Testing of a software system is a resource-consuming activity that requires high-level expert knowledge. Methods based on knowledge representation and reasoning can alleviate this problem. This paper presents an approach to enhance the automation of the testing process using ontologies and inference rules. The approach takes software requirements specifications written in structured text documents as input and produces the output, i.e. test scripts. The approach makes use of ontologies to deal with the knowledge embodied in requirements specifications and to represent the desired structure of test cases, as well as makes use of a set of inference rules to represent strategies for deriving test cases. The implementation of the approach, in the context of an industrial case, proves the validity of the overall approach.

There exist various situations when it is essential for an operator to be able to trace the control flow step by step in a modern computerized manufacturing system, e.g. during the start-up phase of a new manufacturing cell, when a failure has occurred and the fault should be localized to a certain spot, when a new product is introduced into a running system or when a new machine is installed. An operator should be able to perform these types of manual interaction preferably using a single tool. The idea is that the tool is used by the operator from a centrally located computer, typically the cell computer in a manufacturing cell. The operator should be able to trace the flow of different operations in an operation list in a transparent manner, i.e. without having to consider the actual level of control that he is currently on or on which computer the program is running on.

This paper describes a method using production levels and gracefully degradable manufacturing systems. A production level is a way of grouping certain activities in a manufacturing system. The method, based on graceful degradation, makes use of the production levels to uphold the availability of the system during computer failures or data network failures. The method is demonstrated on a case study of a machining cell.

This paper presents part of the results from a research project aimed at increasing flexibility and reusability of cell-control software. First to be discussed are the concept of flexibility and the advantages and disadvantages of various types of modular controllers. Then guidelines, generic models that describe the behavior of manufacturing resources, and a message-passing structure are given. The guidelines and the models should be used as the basis to support system developers when implementing modular control software for machining cells. The two main case studies examined to achieve the models are also briefly described.

This paper describes a series of protocols that increases the ability to automatically cope with failures in a computerized manufacturing system. In order to achieve better automatic failure solutions the behavior of a subsystem must be more precisely defined regarding the possible failure modes. This is called to give the subsystem stronger failure semantics. Then, regarding the systems specific failure semantics, an algorithm to cope with the corresponding failures can be given.>

The authors deal with fault detection and analysis in a partitioned data network in a computer-integrated manufacturing (CIM) system. They use the following definitions concerning the data network and the material network: the data network is the logical communication of information between two production units. The material network is the logical communication of material between two production units. The use of the material network is proposed in a computerized manufacturing system to detect and analyze faulty production units when the system experiences data networking partitioning. This approach can help avoid a costly system reconfiguration. The detection and analysis is done by the other production units in the system, either by a single production unit or by production units working together in voting groups.>

Distributed systems that maintain data communication through a data network always run the risk of having network partitioning. One application of a distributed system is computer integrated manufacturing (CIM), which depends on two different types of network, namely the data network and the material network. A network partitioning in CIM may lead to a stop in the production. The negative effects of network partitioning in a CIM system can be mitigated through the introduction of data and material buffers. Nevertheless, when the partition of the data network remains for a longer period of time, the usefulness of the buffers will be reduced as they eventually will be emptied or invalidated. A way to maintain the distribution of information between different data network partitions is to use the material network as an extra medium for the data transportation, as well as for the material transportation. This approach increments fault-tolerance concerning data network partitioning in CIM systems.>