The European Journal on Artificial Intelligence

Model‐based supervision developed by systems analysts has become an acknowledged supervision aid, ensuring early detection of malfunctions and thereby allowing control of the availability and vulnerability of a process facility. However, it is associated with diagnostics of the process itself, and not of the process control situation, which is the veritable subject of supervision. The operator, facility, control triplet determines a complex situation that must be considered from multiple viewpoints beyond knowledge of the single behavioral model usually advocated in process control approaches. Representing different aspects of process control situation from multiple viewpoints notably allows the on line selection of the behavioral models relevant to the observed situation. Given the size of the application, it was essential not only to structure the knowledge required for the supervision system functions into operating system viewpoints, but also to provide a unique representation method for each viewpoint. The systemic approach SAGACE provides this formal representation framework and the methodology adopted to design and implement our industrial prototype relies on it. All these principles are illustrated by a description of an industrial application in the area of nuclear fuel reprocessing: the size and complexity of the facilities and their high degree of computerization make reprocessing particularly well suited for supervision applications.

Within the European ‘Vehicle Model Based Diagnosis’ (VMBD) project, demonstrator vehicles with built‐in faults provided a serious challenge to model‐based diagnosis techniques and a real‐life test‐bed for their evaluation. One of the guiding applications within VMBD was model‐based on‐board diagnosis of faults in a turbo diesel engine system with a focus on potential origins of increased carbon emissions. This paper focuses on the application aspects. We discuss the requirements imposed, the way they were addressed by the chosen solutions, and the results obtained by the on‐board diagnosis prototype running on the demonstrator vehicle. The most important challenges of the demonstrator were to apply model‐based diagnosis systems to dynamic systems with feedback, to handle systems without a rigorous mathematical model (such as a combustion engine), and to try to provide the response times required for real‐time applications.

Although the area of model‐based diagnosis has developed a number of prototypes with impressive features that promised economic impact and, hence, caught industrial interest, the number of actual industrial applications is still close to zero. One of the reasons is that the successful techniques have not yet been turned into tools that reflect and support the current diagnostic work processes and their existing tools. The INDIA project joined eight German partners (research groups, software suppliers, and end users) in an attempt to take a major step in the transfer of model‐based diagnosis techniques into industrial applications. This paper describes part of the work carried out in this project. Rather than presenting the theoretical foundations of the techniques in depth, we focus on the aspect of how model‐based diagnostic techniques can be related to established tools and systems in order to provide some leverage for today’s work processes and to change them gradually, as opposed to postulating a radical change in current practice and organizational structures. From this perspective, we discuss the utilization of model‐based techniques for the generation of fault trees for on‐line testing and diagnosis of fork lifters, generation of test plans for an intelligent authoring system for car diagnosis manuals, and the exploitation of existing state‐chart process descriptions for post‐mortem diagnosis of processes in a dyeing plant.

Modeling a system is the first step in reasoning about physical devices. By restricting our domain to linear circuits, we can find an efficient algorithm to solve that task.

The algorithm presented in this article is an efficient implementation of the star‐mesh reductions used in Electrical Engineering. By choosing the right representation and based on simple data structures, we can reduce considerably the process of modeling a circuit.

The algorithm has three main sources of efficiency gain: An efficient cluster representation reduces the complexity of the produced model; a simple data structure reduces the search for parallel regions; in the last step, we generate a circuit model where the principle of superposition does not need to be applied. Those three points reduce dramatically both the complexity of the modeling process and the size of the model. The reduction in the size of the model favorably impacts its use in any reasoning task to be performed. Finally, avoiding superposition will allow us to treat this class of circuits more efficiently.