Diagnostic Expert System Research Articles

A diagnostic expert system prototype for CIM

This paper describes the Diagnostic System Prototype (DSP), consisting of a deep knowledge base and several shallow knowledge bases, and shows how it works. Based on the theory of hierarchical systems and hybrid diagnostic reasoning, the DSP tries to find a cause(s) for an observed symptom in CIM (Computer-Integrated Manufacturing), using the concept of entropy. The entropy is calculated using test costs and degrees of belief for each rule in the shallow knowledge bases. The DSP is expected to be applicable to different domains other than just the manufacturing environment.

A neural network in an expert diagnostic system

A neural network was applied for fault diagnosis in a calorimeter, an energy and position detector of high energy particles, composed of tubes assembled in planes, each of which is fired when a particle crosses it. By analyzing the histogram of the number of hits per plane, the network was able to diagnose some types of malfunction of the read-out electronics for each plane and was able to inform the expert system, which in turn took suitable action. The network architecture, the learning algorithm used, the fault classification, and experimental results are given.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

MELDASH: a diagnostic application-specific expert systems shell for network fault diagnosis

Expert systems are receiving considerable interest, but there seems to be credibility gap between the state-of-the-art technology offered by artificial intelligence and expectations fuelled by potential applications. This is because the emerging technology provides us with very sophisticated programming tools and environments, but with almost no basis on which to program. The computing procedure to be implemented must be specified by the applications. This paper emphasizes the need for an algorithm generation tool and discusses the prospects for the task- and domain-specific expert systems shell MELDASH. A prototype expert system for network fault diagnosis is described to explain the methodology.

Probabilistic Models for Uncertainty Management in Expert Systems with Applications to Fossil Power Plants

Human experts and decision makers must typically plan and act without the benefit of complete knowledge and information. In the theory and application of expert systems, uncertainty in human knowledge and reasoning must be effectively managed in order for such systems to be successful in complex, real-world domains. In this paper two probabilistic models for uncertainty management in expert systems are considered: (a) Bayesian probability theory and (b) Dempster-Shafer belief theory. The conceptual foundations of each theory are first summarized and are then applied to a typical diagnostic inference problem drawn from the fossil plant performance area. The advantages and disadvantages associated with each approach are then discussed, concluding with some general guidelines for selecting an appropriate uncertainty paradigm for fossil power plant applications of diagnostic expert systems.

Information processing coupled with expert systems for water treatment plants

The water utility industry is facing increased pressures to produce higher quality finished water at lower costs. The traditional feedforward-feedback control systems used in water treatment plants can be enhanced by using predictive models (with feedforward control) and diagnostic expert systems (with feedback control). The use of predictive models is advantageous as it is virtually impossible to ascertain specific chemical dose requirements from measurements of raw water quality parameters. These predictive models may be either regression relationships or neural networks. When feedback data indicate that the finished water quality is not acceptable and usual corrective actions based on feedback from sensors fails to correct the situation, diagnostic expert systems can be effectively used to aid operator decision making.

Fuzzy Diagnostic Expert System for Ultrasonic Testing. Selection of Fuzzy Connectives and Modification of Membership Functions.

A diagnostic fuzzy expert system for ultrasonic testing (UT). which has been constructed by the authors indicates good agreement between the results obtained by the system and UT experts. The knowledge is represented in terms of fuzzy production rules, and the result is inferenced by the logical product as a fuzzy connective. This paper attempts to improve the jnference result by selecting fuzzy connectives, and membership functions based on the knowledge from several UT experts. When the fuzzy connective is selected as the algebraic product and the membership functions are modified, the results obtained are superior to previous results.

Expert systems for nondestructive inspection and technical diagnostics : Soviet Journal of Nondestructive Testing, Vol. 27, No. 4, pp. 267–277 (Dec. 1991)

Expert systems for nondestructive inspection and technical diagnostics : Soviet Journal of Nondestructive Testing, Vol. 27, No. 4, pp. 267–277 (Dec. 1991)

A diagnostic and observation expert system for large TCP/IP-based internetworks

With the expanding use of the computer networks based on LAN, there is a rapid increase of fault generation. A tremendous amount of time and effort is required in the dissolution of those faults since the origins of those faults are physically distant and the data are difficult to obtain due to the distributed management. In addition, the prevention, as well as the early detection and remedy of faults, are required since a failure may affect a broad range. The authors have constructed a diagnosis/observation expert system LODES for LAN, by constructing a knowledgebase of the experiences in dissolving the failure. LODES discovers the violation of protocol by referring to the TCP/IP-related protocols, automatic detection of fault (or its indication) by observing packets, and points out the abnormality/missetting of network-related information of the host. In the diagnosis, it has the feature that the distributed cooperative problem-solving is executed by communicating among LODES (large internetwork observation and diagnosis expert system) attached to each of the network segments. LODES is actually operated, and it is shown that the phenomena and failures which have not been detected can now be detected.

舶用主機関の過給系診断システム

In recent yesrs, diesel engine has become like a black box for the user due to introduction of mechatronic systems and/or automatic systems. Consequently, it will be difficult to correct troubles or to improve abnormal condition of diesel engine.To overcome the above-mentioned problem, diagnosic systems of diesel engine based on the computer have been developed. However, these system normally require special sensors for the purpose of diagnosis. Thus, initial cost, reliability and maintenance of the sensors are becoming problems.We have developed an expert diagnostic system of engine performance, without adding any special sensor. The system diagnoses diesel engine troubles by using both expert knowledge gathered from experienced and skilled engineers, and information obtained from ordinary sensors such, as pressure gages and thermometers. This pater describes new system which diagnoses air and gas flow performance of diesel engine with a turbo charger.

Expert overseer for mass spectrometer system

An expert overseer for the operation and real-time management of a mass spectrometer and associated laboratory equipment. The overseer is a computer-based expert diagnostic system implemented on a computer separate from the dedicated computer used to control the mass spectrometer and produce the analysis results. An interface links the overseer to components of the mass spectrometer, components of the laboratory support system, and the dedicated control computer. Periodically, the overseer polls these devices and as well as itself. These data are fed into an expert portion of the system for real-time evaluation. A knowledge base used for the evaluation includes both heuristic rules and precise operation parameters. The overseer also compares current readings to a long-term database to detect any developing trends using a combination of statistical and heuristic rules to evaluate the results. The overseer has the capability to alert lab personnel whenever questionable readings or trends are observed and provide a background review of the problem and suggest root causes and potential solutions, or appropriate additional tests that could be performed. The overseer can change the sequence or frequency of the polling to respond to an observation in the current data.

LIED - Liver: Information, Education and Diagnosis

In this paper we describe LIED (Liver: Information, Education and Diagnosis), a diagnostic expert system devoted to medical education in the field of hepatology. LIED combines the facilities of a traditional expert system for clinical diagnosis with several modules designed for educational purposes. The goal of such modules is to train physicians, in particular as regards the improvement of their problem solving ability (at the end of their curricula students in medicine usually lack experience in organizing an efficient and accurate diagnostic process). Training on real and simulated cases has been widely recognized as a proper approach to clinical experience. The architecture of the diagnostic system is introduced, which forms the core of LIED, and which has been derived from our previous experience, and the various educational functions are discussed that were developed on top of the basic architecture.

Soybean Oil Extraction Diagnostic Expert System

A soybean oil extraction expert system (ES), SOYEX, was developed to assist plant supervisors, technicians, and engineers in finding remedies for excessively high residual flake fats during soybean oil extraction processes (Deer, 1988). This diagnostic, knowledge-based system was developed using the expertise of a recently retired technical director of a major soybean oil extraction company. The knowledge engineer was a graduate research assistant in the Department of Agricultural Engineering, Purdue University. SOYEX has been used in one commercial extraction plant since the summer of 1988, and is now operational in four extraction plants. This ES is extensively used to train new managers and to recommend solutions to problems that occur especially during evenings and weekends when human expertise is not readily available.

D-IAL: A diagnostic expert shell with plausible reasoning mechanism

This paper describes an expert shell, Diagnosis Intelligent Automation Language (D-IAL), which was developed for manufacturing facility maintenance expert systems. There is increasing necessity for expert systems which assist machine operators in maintaining and repairing manufacturing facilities at automated factories. D-IAL adopts the best first-search algorithm and has a plausible reasoning mechanism based on the Dempster-Shafer probability [4]. This plausible reasoning mechanism propagates defect probability in a search tree to maintain the consistency of the defect probabilities and requires less computational complexity than Shenoy's method [5] by introducing the restriction of dichotomy. D-IAL is written in the C language so that diagnostic expert systems developed on the expert shell can operate on workstations and/or personal computers. The maintenance tasks at a manufacturing line are explained and the search algorithm and the propagation of defect probabilities which D-IAL employs are described.

Fuzzy Diagnostic Expert System for Ultrasonic Testing. Application of Fuzzy Production Rule.

This paper attempts to build up a diagnostic expert system for ultrasonic testing (UT) which inspects the weld defect encountered in welding structures. The proposed system can handle knowledge and data which include various kinds of uncertaities and ambiguities by use of fuzzy set theory. The knowledge is represented in terms of fuzzy production rules involving certainty factors, which help accommodate reasoning with uncertainty. The ambiguities of knowledge and data are represented by linear membership functions. The computational time required for inference is reduced because the proposed system uses the blackboard model dividing the knowledge source and working memory. In addition, the system has the ability to explain the reasoning that led to solutions of diagnosis. The results obtained by the proposed expert system are compared with those by UT experts, and the usefulness of the system is denoted by high consistency in the comparison.

Supervising the heuristic learning in a diagnostic expert system

The paper presents the organization of a diagnostic expert system with some capability of heuristic learning. In particular the role of the supervisor is analyzed and a set of strategies is defined which allows the system to implement different policies (conservative and non-conservative approaches). The organization of the system has been strongly influenced by the results obtained so far in investigating the properties of neural nets and of human learning. The learning system is able to revise the knowledge bases used by the consultation system by taking into account the experience gained in solving cases as well as the confirmation (disconformation) of diagnoses provided by the external world. In particular the learning system revises the membership functions between findings and diagnostic hypotheses and the membership relations defined among diagnostic hypotheses. The paper focusse its attention on the description of the behavior of the learning system in the conservative approach and discusses some alternative solutions for memory organization.

MANY-VALUED LOGIC SYSTEM FOR LINGUISTIC MODEL PROCESSING

This paper describes mathematical tools and strategy which are incorporated in software system LMPSTM *LMPSTM is trade mark of Janes and Datacoop, 1990. (Druckmüller, Rychly 1989). This system was developed especially for control systems construction and was successfully used in industry. The LMPS system was also used in medicine for diagnostic expert systems construction. Rules of type “if-then” with variable semantic interpretation are used in this system.

On the co-operation between abductive and temporal reasoning in medical diagnosis

On of the basic (and often implicit) assumptions of most first generation diagnostic expert systems is that they operate in a static environment. However, the static domain assumption is very limiting since it requires that all manifestations are observable (and observed) at a unique time point in order to perform diagnosis (and this is unrealistic in medical applications). The adoption of deep and causal models in second generation expert systems provided some insights into how to deal with time in the diagnostic process. There is, in fact, a strong relationship between the notion of causation and the notion of time. In the paper we present an architecture for diagnostic problem solving based on the use of a pathophysiological model in which both causal and temporal relations are explicitly represented. In particular, the architecture is an extension of the causal component of CHECK which has been used to model pathophysiology in the fields of cirrhosis and leprosis. We show that in such an extended framework diagnostic problems can be solved correctly only by means of a strict co-operation between abductive and temporal reasoning. The complexity of such forms of reasoning is analysed and some sources of complexity are singled out. Possible restrictions of the representation formalism are presented and forms of temporal reasoning providing approximate solutions are discussed.

Temporally distributed symptoms in a diagnostic application

For various reasons many diagnostic expert systems do not explicitly represent time. In a number of cases this is a useful abstraction. In other applications — such as MOLTKE's —, however, some symptoms evolve over time in a characteristic manner and require several measurements which have to be carried out in a particular temporal order. In this paper we show how such symptoms can be represented declaratively in the knowledge base and how a general temporal matching algorithm can be smoothly integrated with MOLTKE's otherwise static framework. We emphasize that the extension preserves the general philosophy of the system and does not lead to unacceptable overhead when time-independent symptoms are processed.

Automobile diagnostic expert system by noise and vibration.

A diagnostic prototype expert system was built using characteristics of measured sound and vibration for the automobile. For the utilities of this system, a 1/3 octave filter (bandpass filter) and an A/D converter were used for data acquisition and then the data were analyzed using signal processing technique, statistical analysis, and pattern recognition using a hamming network algorithm. In order to raise the reliability of the diagnostic results by considering many operating variables and the condition of the automobile to be diagnosed, a fuzzy inference technique was applied combining several kinds of information. The results were displayed as a graphical method to help the novice in the diagnostic field. The validation of this diagnostic system was checked through a simple rotating simulator and automobile and it showed acceptable performance for a diagnostic process. Also, for the case of wrong decision making, this system included a learning algorithm and its knowledge base can be extended through that procedure. Therefore, as a result, it is expected that this system can be developed by implementing more simplicity and flexibility to the external environment.

Computerized Biological Treatment Operational Process Control

An integrated data management and operational process control system has been developed to assist in operations of biological wastewater treatment plants. Bio-kinetic process control equations provide daily operational strategies based on moving analysis of plant performance. Statistical analysis of trends and simulation add additional analytical power to the package. Diagnostic and predictive Expert System capabilities are also being developed to assist in plant operations.