Design structure matrix

Abstract: Systems engineering of products, processes, and organizations requires tools and techniques for system decomposition and integration. A design structure matrix (DSM) provides a simple, compact, and visual representation of a complex system that supports innovative solutions to decomposition and integration problems. The advantages of DSMs vis-a-vis alternative system representation and analysis techniques have led to their increasing use in a variety of contexts, including product development; project planning, project management, systems engineering, and organization design. This paper reviews two types of DSMs, static and time-based DSMs, and four DSM applications: (1) component-based or architecture DSM, useful for modeling system component relationships and facilitating appropriate architectural decomposition strategies; (2) team-based or organization DSM, beneficial for designing integrated organization structures that account for team interactions; (3) activity-based or schedule DSM, advantageous for modeling the information flow among process activities; and (4) parameter-based (or low-level schedule) DSM, effective for integrating low-level design processes based on physical design parameter relationships. A discussion of each application is accompanied by an industrial example. The review leads to conclusions regarding the benefits of DSMs in practice and barriers to their use. The paper also discusses research directions and new DSM applications, both of which may be approached with a perspective on the four types of DSMs and their relationships.

Abstract: Spatial-temporal graph modeling is an important task to analyze the spatial relations and temporal trends of components in a system. Existing approaches mostly capture the spatial dependency on a fixed graph structure, assuming that the underlying relation between entities is pre-determined. However, the explicit graph structure (relation) does not necessarily reflect the true dependency and genuine relation may be missing due to the incomplete connections in the data. Furthermore, existing methods are ineffective to capture the temporal trends as the RNNs or CNNs employed in these methods cannot capture long-range temporal sequences. To overcome these limitations, we propose in this paper a novel graph neural network architecture, Graph WaveNet, for spatial-temporal graph modeling. By developing a novel adaptive dependency matrix and learn it through node embedding, our model can precisely capture the hidden spatial dependency in the data. With a stacked dilated 1D convolution component whose receptive field grows exponentially as the number of layers increases, Graph WaveNet is able to handle very long sequences. These two components are integrated seamlessly in a unified framework and the whole framework is learned in an end-to-end manner. Experimental results on two public traffic network datasets, METR-LA and PEMS-BAY, demonstrate the superior performance of our algorithm.

Abstract: Spatial-temporal graph modeling is an important task to analyze the spatial relations and temporal trends of components in a system. Existing approaches mostly capture the spatial dependency on a fixed graph structure, assuming that the underlying relation between entities is pre-determined. However, the explicit graph structure (relation) does not necessarily reflect the true dependency and genuine relation may be missing due to the incomplete connections in the data. Furthermore, existing methods are ineffective to capture the temporal trends as the RNNs or CNNs employed in these methods cannot capture long-range temporal sequences. To overcome these limitations, we propose in this paper a novel graph neural network architecture, Graph WaveNet, for spatial-temporal graph modeling. By developing a novel adaptive dependency matrix and learn it through node embedding, our model can precisely capture the hidden spatial dependency in the data. With a stacked dilated 1D convolution component whose receptive field grows exponentially as the number of layers increases, Graph WaveNet is able to handle very long sequences. These two components are integrated seamlessly in a unified framework and the whole framework is learned in an end-to-end manner. Experimental results on two public traffic network datasets, METR-LA and PEMS-BAY, demonstrate the superior performance of our algorithm.