What is the specialty of "welding expert system" in material processing? Pray for the great gods.

The weldability of metal is the characteristic of whether the metal can adapt to the welding process and form a complete welded joint with certain performance. To evaluate the weldability, we should consider the properties and technological conditions of the metal itself. At present, the evaluation of metal weldability in China is still at the manual level, relying on personal experience. In order to improve efficiency and accuracy, it is the fundamental way to develop and popularize computer-aided metal weldability evaluation. However, in the process of metal weldability evaluation, there are many influencing factors, and the relationship between them is difficult to express clearly, and many problems are statistically poor. Therefore, it depends on experience and knowledge to solve such problems, and it is more appropriate to choose expert system [1]. Compared with manual work, expert system has distinct technical characteristics of expert intelligence, which can improve efficiency and accuracy. There is no such expert system in China at present. Based on the research on the weldability of metals, this paper puts forward an expert system for evaluating the weldability of metals. 1 expert system model 1. 1 overall structure model expert system usually consists of knowledge base and knowledge base management system. Knowledge base is a collection of collected human knowledge and experience, which is stored in computer files in the form of facts and rules and managed by knowledge base management system. Generalized knowledge base includes knowledge base, method base, model base and database. Knowledge base management system is computer software used to establish, maintain, manage and control knowledge base. 1.2 object-oriented model In the evaluation of solderability of metals, metals can be roughly divided into steel, copper and its alloys, aluminum and its alloys, titanium and its alloys, etc. They have their own characteristics, but also have certain * * * characteristics and object-oriented characteristics. Using object-oriented framework can solve the problems of multiple inheritance. The system adopts frame structure, with the same attributes defined in the upper frame and special attributes defined in the lower frame (see figure 1). A class is a collection of closely related objects with * * * attributes, and the class itself does not generate actual objects. Actual objects are generated by instantiation of subclasses, and subclasses generate objects. Private parts of an object, such as part name, material model and workpiece thickness, cannot be inherited. Common parts of objects, such as some welding process options, can be inherited by subclasses. 2 Reasoning Generally speaking, the weldability of metal is evaluated by its chemical composition, physical properties, chemical properties, welding process conditions, phase diagram characteristics, cct diagram, heat treatment and protection methods. This paper focuses on the evaluation of the first few aspects. The expert system has established case base, knowledge base and model base. The case base consists of several cases. The knowledge base is mainly composed of several rules. In this paper, the reasoning method based on cases and rules and the hybrid reasoning method based on cases and rules are given. 2. 1 Case-based reasoning Generally speaking, case-based reasoning includes case representation, organization and retrieval. The expression of the case is to describe the knowledge of metal weldability evaluation in as much detail as possible. The organization of cases is to sort and classify cases according to their characteristics and retrieval needs on the basis of case representation. Case retrieval is to find similar cases according to certain retrieval strategies. In case-based reasoning, it is very important to complete case retrieval quickly and effectively. The retrieval strategy of cases in this system adopts similarity. Representation of similarity: here, the distance between two fuzzy subsets is used to represent similarity. In actual processing, we set a certain threshold, which is greater than this threshold, so we can adopt this method. Is less than the threshold, the following rule-based reasoning can be used. 2.2 rule-based reasoning rules represent mathematical models: rulewhen if < condition1>; Then < conclusion1>; ; ... if < condition n> and then < conclusion n > (n> 1) End Rule < rule name > Every rule in the knowledge base is organized in the form of if ... and then. In the process of operation, interlocking, forming a complex reasoning network, in which knowledge can be transmitted and corresponding analysis and judgment can be made. Carbon equivalent is used to analyze the weldability of metals. Rule < carbon equivalent > when < low alloy steel > If (thickness < 25mm, heat input = 17kJ/cm, σ b = 700mpa, Ce = 0.52%) and then (preheating temperature 75℃) if (thickness < 25mm, heat input =1). Carbon equivalent > among them, it meets the requirements of low alloy steel, and its composition range is: C ≤ 0.2%, Si ≤ 0.55%, Mn ≤ 1.5%, Cu ≤ 0.5%, Ni ≤ 2.5%, Cr ≤ 1.25%, Mo ≤ 0. By inputting the brand of metal material, if it is steel, enter the steel number and call the chemical composition table in the database to get its chemical composition. Through the range of chemical composition, different carbon equivalent calculation formulas are called. According to the calculated values of process parameters such as carbon equivalent, the rules in knowledge base are called to evaluate the weldability of metals. 2.3 Mixed Reasoning Case-based reasoning can only search for existing cases, and it is easy to regard superficial similarity as essential similarity, leading to misdiagnosis. Rule-based reasoning, the system can only infer the rules expected and provided in advance, and the explanation and countermeasures of the conclusion are relatively dull, which can not flexibly adapt to the changes of the situation. Therefore, it is necessary to combine different reasoning methods to realize a hybrid reasoning strategy based on cases and rules. Firstly, search the case base and get the result based on case reasoning. If the result is not satisfied or there is no result, search the rule base and get the rule-based reasoning result. In this reasoning process, hierarchical model is used for step-by-step reasoning, and finally the results obtained by different reasoning methods are comprehensively analyzed. If different reasoning methods get similar results, this result is more credible. The advantage of hybrid reasoning is that it can give full play to the advantages of various reasoning methods, overcome their shortcomings and find the solution to the problem to the maximum extent. 3 system function (1) information consultation and inquiry. The expert system shows the expert experience and knowledge of metal weldability. (2) At the request of users, the weldability of metals is systematically evaluated. (3) Show the welding process under given conditions. (4) Maintenance and printing of the system. 4 system realization In the concrete realization, the inference engine is built by the high-level language visualc++ [4], and the database is built by visual foxpro, which accesses and processes the database through the odbc interface. Conclusion The expert system of metal weldability evaluation based on knowledge base involves a wide range of contents, and the knowledge in the knowledge base is not perfect in the case of short development time. With the development of welding technology and the enrichment of knowledge base, the expert system of metal weldability evaluation will play an increasingly important role in the welding field. I hope this article can play a role in the future work.