Generation Expert Systems Research Articles

On the implementation issues of a parallel expert system

Recent progress in computing hardware technology has resulted in advances in the development of parallel computer systems. This new generation of computers offers advanced architectures and technologies for many AI applications. A new subfield of AI, called Distributed Artificial Intelligence, has emerged which is concerned with the co-operation solution of problems by a decentralized and loosely coupled collection of intelligent agents. In the paper, a parallel expert system called HOPES (Hierarchically Organized Parallel Expert System), is introduced. The system structure and multiblackboard architecture are presented and discussed. The focus of the paper will be on the implementation issues of the HOPES system. Although the so-called second generation expert systems technology has been around for some time, relatively little research effort has been put on implementing such systems. Thus, a major purpose of the paper is to provide general guidelines for implementations of parallel knowledge-based systems. The authors are concerned with two categories of hardware structures. First, multiprocessor system with common memory. Second, multiprocessor system without common memory. This paper will reveal some very important implementation problems and discuss key issues which are believed to be essential for implementing parallel/distributed knowledge-based systems.

A knowledge environment for an interactive control system design package

Abstract This paper describes a generic expert system development environment in a computer-aided control system package called SFPACK. The basic command language of the package follows the well-known MATLAB syntax. However, SFPACK has evolved from the MATLAB-derived control environment to a new generation of control system design environment because of the novel knowledge environment developed within the package. SFPACK now supports a richer set of data structures for representing design objects and knowledge in the control environment. In this paper, the features of the knowledge environment is presented. The issue on coupling of symbolic and numeric processing is also discussed. An expert system for control system design is developed in this integrated environment.

Intelligent Supervisory System for Microwave Telecommunication Networks

To effectively supervise a telecommunication network an intelligent supervisory system is proposed consisting of a traditional process monitoring, a fault diagnostic, a communication handler and a database manager subsystem. The whole system is based on a generic expert system shell designed to operate in industrial environment. The diagnostic subsystem contains a two-level inference engine to operate on structural information and traditional if-then rules. To ensure easy mapping of the supervisory system to any telecommunication network a configuration environment consisting of several compilers integrated into a multi-window editor was also developed. The system was implemented on two interconnected IBM PC-s running MS WINDOWS 3.0.

A second generation expert system for fault diagnosis

We propose a second generation expert system for fault diagnosis in a power system. This expert system combines shallow reasoning with deep reasoning that is based on a model reference approach. The model used for qualitative simulation of the power system in the deep reasoning part is based on a timed Petri net approach. The timed Petri net approach allows for accurate qualitative simulation of the protection system and power system components. The Petri net model permits one to model the function and behaviour of the power system components. It therefore allows for both static and dynamic simulation of the protection system and the associated power system. An application showing the modelling of part of substation configuration is given.

The Model-Based Construction of a Case-Oriented Expert System

Second generation expert systems should be based upon an expert's high level understanding of the application domain and upon specific real world experiences. By having an expert categorize different types of relevant experiences and their components, hierarchies of abstract problems and operator classes are determined on the basis of the expert's accumulated problem solving experiences. The expert's global understanding of the domain is integrated with the experiences by a model of expertise. This model postulates problem classes at different levels of abstractions and associated skeletal plans. During a consultation with the expert system previously unseen types of input may be used to delineate a new problem. The application of the expert system can thus be situated in changing environments and contexts. With increasing dissimilarity between the cases that were analyzed during knowledge acquisition and the specific problem that is processed at the time of the application of the system, its performance gracefully degrades by supplying a more and more abstract skeletal plan. More specifically, the search space which is represented by the skeletal plan increases until the competence of the system is exceeded. This paper describes how such a case-oriented expert system is developed for production planning in mechanical engineering.

Seamless design-to-manufacture (SDTM)

This article presents a seamless design-to-manufacture (SDTM) methodology that can dramatically shorten design and manufacturing cycle times within the paradigm of concurrent engineering (CE). This article also identifies the need for a software-development environment that enables the creation of a set of software building blocks and their interconnection to form concurrent SDTM systems. The expert system generator (ESG) and the server network generator (SNG) combine to form such an environment. Two existing SDTM systems are described that demonstrate concept-to-manufactured-part in 48 hours for complex designs. They illustrate rapid design and manufacturing cycles that are feasible through the SDTM methodology.

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.

Mesh generation expert system for engineering analyses with FEM

The finite element method is used extensively by engineers and modeling scientists to analyze deformations and stresses in physical structures. These structures are represented quantitatively as finite collections of elements. The displacement of each element is computed using algebraic equations. In order to design a numerical model of a physical structure it is necessary to decide the appropriate resolution for modeling each component part. Considerable expertise is required in choosing these resolution values. A too fine mesh leads to unnecessary computational overheads when executing the model. A too coarse mesh produces intolerable approximation errors. In this paper we demonstrate that rules for deciding on appropriate resolution values can be inductively constructed from expert provided examples. The Inductive Logic Programming algorithm Golem is employed for this purpose.

Predicting the survival of immigrant insect pests in new environments

Insect pests and diseases of crops will continue to be spread around the world, despite all efforts to stop them. Their movements will be paralleled by the spread of weeds. On arrival at their new destination these pest species encounter a new environment. In order to establish, they need a favourable climate, suitable host plants, absence of severe competition, and other physical needs such as shelter or specific soil types. The response of an invading species to the climate in a new environment can be estimated using the CLIMEX model. A full pest risk analysis needs to combine such information with that on non-climatic factors. PESKY is a generic expert system for pest risk analysis. It integrates the risks from a potential source location of infestation and a pathway of entry to the new destination with the suitability of the new environment, to estimate the overall risk posed by the pest at the destination. To support such information systems there is a need for comprehensive databases of pests, their distributions, hosts and economically important attributes.

From CLIMEX to PESKY, a generic expert system for pest risk assessment1

CLIMEX, a climate‐matching model, is described in relation to pest risk assessment. The rationale for the development Of CLIMEX is given and its various functions are illustrated with examples, using the Colorado beetle Leptinotarsa decemlineata. The beetle's climatic requirements are first inferred from its native distribution in North America. They are then used to project the relative favourableness of Europe and Asia for population growth and persistence. The role is illustrated Of CLIMEX, in association with PESKY, a prototype version of a generic expert system for pest risk assessment, in the assessment of the risk of the beetle being transferred from France to China. Finally, CLIMEX is used to search for places in North America with climates best matching that of Beijing (CN), in order to target collection of possible biocontrol agents for use against the beetle in that area, should it become established there.

A prototype expert system for automatic generation of image processing programs

A prototype expert system for generating image processing programs using the subroutine package SPIDER is described in this paper. Based on an interactive dialog, the system can generate a complete application program using SPIDER routines.

The Boston University Manufacturing Expert System (BUMES)

Manufacturing engineering involves the processes and equipment required to design, analyze, fabricate, and test products. The transition from an engineer's first sketch to a finished, machined part is traditionally a long, expensive, difficult path. This path includes a variety of separate software solutions to individual subproblems. Together, these separate solutions raise many confusing issues of capability, compatibility, and design integrity. This paper presents the design and implementation of a fully integrated solution to this problem. This system enables a designer to sketch a part graphically, and immediately have it machined. Because this rapid-prototyping system was designed as an integrated solution, it allows the process rules to be available up-front in the part design stage, so that unmanufacturable design features are identified immediately, as the designer attempts to add them to the part. This rapid-prototyping capability also addresses the need to modify the part by changing its dimensions or its features. Design changes and subsequent machining are rapid because of the use of relational geometry, which remembers the geometric relationships between the part's geometric elements, so that if some are changed, all can be adjusted automatically. The prototype Boston University Manufacturing Expert System (BUMES) was designed and implemented using the Expert System Generator (ESG) CASE tool, in APL2.

Generic Expert System for Identification of Industrial Processes

The final goal of our Expert System in Identification is to provide a helpful tool for the user of Identification techniques. Because the Identification theory is still growing, we have to implement our ES in such a way that it allows further developments of theoretical results or of the way to use them. Moreover, each user could wish to describe easily its own expertise. So we aim to design a generic knowledge base, for evolutivity and flexibility. for that purpose we use the Static, Semiotic and Dynamic Design method, as described in the first part of the paper. The most relevant aspect of this method is the structuration of the knowledge base into three components: the generic component, the static knowledge base and the dynamic knowledge base. The second part of our paper presents the description and the structuration of the objects involved in identification procedure and the links between them (the static description), the strategy of an user for solving a specific problem of identification (the dynamic knowledge base). The third and last part describes the generic component, which contains the control structure under the form of generic rules. In this paper we focus on a part of identification general problem, namely the choice of Model Structure, Parameter Estimation Method and Computation Algorithm.

Expert systems as a tool for statistics: A review

The current state of using expert system (ES) techniques for statistics is expounded through the difficulties of defining the notion of statistical expert system (SES) and through several well-known examples. The main characteristics of statistical expertise, and particularly the important concept of strategy are emphasized. The current state of SESs is seen in the context of the global trend of the evolution of ES, i.e. within the growth of second generation expert system (SGESs). The knowledge base of these ESs represent an attempt to represent not only experts' knowledge but also the way experts acquire their expertise, thanks to the existence of an underlying model. We show through many examples how far SESs are and/or will be able to be SGESs.

A Knowledge-Based Framework for the Diagnosis of Sludge Bulking in the Activated Sludge Process

Despite advances in the understanding of the activated sludge process for treating wastewater, the operation of an activated sludge process plant - in particular, the detection, diagnosis and remedy of operational problems - still involves a significant amount of qualitative knowledge derived from empirical observations. Expert systems can be of assistance to plant operators in problem diagnosis by automating the problem-solving behavior of human experts and retrieving the appropriate chunks of qualitative knowledge from a large collection of such knowledge as the context of the problem dictates. A new generation of expert systems shell delivers better performance by providing (i) an object-centred framework with interesting computational properties to organize the considerable amounts of information about the physical world, (ii) flexible, context-dependent and programmable inference strategies to better model the problem-solving behavior of human experts, (iii) a reliable means of integrating numeric and symbolic computation, and (iv) a means for the expert system application to couple its inference procedure with its ability to interact with events in the real world through sensors and actuators. A prototype expert system employing a new generation expert system shell has been developed for diagnosing the sludge bulking problem in the activated sludge process. The paper discusses the knowledge representation scheme employed in the prototype, which is general enough to be extended to cover other operational problems occurring in sewage treatment plants. A study was performed to validate the knowledge in the prototype by comparing the conclusions of a panel of human experts reported in the literature with those of the prototype in response to a wide range of operating conditions. The study shows close agreement between the two sets of conclusions.

Transformation reborn: A new generation expert system for planning HST operations

The Transformation expert system (TRANS) converts proposals for astronomical observations with the Hubble Space Telescope (HST) into detailed observing plans. It encodes expert knowledge to solve problems faced in planning and commanding HST observations to enable their processing by the Science Operations Ground System (SOGS). Among these problems are determining an acceptable order of executing observations, grouping of observations to enhance efficiency and schedulability, inserting extra observations when necessary, and providing parameters for commanding HST instruments. TRANS is currently an operational system and plays a critical role in the HST ground system. It was originally designed using forward-chaining provided by the OPS5 expert system language, but has been reimplemented using a procedural knowledge base. This reimplementation was forced by the explosion in the amount of OPS5 code required to specify the increasingly complicated situations requiring expert-level intervention by the TRANS knowledge base. This problem was compounded by the difficulty of avoiding unintended interaction between rules. To support the TRANS knowledge base, XCL, a small but powerful extension to Common Lisp was implemented. XCL allows a compact syntax for specifying assignments and references to object attributes. XCL also allows the capability to iterate over objects and perform keyed lookup. The reimplementation of TRANS has greatly diminished the effort needed to maintain and enhance it. As a result of this, its functions have been expanded to include warnings about observations that are difficult or impossible to schedule or command, providing data to aid SPIKE, an intelligent planning system used for HST long-term scheduling, and providing information to the Guide Star Selection System (GSSS) to aid in determination of the long-range availability of guide stars.

SODEXPERT: A control room expert system for PWR secondary system chemical contamination diagnosis

SODEXPERT is the first version of an expert system designed to aid pressurized water reactor (PWR) control room operation supervisors. The system enables them to make diagnoses in the event of abnormal excursions detected by the on-line monitors controlling the chemical composition of the secondary system water. In the event of a major chemical parameter excursion, i.e. an incident situation, using it must not add to the operator's workload. Conversely, a slight parameter excursion, but constituting a sign of the onset of contamination, must trigger it. For these reasons, we have preceded SODEXPERT by a program which automatically processes the sensor data relayed to the control room which invokes it in accordance with suitable criteria. This software carrying out processing controlling invocation is itself rule-based and constitutes an auxiliary expert system written in the same language as SODEXPERT. SODEXPERT presents the operator with a diagnosis for a minimal contribution by the latter simply consisting of replying to questions concerning the operating condition of the unit at the time of the consultation. The SODEXPERT diagnosis is generally straightforward. However, in the case of doubt, it proposes a number of possible causes of contamination. When the entry of raw water is detected, it gives an estimation of the flow rate; for other types of contamination—the entry of caustic soda, trisodium phosphate or acids—it normally locates the origin of the contamination. The diagnosis is, insofar as possible, supplemented by operating instructions. SODEXPERT, as described here, runs on a PC-compatible microcomputer and can be used for 900 MW PWR on river sites under normal operating conditions. It currently uses the ALOUETTE expert system generator, designed and implemented by the EDF Research and Development Division. The time required for manual consultation is of the order of 1 min. However, in its present form, SODEXPERT analyses a static situation— point in time and an earlier reference series—whereas a rise in the contamination is a dynamic phenomenon rather than continuous. Furthermore, control room operators generally wish to know, from the onset of contamination, what the likely development will be. It is this forecasting capability, not merely that of instantaneous diagnosis, which needs to be provided if the expectations of the operating staff are to be met. Thus SODEXPERT is, as it were, bracketed by two functional resources indispensable for its implementation in the control room: • —upstream: a program for invoking it and processing data provided by sensors • —downstream: allowance for the dynamic forecasting aspect of • contamination at different stages of its development.

A Knowledge-Based Approach to Process Planning: A Case Study

This case study examines the background, and problems encountered building the EXCAST expert process planning system for use in the casting room of a collaborating manufacturing company. Knowledge-based expert system prototyping and delivery using particular computer software (the Prolog programming language), and the facilities of both first and second generation expert system shells are discussed. The paramount importance of providing an effective interface between the computer and user for a system sited in the work environment is featured. This project was funded by Objex Ltd, UK and the SERC/ACME directorate.

The current status of expert system development and related technologies in Japan

General trends and statistics obtained from responses to a questionnaire are reviewed, showing that more than 400 expert systems are now under development or in practical use in Japan. Three expert systems selected as representative examples from among planning-and-design-system candidates are described: IBM Japan's Scheplan, Kayaba's OHCS, and Hitachi's automatic pipe-routing system for a power plant. Current Japanese research in related methodologies, including knowledge compilation, knowledge acquisition, generic methods for building expert systems, the automatic generation of task-oriented expert systems, and machine learning, is examined. It is noted that AI is spreading into almost every Japanese industry, and Japanese companies are seeking the most efficient way of introducing AI technologies, particularly expert systems. >

A C-oriented tool for building second generation expert systems

Considerable interests have been expressed in developing a tool for building Second Generation Expert Systems (ESs). The C-Oriented tool ZIP/E ++ is developed to verify the authors' ideas on what characteristics such tools should have. New mechanism such as four-level inference control scheme with news posting/delivering model and built-in explanation facility are introduced; rich semantical knowledge representation and multiple inference engines are integrated; a knowledge acquisition shell ZIFS is offered to help domain experts to acquire the domain knowledge; a truth maintenance system for ESs built by ZIP/E ++ and multiple knowledge bases organizing, scheduling and intercommunication are provided; a knowledge-based expert system building model CIA-model also is considered. The most important features of ZIP/E ++ are: (1) It is a general tool suitable for building various domain ESs; (2) It is a flexible tool for integrating different kinds of inference engines and knowledge representations into a single system; (3) It is a tool with deep-level reasoning model for coupling heuristic reasoning and domain reasoning [6]; (4) It is a very efficient tool for problem solving; (5) It is a tool with well-developed environment to support the ESs building.