Artificial intelligence technology has a unique effect when dealing with objects that are highly non-linear, time-varying and uncertain, which are not very effective in traditional methods based on precise mathematical models. Looking back at the development process of control theory, it can be seen that its development process reflects that human beings have entered the era of electrification from the era of mechanization, and are moving towards the era of automation, information, and intelligence. Since the 1960s, due to the development of space technology, computer technology and artificial intelligence technology, scholars in the control field have begun to apply artificial intelligence technology and methods to improve the self-learning ability of the system based on the study of self-organization and self-learning control. Actual control system: 1. Since computer numerical control technology has developed by leaps and bounds from hardware numerical control to software numerical control, it is still in a period of continuous improvement in the past two decades. Although there are some reports of novel numerical control technology, it has not broken the tradition. System framework.
Based on the above-mentioned situation, this paper proposes to apply artificial intelligence technology to computer numerical control systems, seeking intermediate intersections and new ways of integration. Because the cnc machine tool is a complex controlled object that combines mechanical, electrical, hydraulic, and gas phases, it is difficult to establish an accurate mathematical model. Although some control problems can be solved by using classic control theory, it is in the control of the machining process and fault diagnosis and maintenance. It seems a little difficult. To this end, this article aims at a certain functional module of the numerical control system, using artificial intelligence technology to achieve the purpose of replacing or improving system performance.
1 The application of fuzzy control in numerical control systems. Fuzzy mathematics (also called fuzzy set theory) was created by L.A.Zadeh in 1965. In the process of in-depth exploration and research on the relationship and contradiction between “large system”, “fuzziness”, “computer” and “human brain thinking”, he started from the separation of mathematics and human brain thinking, and found that Contor created The set theory is essentially a mathematical concept that removes ambiguity and abstracts it. It is to absoluteize the thinking process to achieve the goal of precision and strictness. For this reason, he unified fuzziness and mathematics, and instead of letting mathematics abandon its rigor to accommodate fuzziness, he let mathematics go back and absorb the advantages of the human brain in understanding and reasoning about fuzzy phenomena. So in Information and Control in 1965 )> published a seminal classic paper “FuzzySets at this time, you can use a two-dimensional fuzzy controller to achieve gain control. Its structure diagram is shown in the figure. The FPD fuzzy look-up table in the figure can use the membership function And rule reasoning is obtained, and simple and practical rule self-adjusting control strategy can also be adopted.
The principle of position loop gain fuzzy control. The application of artificial neural networks in numerical control systems. The research on artificial neural networks (ANN) has a long history, and there was a research boom in the 1960s. But after a ten-year trough, the revival began in the 1980s, which attracted great interest from researchers. In summary, ANN has the following salient features: distributed storage of information, even if a certain part of the network is damaged, the original information can be restored by relying on the associative memory function.
The information is processed in parallel, which greatly speeds up the operation.
Continuous learning, the method is simple.
A network is composed of many neurons, which can approximate any nonlinear system.
Therefore, the control systems designed based on ANN have good adaptability, intelligence and robustness, and can handle complex objects with high dimensionality, nonlinearity, strong interference, uncertainty, and difficult to model, which are specifically reflected in the following Several aspects: use adaptive neuron to realize the adjustment control of the position loop software gain of the numerical control system.
Using ANN to realize the fault diagnosis of CNC system.
Using ANN to realize the interpolation calculation of the numerical control system.
The interpolation calculation in the CNC system is one of the core modules. It is a process of inserting some intermediate points between the start point and the end point according to the line type, start point, end point, speed and other information of the contour of the processed part. It is also equivalent to ” Densification of data points’. The BP artificial neural network has a strong ability to approximate complex functions, and it has been proved that using a three-layer BP neural network, as long as the number of nodes in each layer is sufficient, it can theoretically approach any degree of complexity We can use this feature to construct a three-layer BP neural network to interpolate non-circular curve contours, as shown.
3 The application of the expert system in the numerical control system in the 1960s, and obtained extremely valuable applications. The so-called expert system is a computer program system that provides human expertise to solve important problems in the professional scope. It can solve the problem of knowledge reasoning in specialized fields where the structure is not clear or the algorithm is difficult to determine. Generally speaking, a well-functioning ES is composed of seven parts: knowledge base, reasoning mechanism, problem understanding, user interface, conclusion, learning mechanism, and knowledge acquisition, as shown in detail.
However, the general expert system is relatively large in scale, belonging to the category of static knowledge processing, and cannot realize online control in the face of actual control systems.
For this reason, Astrom introduced it into the real-time control field in 1984, and first proposed the name of Expert Control (ExpertControl), which has now developed into an important branch in the field of intelligent control.
Computer numerical control system (machine tool) is a fusion (Next page 40) 3 Conclusions The numerical control process has been studied in depth and detail, and the generalized process idea in numerical control machine tool processing and the optimization method of cutting consumption suitable for numerical control processing have been put forward. The optimized mathematical model of the main processing methods in Cnc Machining has been systematically established, and the corresponding application software has been developed. Through actual verification in the precision machining branch of the 457 Factory, the economic benefits are very significant.
Please keep the source and address of this article for reprinting：Research on Application of Artificial Intelligence Technology in Numerical Control System
Reprint Statement: If there are no special instructions, all articles on this site are original. Please indicate the source for reprinting.：Cnc Machine Wiki，Thanks