Techniques for machine condition monitoring and diagnostics are gaining acceptance in various industrial sectors. They have proved to be effective in predictive or proactive maintenance and quality control. Along with the fast development of computer and sensing technologies, sensors are being increasingly used to monitor machine status. In recent years, the fusion of multisensor data has been applied to diagnose machine faults. In this study, multisensors are used to collect signals of rotating imbalance vibration of a test rig. The characteristic features of each vibration signal are extracted with an auto-regressive (AR) model. Data fusion is then implemented with a Cascade-Correlation (CC) neural network. The results clearly show that multisensor data-fusion-based diagnostics outperforms the single sensor diagnostics with statistical significance.

The efficient use of critical machines or equipment in a manufacturing system requires reliable information about their current operating conditions. This information is often used as a basis for machine condition monitoring and fault diagnosis—which essentially is an endeavor of knowledge extraction. Rough set theory provides a novel way to knowledge acquisition, especially when dealing with vagueness and uncertainty. It focuses on the discovery of patterns in incomplete and/or inconsistent data. However, rough set theory requires the data analyzed to be in discrete manner. This paper proposes a novel approach to the treatment of continuous-valued attributes in multi-concept classification for mechanical diagnosis using rough set theory. Based on the proposed approach, a prototype system called RClass-Plus has been developed. RClass-Plus is validated using a case study on mechanical fault diagnosis. Details of the validation are described.

This paper presents a method for the automatic deduction of priority lists of input information as well as for the extraction of task relations from the available design knowledge. The method is based on multiple extensive searches of the design space and produces a final priority list for the instantiation of the primary design entities. If followed, this list ensures the generation of the most decisive design information at the very beginning of the design process. Additionally, the method can produce a priority list of design tasks, which represents the order of completion of these tasks. Finally, it offers a representation platform for tracking the evolution of the design process. First, a brief overview of the current literature is presented, after which the method is presented in detail. Entities, descriptors, and tasks are used for the representation of the design knowledge. They are linked in order to form multiple design relations, formally represented by digraphs. Simple set relations and graph theories are used as mathematical background to the method. Design experience is also taken into account through a weighting process of the primary design entities. Finally, an example of a belt conveyor design is presented, followed by a discussion of the results and some general conclusions. The method may be considered as a design-assisting tool that dynamically processes pieces of design knowledge and suggests corresponding design paths. Additionally, it relates the design tasks in a ordered form. Its extension—currently under elaboration—is expected to treat systematically the problem of identifying and handling the design knowledge inconsistencies.

A crucial early stage in the engineering design process is the conceptual design phase, during which an initial solution design is generated. The quality of this initial design has a great bearing on the quality and success of the produced artefact. Typically, the knowledge required to perform this task is only acquired through many years of experience, and so is often at a premium. This has led to a number of attempts to automate this phase using intelligent computer systems. However, the knowledge of how to generate designs has proved difficult to acquire directly from human experts, and as a result, is often unsatisfactory in these systems. The application of inductive machine learning techniques to the acquisition of this sort of knowledge has been advocated as one approach to overcoming the difficulties surrounding its capture. Rather than acquiring the knowledge from human experts, the knowledge would be inferred automatically from a set of examples of the design process. This paper describes the authors' investigations into the general viability of this approach in the context of one particular conceptual design task, that of the design of fluid power circuits. The analysis of a series of experiments highlights a number of issues that would seem to arise regardless of the working domain or particular machine learning algorithm used. These issues, presented and discussed here, cast serious doubts upon the practicality of such an approach to knowledge acquisition, given the current state of the art.

A common way of dynamically scheduling jobs in a flexible manufacturing system (FMS) is by means of dispatching rules. The problem of this method is that the performance of these rules depends on the state the system is in at each moment, and no single rule exists that is better than the rest in all the possible states that the system may be in. It would therefore be interesting to use the most appropriate dispatching rule at each moment. To achieve this goal, a scheduling approach which uses machine learning can be used. Analyzing the previous performance of the system (training examples) by means of this technique, knowledge is obtained that can be used to decide which is the most appropriate dispatching rule at each moment in time. In this paper, a review of the main machine learning-based scheduling approaches described in the literature is presented.