The main outcomes of the research activities of the EXTRAS project are the generic knowledge framework, and an extensive amount of the software produced which has allowed further steps in developing explanation systems. Explanation systems that are quite difficult, if not impossible, to obtain by other means. The project has had three major outcomes that can be used for explanation.
A methodology which will allow the use of intelligent simulation systems of on-line decision support and off-line training.
A set of generic software tools for interaction and explanation that can be combined with most existing simulation systems for developing future EXRAS systems;
The application of such toolkit to one important industrial process which is the power distribution substations.
Generally speaking, additional effort would be necessary in some methodological aspects, such as the multiple models switching strategy, as well as in the software toolkit to improve the robustness, reliability and integration of the whole system to allow a proper industrial exploitation. However, the system as it is, can be definitely profitable and the experience gained in the framework of the project will allow developing future simulator-based explanation systems in a cost-effective way.
The results of the applications are:
The Fossil Fuel Power Plant Application (FFPPA), an ISS for off-line training of power plants control room operators by means of real-time simulators. The training goal is the assessment and the improvement of the personnel capabilities related to the plant operation, mainly in usual and faulty. FFPA can constitute an effective support to the current training activity carried out by the Specialist Training Service of ENEL: it can be used both as an exercising tool for in-house training courses or as a self-training system to be used by the operators directly in the power plants.
The Paper Machine Application has been evaluated with the engineering and operating staff of the target process. The results have shown four tools to be interesting, but due to the simplistic and restricted scope have not been generally used. The operators are very familiar with the operation of this process and do not require development of the Extras tools and on-line operator decision support.
The Power Distribution application is going to have a key role for the exploitation of the project results and for the dissemination of the generic technology to other areas within the electrical sector or related to it. The validated final application will act as the main vehicle for demonstrating to potential customers the significance of the results of the project.
The Combined Cycle Power Plant application has been extremely useful in developing advanced training ideas and has received positive feedback from the operators. The validated application will provide an excellent vehicle for demonstrating for potential customers, and in fact has already lead to a possible contract with one customer.
To conclude the ideas exchanged between the partners and feedback from the each of the four applications has provided a good starting point. However the lack of an earlier input from the methodological side, as this task and the applications development tasks were running in parallel, prevented a stronger implementation of the methodology in the design of the demonstrators.
Finally, a broader exploitation of EXTRAS requires an improvement in the robustness and reliability of the software package as well as the development of some additional support tools.
BE95-1245 Explanation in on-line Help and Offline Training Based on Continuous Simulation
The ability to cope with rare, uncommon or abnormal situations is an increasingly essential requirement on the operation of complex industrial processes. The need for the continual reconfiguration of plant and the constant modification of operating conditions to meet production demands means that operating staff are faced with situations where they cannot react or decide on the basis of their experience alone. Further, the large number of possible situations makes it very difficult, if not impossible, to pre-enumerate the correct responses for use in a decision tree or expert system based approaches.
The currently preferred solution to this problem is to use simulation systems to train operators and to provide on-line decision support. However, these simulations can be difficult for operators with limited technical knowledge to use and interpret and can therefore, result in a loss of confidence in the system, thereby reducing its effectiveness. This requires that lucid explanation facilities are developed that can automatically interpret the responses of the simulation in a given situation and provide a causal explanation of how that behaviour comes about and why that behaviour results.
We propose to develop generic intelligent simulation systems that utilise Artificial Intelligence based techniques and advanced simulation techniques to provide perspicuous interaction and explanation systems that can be used as the basis for:
- Intelligent Simulation Systems for on-line decision support. - Intelligent Simulation Systems for off-line training. Intelligent Simulation Systems for combined training and on-line support
Based on pre-existing simulation systems, four application systems will be built as examples of the common generic technology. These represent important sectors of the continuous process industries, that now accounts for almost half of current European manufacturing capability. Specifically these applications are: fossil fuel power generation, paper and pulp production, gas turbines for combined power generation and control of regional energy distribution centres. The benefit derived from the usage of these systems will ensure increased efficiency of production, improved safety and reliability of operations and reduced environmental impact through tighter controls on product quality. The project is structured so that the results will be validated in four different European countries, so that the approach can be tested and readily adapted to different systems and various cultures within the process industries.
The Partnership structure therefore contains four companies that are very experienced in industrial simulation systems: Marconi Simulation and Training, CISE, ENEL and Tecnatom; three centres of excellence in developing and applying the background in software development and artificial intelligence techniques: CISE, KCL and Heriot Watt University; two major national suppliers and distributors of electrical energy, Iberdrola and ENEL. Finally, UPM provides an important application in the paper and pulp industry that tests the scope of the methods to be developed.
This powerful mix of expertise and geographical distribution ensures that generic technological solutions will be generated that are applicable to the diverse cultural backgrounds of operating personnel across Europe and beyond.
Funding SchemeCSC - Cost-sharing contracts
EH14 4AS Edinburgh