Automatic programming method of the hottest sheet

Abstract: with the development and application of artificial intelligence, many researchers use expert systems, neural networks and other methods to explore the research of generative CAPP. According to the sheet metal computer aided process planning system developed by the author, the principle and method of automatic process planning using expert system principle are introduced in detail, A network diagram is proposed to describe the part information and express the process decision-making route, and the hierarchical reasoning method and flexible knowledge expression are used to solve the complexity of CAPP system development, realizing the requirements of practicality, openness and intelligence of the system

key words: sheet metal parts; Process design; Expert system

Computer Aided Process Planning (CAPP) has developed so far, and many systems and design and development methods have emerged. However, due to the complexity of process design, it is difficult to meet the requirements of CAPP system only by relying on a certain technology. How to comprehensively apply various relatively mature technologies and relatively general methods to simplify the development of CAPP and make it meet the openness and practicality of CAPP system has always been one of the goals pursued by researchers. With the development and application of artificial intelligence, many researchers have used expert systems, neural networks and other methods to explore the research of generative CAPP, in order to replace the previous derivative CAPP, and achieved some theoretical and practical results []

the author believes that there are two main difficulties in developing CAPP:

1) how to completely, accurately and quickly describe and input thousands of different parts into the computer, that is, the problem of part input recognition. The parameters of parts include two categories: one is the overall characteristic parameters, such as material, heat treatment requirements, topological shape and size, etc.; the second is the parameters with low iron content, including the type, size, position relationship, accuracy, etc. of each constituent element. Different from the data required and output by CAD, what CAPP needs is not the specific coordinates of some points, lines and surfaces, but the so-called feature parameters and implicit process information of parts and their constituent elements, and it requires fast and convenient input

2) processing technology is the orderly arrangement of various processing methods. It is not difficult to obtain the processing method of each feature element of the part, but its processing sequence is relatively difficult to solve. Although there are some process compilation principles, such as "rough first and then fine", how to be specific in practical application is a problem

the CAPP system for sheet metal parts developed by the author uses the basic principle of group technology to classify parts, and uses the description method based on feature elements to describe the geometric features of parts. Users only need to select the corresponding items according to the graphic or text prompts and input the necessary parameters, as shown in figure 1. In this way, the parts are described intuitively, quickly and accurately, which is convenient for part identification in process planning

process preparation is divided into two parts: pretreatment and process generation. Preprocessing is to obtain the local process information of each element through calculation, analysis and reasoning according to the parameter information of feature elements, such as the positioning of each element, dimensional accuracy requirements, tooling conditions, etc., and then conduct comprehensive induction to obtain the overall information of parts

because when inputting part information, the feature parameters of the part are described respectively, and the processing elements, processing methods and steps affect each other, it is necessary to integrate all aspects of information to make a unified analysis and identification of the part. This part infers the process information from the geometric information of the part, which can be considered as the process identification of the part, and finally obtains the detailed information of the part from part to whole, which provides the basis for the decision-making of the process route and the selection of processing methods, and prepares the processing data

process generation is to generate specific processes, machining allowances and tooling equipment, process cards and sheet metal unfolding drawings according to the partial and overall information of parts obtained from the previous reasoning, combined with the basic principles of process preparation, such as process concentration, benchmark overlap, etc., as shown in Figure 2

Figure 2 process route preparation

1 working principle of reasoning system

this sheet metal CAPP system adopts the same reasoning machine for global reasoning and local reasoning. When reasoning, it adopts two-way hybrid reasoning, that is, forward reasoning combined with reverse verification. When reasoning, the reasoning machine establishes a local dynamic database for each feature element and a global dynamic database for the whole part information [3]. The global dynamic database stores the overall feature information of parts, which provides the basis for process route decision-making and processing method selection. The local dynamic database stores the information of each feature element, which is the main basis of global reasoning. When compiling the process, it can also deal with each feature element according to this information

in local reasoning, the inference engine infers according to the current feature elements and the knowledge in the local dynamic database and local knowledge base (common), and the global reasoning is summarized from the conclusions in each local dynamic database according to the knowledge in the global knowledge base. Its working principle is shown in Figure 3 and figure 4

Figure 3 local reasoning

Figure 4 global reasoning

this hierarchical processing method is more in line with people's process and habits of part recognition. The purpose and object of reasoning are clear, which reduces the complexity of knowledge expression and part recognition, and less data (such as dynamic database) is accessed each time. Finally, the part object with sufficient process information can be obtained

the reasoning system of local reasoning and global reasoning is compiled in C language [4], which adopts depth first search. The reasoning process is as follows:

read the conclusion of a certain knowledge (production rule), check whether it already exists. If it does not exist and the rule has not been processed, read its conditions and verify it. If there are conditions unknown, verify it. If all conditions are met, verify it, As shown in Figure 5:

Figure 5 reasoning process

2 knowledge expression and knowledge base

system knowledge base is divided into local knowledge base and global knowledge base. Knowledge is expressed by production rules, stored in user-readable form, and can be modified directly

the knowledge of the knowledge base includes two classes that need to be discussed: symbolic knowledge and process knowledge expressed by first-order predicates. Symbolic knowledge is omitted here. There are mainly three kinds of process knowledge []:

1) expressions: because there are many digital information of processing elements in this system that need to be converted and analyzed according to specific conditions, it requires a large number of random and simple mathematical operations, considering the expansibility requirements, They are compiled in the knowledge base in the form of text and executed by a general expression processing program. For example, a knowledge expressed in natural language:

if the distance from the round hole to the bending edge is less than twice the thickness of the plate, it must be processed after bending, After decomposition conditions: (name "round hole") and (L 33 process route generation

due to the complexity of process design, it is a difficult point for expert systems to take into account the optimization of processing processes in various situations. Usually, low-level repeated construction and homogeneous competition are avoided. The process knowledge provided by users is the process decision-making idea of technologists in considering the process of parts processing. It can be expressed by processing decision tree, as shown in Figure 6.

Figure 6 decision route map

you can see the tree The vertical direction is the part process route arrangement, while the horizontal direction is the processing scheme under different conditions considered in the corresponding processing stage. Among them, the basis of route branching is the result obtained from the previous reasoning

considering that the tree structure is open, a large number of branches will be generated for the comprehensive consideration of the complex part tree. In fact, each branch has the same or similar processing decisions and schemes at the corresponding stage, so the following methods are used to simplify

1) merge the routes with the same decision tree to form the data structure of "graph". The graph is a weighted directed graph, the vertex represents a state of processing action, including the current state, result and decision-making scheme of the next step route, and the edge represents the direction of transition from the current state; The weight represents the operation of state transition

2) some decision schemes are highly repeated in the processing tree, especially some decisions that have less relationship with the part structure and feature element processing method sets that are widely used but have few changes. If they are compiled in the process decision route, it will complicate the route, increase the number of vertices and join pointers (edges), and extract them as a decision subset, When used, it is called by the operation represented by the weight in the roadmap like a function. Therefore, the whole part processing decision diagram includes a series of main route diagrams (main process routes) and secondary route diagrams (secondary process routes), as shown in Figure 7

Figure 7 main route map and sub route map

the practice shows that using the data structure of the map to represent the process decision-making route, and searching and matching it to produce the specific part process, can better solve the problem of processing sequence arrangement, and has the ability to optimize the process; Reduce the storage of process decision-making route, improve the search efficiency, express flexibly and have better flexibility; The drawing and program are stored separately in external storage, and the format has certain rules, which is close to the thinking mode of the process personnel, clear and easy to understand, easy to maintain and expand; Using recursive programming for local programming can be extended to the whole road map, and the design is not too complex

based on the above principles and methods, the author developed a CAPP system for sheet metal parts. The system is mainly composed of part information description module, process compilation module, process module, process file output module and part information retrieval module, as shown in Figure 8. Each component module is relatively independent in software design, and is controlled by the general control module of the system to establish contact with the flow of data

figure 8 sheet metal CAPP this project takes the overall structure of the system as the situation of the presidential decree

4 conclusion

artificial intelligence technology is a very important development direction of computer technology, which is widely used in various fields, and CAPP has always been the difficulty of manufacturing automation. At present, it is still difficult to solve the preparation of processing process with pure expert system or other artificial intelligence methods. The principle of group technology similarity, the principle of artificial intelligence and other available technologies are comprehensively used to realize the compilation of technological process, so as to achieve the practicality, openness and intelligence of the system, which is also the purpose of all artificial intelligence systems. Generally speaking, this method and system have the following characteristics:

1) the system compilation idea conforms to the customary method of engineering and technical personnel's compilation process, so the establishment of its knowledge base is very convenient and intuitive, and the establishment of a perfect knowledge base is the most difficult for general expert systems

2) the system knowledge base and inference engine are independent of each other, which is easy to expand and modify. The function base includes all kinds of general methods, and the application library includes relatively general algorithms and applications. The system is a data-driven system, and the specific process documents and process diagrams can be stored and modified, which can be said to be an open system

3) using the similarity principle of group technology, various sheet metal parts in actual production are classified and summarized, which simplifies the complexity of system design and operation, and makes the system truly practical