Experts take you to answer the hydraulic system failure
Before beginning to collect expert knowledge, knowledge engineers should go deep into the field of expert work, and familiarize themselves with, understand and understand the nature, content, conditions, environment and expertise of expert work as much as possible.
At the same time, we should also introduce the basic knowledge of ES to the experts, especially pay attention to the form of knowledge, how to store the computer and solve the basic principle of the problem. The experts have a clear understanding of the structure of the ES, so that experts can provide specialized knowledge.
Classify the problem to be solved and establish the image of the fault tree
The expert domain knowledge understanding to a certain extent, object to collect expert knowledge more clearly, to conduct a comprehensive discussion and analysis to all objects, its content will be sorted out to classification, the research object (system) is divided into several sub classes (subsystem), subclass (sub system subdivisible). The principle of classification is conducive to the rapid and comprehensive collection of system knowledge, and is conducive to the application of expert system to knowledge.
A hydraulic equipment fault diagnosis expert system as an example, the equipment erection and dismantling operations are combined with electromechanical control use of electricity, gas and liquid phase, to achieve the hydraulic part of the work, electricity, gas and liquid components of many kinds (12 cylinder, 2 cylinder), and various electric systems hydraulic system and hydraulic system arrangement structure of electricity and gas, more complex. Through investigation, it is found that the hydraulic system of the equipment is the most fault, which often leads to the failure of the erection or erection of the bridge body, the failure of recovery or recovery of the bridge and the failure of the control.
According to the operating characteristics of the equipment, the hydraulic system according to the different function division, classification of each subsystem, such as the supply circuit part is to provide the hydraulic oil for each subsystem to work properly, it is assisted by double gear pump, relief valve, oil tank, oil filter, oil and other components. The main part of the push; pushing hydraulic cylinder, balance valve, reversing valve and other components....... In this way, the collected objects are well organized and well organized. It is convenient for expert knowledge to collect correctly, conveniently and comprehensively. And then go down until you reach the diagnostic requirements.
Choose one of the most representative objects, from the bottom class object in hand
Choose one of the most representative objects, and starting from the bottom class of this object, using a variety of methods to collect knowledge acquisition knowledge. First collect a sub class object, and then collect brother class object, all brother class object after collection, and then complete the collection of parent object knowledge.
To solve the most representative object, other objects have a reference and reference, in the collection of the main energy can be placed on the characteristics of these objects, you can save time.
Another advantage of doing this is to collect together the knowledge of relevant and strong, for knowledge organization, knowledge base, to solve the problems when the speed of the search progress.
(1) according to the system of elements of knowledge in the division of the hydraulic system of the equipment in accordance with the executive function, can start according to the fault characteristics of each part of knowledge collection and collation, such as the common faults of gear pump oil supply part and other parts of the analysis, to sort out the corresponding fault phenomenon at the bottom of the knowledge, including fault phenomenon, fault reason, fault countermeasures (solution), then in accordance with the components from the component to the system to sort out the order of all the knowledge to affect the normal operation of the system, we must pay attention to comprehensive knowledge, and have certain rules to express.
(2) according to the system fault phenomenon to classify the consolidation of knowledge we can first determine the fault tree, the equipment pushing part, lifting part and balance adjustment and other actions by the hydraulic motor, hydraulic cylinder to implement, the fault tree is established for each subsystem according to the functions of each subsystem and its common fault phenomenon and the whole system, the corresponding elements of knowledge extraction.
For the most representative objects, we interview with experts to collect knowledge, in order to verify the preliminary knowledge, a case analysis method is used. In the collection of all objects, more use of the literature to introduce the knowledge acquisition method, from the grasp of the field knowledge extraction.
Choosing an appropriate knowledge representation method -- building the image of the rule base
The knowledge that have a variety of methods, such as the state graph representation, logical representation, semantic network representation and frame representation etc.. Various methods of representation have their own characteristics, and they are described in detail in many literatures. According to the characteristics of hydraulic system fault diagnosis, the appropriate expression method is selected. Choosing (or designing) an appropriate language is the key to many problem solving. The selected knowledge representation language should be able to express the domain expert knowledge conveniently and comprehensively, and also consider how to construct the knowledge base with this kind of representation, whether to use the knowledge base to solve the problem is convenient, and whether the efficiency is high or not. We use the object-oriented knowledge representation method, the writing format of the fault rule library file is modeled according to the format of the *.ini file in the Windows system, and notice the condition or relationship between the antecedent parts of the rule.
Domain experts or authorized users can modify, delete, extend and manipulate the attributes, rules and methods of this object, and the knowledge can be stored in the corresponding database table. As mentioned above, the knowledge base of the expert system is to establish the whole system design of the "bottleneck", its core is the knowledge and rules is the embodiment of knowledge, have much knowledge in the knowledge base (i.e. the number of rules) quality and knowledge (the correct degree of rules) determines an expert system to solve the problem the ability of. Therefore, the construction of an expert system first is to obtain a large number of concepts, facts, relationships and methods in the field of expertise, including various heuristic knowledge of human experts dealing with practical problems, so as to construct a content rich and perfect knowledge base.
Building object oriented knowledge base
The traditional knowledge based expert system structure is shown in figure 1. In this system, the inference engine uses the data in the database (fact) and the knowledge in the knowledge base to reasoning, so as to achieve the solution of the problem. The object oriented knowledge base system scheme is shown in figure 2. In such a system, the inference engine no longer appears as a single module, but is defined as a method in the knowledge base. The solution of the problem is controlled by the controller according to the process knowledge defined in the knowledge base.
In the object-oriented knowledge base, the concepts and entities involved in the problem solving are taken as objects, and they are represented as a frame, that is, each frame is an object. Each object forms a hierarchical network in relation to the superclass, subclass, and instance between them (Fig. 2). All the attributes of the object, the method of manipulating the property of the object, and the rules used during the manipulation are encapsulated in the object framework.
The methods involved in the object can be rule reasoning, and can also be any other functions, such as access to the database or access to the artificial neural network, and the user defined solution process and so on. The whole solution process is the process of transferring messages between objects, that is, the only way to communicate with each other is message passing (process call). Due to the encapsulation of objects, various types of solving operations do not interfere with each other.
In the process of designing a hydraulic fault diagnosis expert system for a equipment, based on the above method, we fully communicate with the domain experts, and build a knowledge base composed of fault tree file and rule base file. To solve the blindness of manual knowledge acquisition, improve the efficiency of knowledge acquisition, we obtained a knowledge 200 equipment fault diagnosis of hydraulic system.
The actual operation test proves that the knowledge is comprehensive and effective, and can cover almost all the fault phenomena and Countermeasures of the hydraulic system of the equipment. Therefore, we believe that the acquisition of expert knowledge by these methods takes a short time, a comprehensive collection of knowledge, and easy to build knowledge base, governance and progressive reasoning efficiency.
Wonderful recommendation:
Work group together to create a strong Sandvik venture
Ouweimu successfully cast steel cable for your Guizhou Expressway
China heavy truck J7G gas truck entered South Xinjiang market for the first time
