Integrated Computer-Aided Engineering

This paper describes the architecture of the Dynamic Scheduling Package (DSP), a configurable scheduling system currently in beta testing, performing schedule generation, schedule execution monitoring, schedule revision, and decision support for manufacturing and operation management. DSP makes extensive use of Object Design Methodology to define the structure, dynamics, and constraints of the manufacturing process to be scheduled. The scheduling engine solves the scheduling problem using an incremental, constrained-heuristic search process that can be guided by user-defined search control guidelines to generate and revise schedules within deadlines. DSP has been installed in several beta sites including discrete and batch manufacturing environments. The beta testing phase required several enhancements to the initial design of the system and has provided insights for further enhancements currently being designed. However, the architecture of the system proved to be very effective in addressing dynamic scheduling problems in a variety of manufacturing domains.

The discrete-event nature of on-line data from an automated manufacturing system offers many challenges for on-line fault detection. In this paper, the template monitoring method is developed as a simple but versatile method of using discrete event signals for fault monitoring of manufacturing systems. This monitoring method is easily distributable, suitable for monitoring highly concurrent systems, and has low computational requirements. This paper describes template monitoring using event observations, and demonstrates a method for automatically generating template monitoring models from timed automata specifications.

Error detection and recovery are essential aspects of the continuous and safe operation of automated manufacturing systems. In the past these were sufficiently simple to allow informal approaches to supervisory control and therefore to error recovery. The increasing level of automation, integration, and flexibility encountered in automated manufacturing systems renders formal approaches to the supervisory control system development a necessity. The controlled-automata based approach to supervisory control development considered in this work is one such approach, and provides a solution to the error recovery problem. It offers important advantages over other approaches. It guarantees that: (i) The resulting controlled behaviors do not contradict the behavioral specifications and are nonblocking (Ramadge and Wonham, 1989; Wonham and Ramadge, 1988), and (ii) the controlled behaviors are maximally permissive within the behavioral specifications (Ramadge and Wonham, 1989). Typically, these include safety and error recovery specifications, and production specifications such as routing and sequencing: In this paper, errors or their symptoms are interpreted as any event occurring in the plant behavior, and error recovery is approached as any specification on the behavior of the plant. This paper introduces supervisory-control-system development and illustrates the key concepts through a small manufacturing cell example. Subsequently, the development of an error-recovering supervisory control system for an automated assembly cell, in which the robots of the various assembly stations may break down and perform faulty assemblies, is described. “Off the shelf” industrial programmable controllers are used in the corresponding implementation.

A new bearing defect detection scheme is proposed using the data dependent systems (DDS) methodology. The detection scheme has two parts: monitoring and diagnosis. Monitoring indicates the stability of the defect and diagnosis identifies when it is significantly large and its location. The detection scheme uses two separate indices: band energy and detection vector. Two life-tests were used to judge the effectiveness of the scheme. In both tests, the scheme identified the defect (on the inner-race and the outer-race) several days before failure occurred. The detection scheme thus provides excellent results when predicting both types of defects.

Process modeling methodologies have become widely used in defining and analysis of manufacturing systems. Two specific types of models, functional analysis and process flow descriptions, are frequently discussed in the literature. However, most published work presents the basic concepts, industrial applications, and critiques of various methodologies. One drawback often cited is that, beyond preliminary ad hoc analysis, process models are seldom used. Furthermore, the lack of quantitative information in models prevents more detailed and interesting analysis. This paper adapts traditional system reliability methods to manufacturing and design process modeling. The logical relationships between activities in the process are used to quantify the overall reliability of the system and identify high-risk paths of decisions in the process. The concepts of path sets and cut sets are then used to identify critical activities and decision points. The analysis is illustrated with an industrial case study.

Making manufacturing systems more robust (i.e., able to carry out their function under the presence of faults) is an issue of paramount importance: It involves on-line and real-time detection of faults, diagnosis of faults, and recovery from the faults. In this paper, we present a system that is able to generate practical and efficient solutions for these problem. This approach we present is available as part of IPCS (Intelligent Process Control System), which is a model-based environment for generating monitoring, control, simulation, and diagnostics applications for large-scale, continuous process plants. IPCS has been used to generate practical real-time diagnostic and recovery applications in chemical and cogenerator plants.