Krishna R. Pattipati - US grants

Human decision making and coordination strategies within a distributed organization are known to be less than optimal, especially as the workload under which the team must operate increases. This project will employ a normative-descriptive approach to systematically investigate team performance in such contexts. The fundamental tenet of the normative-descriptive theory is that motivated expert decision makers strive for optimality, but are constrained from achieving it by their inherent limitations and cognitive biases. The normative-descriptive theory employs normative solutions as a baseline, and modifies these solutions by placing psychologically interpretable constraints and structure on the team's cognitive processes to provide accurate predictions of team performance. These models, which are experimentally validated, provide a relevant basis for designing distributed database and communication subsystems that best support the needs of human decision makers. The working hypothesis put forward in this project is that human teams adapt their coordination strategies to workload demands. At low workload teams prefer to coordinate explicitly using communication channels. Under moderate workload teams rely on implicit coordination, exercising internal models to anticipate the needs of other team members. Under high workload off-line pre-planning replaces on-line coordination. This project will investigate, using the normative-descriptive approach, explicit and implicit coordination in the areas of team information processing, resource allocation, and task sequencing. The research on information coordination characterizes the amount of communication needed among team members for superior explicit coordination, and the role of feedback as a means of improving implicit coordination. The research on resource and task coordination focuses on when a leader is necessary to improve coordination, and on how coordination strategies change when different mixes of sequential, parallel, and multiple actions must be taken by the team. The research on database support addresses the questions of database modeling structures needed to support the requirements of different forms of coordination in an organization, and develops algorithms for dynamically (re)locating the data in the network for effective coordination.

This ongoing research effort on job shop scheduling explores three generic problems in the manufacturing industry. The first problem is to schedule products with bills of materials to improve on-time deliveries of products. The second problem pertains to the scheduling of facilities capable of batch processing several jobs concurrently, such as an oven or a plating facility. Since these facilities are often expensive and have long processing times, they constitute bottlenecks in many manufacturing systems. The scheduling of a flexible manufacturing cell (FMC) is the third problem for study. The purpose of this research is to obtain an efficient, effective and consistent solution methodology for scheduling general job shops. Through extensive collaboration, an optimization-based scheduling methodology using the Lagrangian relaxation technique has been developed. The method has many positive features, and has been tested at the Development Operations shop of a major aerospace manufacturer. The abilities to handle bills of materials, batch facilities, and FMC's, however, are of crucial importance for the method to be truly effective for general job shops. This research will thus advance the state of the art and practice of scheduling methodology, and enhance the profit-making ability of many manufacturing companies.

This award is made under the high performance connections portion of ANIR's "Connections to the Internet" announcement, NSF 98-102. It provides partial support for two years for a DS-3 connection to the vBNS. Applications include projects in studying educational outcomes of networked multimedia, networked-based monitoring and fault diagnosis, distributed services telemedicine, interconnecting distributed biological databases to establish a "virtual computational resource center"; network-based scheduling and supply chain coordination and networking controls for network edge multimedia appliances. Collaborating institutions include the Jackson Laboratory, the Institute of Genomic Research, Lawrence Livermore National Laboratory, Lawrence Berkeley National Laboratory, Los Alamos National Laboratory, Brookhaven National Laboratory, Oak Ridge National Laboratory, and the National Human Genome Research Institute, the Jefferson Laboratory at the University of Virginia and others.

The objectives of this research are to design a heterogeneous network of embedded systems so that faults can be quickly detected and isolated and to develop on-line and off-line fault diagnosis and prognosis methods. Our approach is to develop functional dependency models between the failure modes and the concomitant monitoring mechanisms, which form the basis for failure modes, effects and criticality analysis, design for testability, diagnostic inference, and the remaining useful life estimation of (hardware) components.

Over the last few years, the electronic explosion in automotive vehicles and other application domains has significantly increased the complexity, heterogeneity, and interconnectedness of embedded systems. To address the cross-subsystem malfunction phenomena in such networked systems, it is essential to develop a common methodology that: (i) identifies the potential failure modes associated with software, hardware, and hardware-software interfaces; (ii) generates functional dependencies between the failure modes and tests; (iii) provides an on-line/off-line diagnosis system; (iv) computes the remaining useful life estimates of components based on the diagnosis; and (iv) validates the diagnostic and prognostic inference methods via fault injection prior to deployment in the field. The development of functional dependency models and diagnostic inference from these models to aid in online and remote diagnosis and prognosis of embedded systems is a potentially novel aspect of this effort.

This project seeks to improve the competitiveness of the U.S. automotive industry by enhancing vehicle reliability, performance and safety, and by improving customer satisfaction. Other representative applications include aerospace systems, electrification of transportation, medical equipment, and communication and power networks, to name a few.

The proposed collaborative project with GM R&D center, Warren, MI seeks to develop hybrid model-based/data-driven/knowledge-based prognostic framework, and the associated prediction and inference algorithms, to detect and isolate incipient component degradations in coupled systems. The focus will be on chassis health determination (meaning the health status of steering, braking and suspension components, as well as functionalities that use these components such as the StabiliTrak system) to replaceable components (e.g., broken tie rod, worn out brake pads or blown tire, malfunctioning ABS control module). The diagnostic and prognostic framework will be validated on test fleet prior to implementing in the production versions.

Intellectual Merit: Existing prognostic and diagnostic algorithms employed in automotive systems tend to be component-centric. They often fail to provide correct diagnosis due to neglect of cross-subsystem failure propagation and unreliable tests. The proposed research seeks to overcome these limitations by explicitly modeling the cross-subsystem effects via a graphical model and by developing a multi-layer probabilistic reasoning process spanning sensed data → predicted features → diagnostic trouble codes → failure modes → replaceable components and subsystems. Modeling the time evolution of coupled component states as factorial hidden Markov models and the development of computationally-efficient inference algorithms in multi-layer graphical models is a novel aspect of the proposed effort. In addition, combining model-based, data-driven and knowledge-based approaches in a unified way to solve practical diagnosis and prognosis problems in the next-generation automotive vehicles is another contribution of this work.

Broader Impact: The proposed research improves the competitiveness of American automotive industry by reducing warranty costs, and enhancing vehicle availability and customer satisfaction. The PI plans to promulgate the results of this research to the broader industrial community via short courses, tutorials, conference presentations and journal manuscripts.

The goal of this multi-disciplinary project is to develop a simple, robust, generic, and scalable model-based and data-driven Fault Detection, Diagnosis and Prognosis (FDDP) process and the associated detection, inference and predictive analytics that are applicable to a variety of buildings. The research is motivated by the observation that buildings account for more than 40% of US energy consumption. Heating, Ventilation and Air Conditioning (HVAC) constitutes 57% of energy used in commercial and residential buildings, valued at $223B in 2009. About 20% of the energy consumed by HVAC is wasted due to abrupt faults (e.g., stuck dampers), performance degradations (e.g., air filter clogging), poor controls (e.g., biases in set points), and improper commissioning (e.g., poorly balanced parallel chillers). This project will develop FDDP methodologies for HVACs to improve equipment availability, lower energy and operating costs, extend equipment life, and enhance occupants' comfort. The FDDP process will be validated and evaluated by applying it to UConn's Tech Park Building; Duncaster, a life-care retirement community, located in Bloomfield, CT; and potentially to others. The project contributes to the vision of green and sustainable buildings equipped with cyber-physical substrata consisting of HVAC modules, networked sensors providing information on spatial and temporal distribution of occupants, smart building management systems providing situation awareness and decision support to human operators, and improved tenant comfort.