM can facilitate the reuse of existing data, find the inconsistency and errors in data, reduce the work associated with populating the knowledge base of the ontology, and help designers make decisions by considering complex technical and economical criteria.

A Comparative Study on Machine Learning Algorithms for Smart Manufacturing: Tool Wear Prediction Using Random Forests

Manufacturers have faced an increasing need for the development of predictive models that predict mechanical failures and the remaining useful life (RUL) of manufacturing systems or components. Classical model-based or physics-based prognostics often require an in-depth physical understanding of the system of interest to develop closedform mathematical models. However, prior knowledge of system behavior is not always available, especially for complex manufacturing systems and processes. To complement model-based prognostics, data-driven methods have been increasingly applied to machinery prognostics and maintenance management, transforming legacy manufacturing systems into smart manufacturing systems with artificial intelligence. While previous research has demonstrated the effectiveness of data-driven methods, most of these prognostic methods are based on classical machine learning techniques, such as artificial neural networks (ANNs) and support vector regression (SVR). With the rapid advancement in artificial intelligence, various machine learning algorithms have been developed and widely applied in many engineering fields. The objective of this research is to introduce a random forests (RFs)-based prognostic method for tool wear prediction as well as compare the performance of RFs with feed-forward back propagation (FFBP) ANNs and SVR. Specifically, the performance of FFBP ANNs, SVR, and RFs are compared using an experimental data collected from 315 milling tests. Experimental results have shown that RFs can generate more accurate predictions than FFBP ANNs with a single hidden layer and SVR. [DOI: 10.1115/1.4036350]

Principles for Managing Technological Product Obsolescence

Products evolve to accommodate competitive market pressures, rapid rates of technology change, and constant improvements in performance and functionality. While adding functionality and value, the fast moving technologies also make products obsolete quickly. One of the primary reasons for product obsolescence is technological obsolescence which results when consumers are attracted to functions in newer models of products that are more technologically advanced. One way to deal with problem is ldquopiggybacking,rdquo a strategy that enables renewed functionality of a technologically obsolete product through the integration or add-on of a secondary device or component. Not to be confused with upgrading strategies, piggybacking requires a device that fits adjacent to, upon, or within the existing product architecture. Piggybacking is an attractive strategy for consumer electronic products that are particularly prone to technological obsolescence as it offers a means to accommodate fast and slower changing technologies within a single product. Currently, piggyback products are realized with ad hoc methods that rely on the experience and intuition of the designer, often applied inconsistently and not well known by less experienced designers. In this paper, a set of formal principles is presented for guiding the design of piggyback products. These principles are derived from the results of an empirical study of 72 different products. As part of the study, various products are analyzed with a dissection tool with representative principles derived from the data. The utility of these principles is demonstrated via the conceptual design of a novel piggyback products.

APPLICATION OF A GENETIC ALGORITHM TO CONCEPT VARIANT SELECTION

This paper outlines a framework for applying a genetic algorithm to the selection of component variants between the conceptual and detailed design stages of product development. A genetic algorithm (GA) is defined for the problem and an example is presented that demonstrates its application and usefulness. Functional modeling techniques are used to formulate the design problem and generate the chromosomes that are evaluated with the algorithm. In the presented example, suitable GA parameters and the break-even point where the GA surpassed an enumerated search of the same solution space were found. Recommend uses of the GA along with limitations of the method and future work are presented as well.Copyright

A hybrid ontology approach for integration of obsolescence information

Information sharing among distributed obsolescence management systems is a challenge because of the heterogeneity of data (data with different forms and representations). Indeed, this is the main hurdle that exists for current tools managing product obsolescence. This paper presents a hybrid ontology approach for the integration of obsolescence information that combines a global ontology that provides a shared vocabulary for the specification of the semantics of obsolescence domain knowledge, along with local ontologies that describe structures of multiple data sources distributed in various obsolescence management tools. A procedure is provided for mapping local ontologies to the global ontology by quantifying relationships between classes and identifying groups of classes with a clustering method. Ontologies and rules of identifying relationships are realized with OWL (Web Ontology Language) and SWRL (Semantic Web Rule Language). With the application of the hybrid ontology approach, a unified view of data is provided to support decision making for efficient obsolescence management and a structure where new sources of information can be easily added with little modification in the future.

Cooperative body–brain coevolutionary synthesis of mechatronic systems

Abstract To support the concurrent design processes of mechatronic subsystems, unified mechatronics modeling and cooperative body–brain coevolutionary synthesis are developed. In this paper, both body-passive physical systems and brain-active control systems can be represented using the bond graph paradigm. Bond graphs are combined with genetic programming to evolve low-level building blocks into systems with high-level functionalities including both topological configurations and parameter settings. Design spaces of coadapted mechatronic subsystems are automatically explored in parallel for overall design optimality. A quarter-car suspension system case study is provided. Compared with conventional design methods, semiactive suspension designs with more creativity and flexibility are achieved through this approach.