The relatively young functional modeling approach is becoming a promising and leading technique in various areas such as fault diagnosis, alarm analysis, and supervisory control, especially in the process plants and indus-tries involving complex physical facilities. The primary objective of functional modeling approaches is to pro-vide a concise description of a system in terms of the functional properties of its objects in addition to structural properties. Multilevel Flow Models (MFM), firstly proposed by Morten Lind, is a graphical functional modeling method based on the notion of flow of mass, energy and information. It aims at providing a systematic basis for using means-end and whole-part decompositions in the modeling of complex industrial plant.
The first step of a MFM modeling approach is to provide a concise description of a system according to the functional properties of its objects as well as its structural characters, and thus requires not only a profound un-derstanding of system purpose, function, behaviour and structure but also skills on MFM modeling and IT. Un-fortunately, the experts who are familiar with the target system usually lack in expertise on MFM modeling method and IT skills as they are engaging in modeling the target system for various purposes such as supervisory, diagnosis, and analysis.
This paper proposed an integrated graphical interface based system named Multilevel Flow Models Studio (MFMS), shown as Fig. 1, that provides assistant from cover to cover, namely, modeling system and generate final application for monitoring, diagnosis, and operational instruction. With a friendly graphical interface, MFMS mainly consists of three components, an Extensible Markup Language based file structure to represent the MFM model; an editor to intelligently assist user establish and maintain the MFM model; and an executor to generate the application for diagnosis, monitoring and operational instruction in terms of the established MFM model. The executor can implement different applications by loading different MFM models. The MFM model in this MFMS includes not only functional, structural and behavioural properties of target system, but also vari-ous information and mechanisms for specific application. The MFM model can be easily revised for improving the performance and usability of the application. In this way, user can concentrate on the fields that they are fa-miliar with.
In order to model the complex process system and implement the auto propagation search through the pas-sages of MFM model, several new features are introduced to the previous MFM. A compose relation represents that a function in a MFM structure can be divided into a group of functions that are called as its sub-functions and integrated to another MFM structure. A sub-MFM-model function represents another integrated MFM model that is described as a single symbol in the current MFM model. A connector relation represents that there is a consequence between two functions, and it signifies that these two functions depend on each other in a direct consequential way.
By using Visual C++, DirectX SDK, MSXML Parser SDK and Microsoft Agent SDK, a prototype MFMS system has been developed and applied to establish a demo operation support system for a co-generation system, Micro Gas Turbine System. The demo operation support system with function of monitoring, diagnosis and op-erational instruction has been tested by operation data from several actual fault cases in order to validate and im-prove the MFMS.
In addition, this MFMS is now being applied to establish a demonstration system for Nuclear Fuel Cycle in Japan. This demonstration system will be used for communication between public and atomic society aiming at improving public acceptance for nuclear energy in Japan. Furthermore, after being integrated with Geographical Information System, the MFMS is also being used to analyze the trend of the deregulated Japanese Natural Gas Market.