【文章內(nèi)容簡(jiǎn)介】 neural work was trained through NeuroShell (Samarasinghe, 2021。 Marques de Sa180。 et al., 2021) with data obtained from diagrams for determination of permitted exposure time on specific vibration level in dependence of frequency and acceleration. PROTECTOR contains the ‘‘standard’’ elements of an expert system: the knowledge base, an inference engine, the user interface and a working memory, but also a module for the interface with routines for relevant parameters determination, the Visual Basic routines themselves, and a database used by these routines (Fig. 3). Fig. 3. Architecture of PROTECTOR expert system ponent. The main purpose of the user interface is to provide means for a successful dialogue, . an exchange of information between the user and the system. It is the user interface that enables PROTECTOR to obtain all the necessary information from the user, on one hand, and that transforms system’s results and conclusions into information the user can understand, on the other. The PROTECTOR knowledge base is a formalization of the mine safety expert’s knowledge. Knowledge inexpert systems basically consists of facts and heuristics which can be represented by means of rules, frames, semantic works and other formalisms. Since knowledge is the key factor in problem solution and decision making, the quality and usability of an expert system is basically determined by the accuracy and pleteness of its knowledge base. The selection of the representation formalism is very important and plays a significant role in knowledgebased design. The problemsolving strategy is realized by the expert system’s inference engine. This reasoning mechanism infers conclusions based on knowledge from the knowledge base and the available information pertaining to the safety problem at hand. The inference engine stores intermediate results in the working memory. 4. Implementation issues The objectoriented approach in system structuring and modeling (Fowler, 2021。 Stevens, 2021) was used as the strategy for defining the model of processes and data in the development of PROTECTOR, and UML for the software development analysis phase. The task of visual modeling of the system is to define the objects and logic of the real system using the adopted graphic notation. Visual Studio was chosen as the programming environment for the development of PROTECTOR. UML has been used as the most appropriate notation, and the system’s architecture was conceived in the form of a threelevel class diagram. This architecture supports well the objectoriented approach in model development for plex applications. Its main characteristic is a separation of the domain model, which is represented by business services and data services, from user interface, represented by user services. Fig. 4 depicts the threetiered service model of PROTECTOR. Fig. 4. Threetiered service model of PROTECTOR. The nine classes identified within the user services of PROTECTOR represent its interface forms. They are used for data manipulation (entering, viewing and searching the data), textual and graphical presentation of results and munication with other modules in PROTECTOR (expert system and neural work). The classes related to user services municate with classes at the business service level by sending messages that initiate the execution of specific applications. Two interface forms belonging to user services are shown in Fig. 5. They enable text search and editing, munication with the database, creation of business diagrams, etc. Fig. 5. Interface forms of PROTECTOR. Three classes were identified within the business services, all of them are Visual Basic application modules and they are used for safety assessment. In Fig. 6 an activity diagram depicts the dynamic model of one of these classes the Neuro class. This class includes specific procedures based on the model of permitted time determination for specified level of vibration using a neural work and linear interpolation. Fig. 6. Activity diagram for dynamic model of Neuro class Data services provide data maintenance, data access and modification functions. In view of the plexity of the PROTECTOR system’s global model data structure, which had to model all relevant parameters of plex surface coal mine safety analysis, the design and realization of the database was executed in the MSAccess relational database management system. The system offers safe data archiving for plex data models as this one, as well as all procedures for data manipulation. The use of SQL as a standard query language for data manipulation secures the openness of the hybrid system PROTECTOR for a connection with different environments. Fig. 7 depicts the structure of the database relevant to PROTECTOR through the MSAccess Relationships panel. Fig. 7. Database structure of PROTECTOR. PROTECTOR was developed using an expert systems shell, the KAPPAPC applications development system. KAPPAPC is a MS Windows application which provides a wide range of tools for constructing and using applications by means of a highlevel graphical environment which generates standard C code. In the KAPPAPC system, the ponents of the domain are represented by objects that can be either classes or instances within classes. The relationships among the objects in a model can be represented by linking them together into a hierarchical structure. Thus the modified OOA model based on the strategy for evaluation of the general safety state of the surface coal mine could be easily mapped onto the appropriate elements of KAPPAPC. Objectoriented programming tools within KAPPAPC were used to endow PROTECTOR objects with methods that specify what objects can do. First the objects and methods for the knowledge base were constructed. Then mechanisms were build that specify