The problems and challenges faced in designing high-performance motion controllers in high-speed and high-precision machining are analyzed. The development status of motion controller design methods for high-performance numerical control systems is reviewed, and the main problems and possible future development directions are reviewed. A further discussion. 1 Research Background Hysteresis is an inherent characteristic of the industrial process and is considered to be the most difficult dynamic link that would otherwise exist in a physical system. A large time lag will result in larger overshoots and longer adjustment times, even The wrong control seriously affects the quality of control in the production process. The most popular method for the time delay system is the Smith predictor proposed by Smith (1957). However, this method is sensitive to the accuracy of the model. To eliminate this effect, Al2Majed proposed Based on the DOB-based Smith control scheme [2]. Since the 1970s, on the one hand, in order to improve the accuracy of the mathematical model and consider the influence of uncertain factors to strengthen the system identification! Industrial process modeling! Adaptive control! Robust Research on control and other aspects began to break through the constraints of traditional control ideas, trying to face the characteristics of the actual industrial process research and development of a variety of low requirements on the model! On-line calculation is simple and convenient! Real-time performance is good! New control algorithm with good control effect. On the other hand, the rapid development of computer technology also provides a good platform and foundation for new control strategies. Predictive control is a type of development in this type of situation. The new control algorithm [20]. The emergence of predictive control opens up a new way to solve the problem of large delay system control. In the past 20 years, the research and application of predictive control at home and abroad have become increasingly widespread. The research scope [21] has already involved predictions. Model Types! Types of Optimization Targets! Types of Constraints! Control Algorithms and Stability, Robustness, etc., Also Including Multivariable Systems! Nonlinear Systems. With Advances in Predictive Control Theory Research! More and more scholars began to try to combine other control theory with predictive control and formed many new predictive control methods [22,23], such as: neural network-based predictive control; fuzzy predictive control; grey predictive control; neural-based Network adaptive fuzzy predictive control etc.
There are many problems in the recovery of copper chips directly into the furnace, the recycling of copper brass chips reduces the production cost of foundry enterprises and greatly improves the market competitiveness of the enterprises. The copper scraps are small, loose, directly into the furnace will be seriously blocked, the material block clearance is the furnace gas rising channel, directly affect the smelting process, the loose steel scrap without pressing, extremely easy to float on the furnace lining, corrosion lining, and large loss of raw materials will cause low molten iron rate.
Copper Briquetter, Brass Briquette Machine, Copper Chips Briquette Machine, Copper Briquetting Press, Copper Briquetting Machine Jiangyin Metallurgy Hydraulic Machinery Factory , https://www.eco-balingmachine.com
Modern science and technology have brought profound changes to the machinery manufacturing industry. On the one hand, they have promoted the formation of a fiercely competitive market in the world, and have made users more and more strict with the product quality! The requirements for varieties and prices have appeared in the machinery manufacturing industry. High-precision, multi-variety, low-volume, low-cost and fast-cycle production requirements. On the other hand, their combination with mechanical manufacturing science and technology provides important system theory and technology for the mechanical manufacturing industry to adapt to this trend. Foundation [1].
CNC machine tools are developing to meet this demand. They have developed machining centers, flexible manufacturing units, etc. They are the products of electronic technology! The combination of information technology and machine tool technology. In recent decades, the development of foreign CNC machine tools has been very advanced. Fast. China's CNC machine tool technology development relative to industrialized countries, regardless of the output, output value and the number of possession of the NC rate, or in the variety! Performance and reliability, and other technical aspects, there is still a big gap. Therefore, according to The status quo of domestic and foreign machinery manufacturing automation development, starting from the reality of our country, in addition to the necessary follow-up research, focus on the necessary manpower and funds to study the key equipment and basic theory and technology of mechanical manufacturing, such as high-performance new type CNC system! High-precision servo control Technology and spindle drive technology, etc., have an important significance for the improvement of China's machinery manufacturing level, especially the level of basic technology, for the development of higher levels of integrated automation and for the development of the national economy. A new generation of numerical control devices to achieve high speed! High-precision! Processing with high efficiency and high reliability is a priority [1]. It should be noted that high speeds are high! Degrees! High performance and high reliability Four high-performance indicators are unified as a whole. To achieve high-performance control, high-performance CNC servo controller design is one of the basic and key technologies. This article is one of the important research aspects.)) The application of high-performance motion control in numerical control system is reviewed.
2 Research Status (Presentsituation)
2.1 Challenges High-speed, high-precision machining is currently one of the most important research areas. Its purpose is to improve machining productivity and improve machining quality. However, high-speed, high-precision machining faces many challenges before it is put into practical use. The main problem is that Perturbations exist! Nonlinearity! Design high-performance servo controllers with model and parameter uncertainties [2]. When using a servo controller with limited bandwidth, the servo delay becomes the main cause of position error, and will follow The high feed rate at high speed machining is even more serious. The modern processing system is supported by the servo system. The servo controller's performance and processing quality are closely related to the efficiency. From the perspective of the machine control system, the machine tool control is a dynamic system. The main reason for the uncertainties in the control system is due to [1]: 1) the input of the system contains random disturbances; 2) the measurement sensor of the system has measurement noise; 3) the parameters and even the structure of the system mathematical model are uncertain. The first and second types of uncertainties are called uncertain environmental factors, and the third type of uncertainties are called uncertain model factors. Traditional The control system controls the machine tools using the classical control theory method [3]. Most of them are PID control. The PID controller is often used in motion control because of its simple structure, convenient use and reliable operation. However, it is used to solve the system. When there are control problems with nonlinear factors that are not easily quantifiable, the PID controller is incapable of understanding. Because the structure of the PID controller itself and the design of the algorithm depend on the limitations of the object, the improvement of the precision leads to a decrease in the dynamic performance, while the dynamic performance Improvements also make actuators bulky and increase energy consumption. Furthermore, in the design of motion control systems with fast, high precision and robustness requirements, the same controller is not only used to improve the dynamic performance of input and output, but also to use To eliminate load disturbances, it is very difficult to obtain PID parameter tuning algorithms that make the system have satisfactory dynamic and static performance indicators. Since the classical control theory method completely relies on precise mathematical models and can only realize random control, such systems do not Determining the factors of the model is powerless. The treatment of uncertain environmental factors depends on the nonlinear energy of the control model. The control capability of uncertainties in such control systems is extremely limited. The role of adaptive control techniques in modern cybernetics is to track system parameters, environmental conditions and input signals, etc., and then by changing the parameters of the compensation components of the inner loop. Satisfactory performance. Due to the proprietary architecture of the traditional CNC system, the control strategy of the system is difficult to update. The implementation of adaptive control requires higher costs and costs, and the adaptive control does not improve the control performance enough. Control does not become the mainstream technology in machine tool control.
Another challenge is the realization of high-speed machining processes and the monitoring of CNC machining conditions [2]. The machining process and NC machining conditions mainly include the following factors [3]: 1) control of actuator state; 2) status of each movement axis; 3 ) Tool status; 4) Machine tool auxiliary function work status, etc. The status feedback signal mainly comes from various types of sensors. Due to the limitations of current sensor technology, it is still unrealistic to fully feed back the above machining status information. Especially for tool status and The measurement of workpiece status currently lacks effective means. In traditional machine tool control systems, position measurement and feedback are relatively mature technologies. The current hotspot technology of position measurement sensors is to realize digital feedback. It is used to monitor the force measurement of servo motors and other institutions. The speed sensor is also basically able to meet the requirements of the control system. The real-time measurement of the status of the tool and the workpiece is still mainly in the research stage. Although there are new methods, new technologies emerge, but overall, the realization of the technology is more complicated! It is more expensive and difficult to use for practical purposes. The lack of feedback capability has severely constrained the improvement of the level of intelligent control of machine tools. Step-by-step improvement of machine control performance bottlenecks. So new sensor technology for measuring tool states, workpiece status, machine characteristics, etc., and the use of current sensors for fusion estimation or soft measurement of these states is an important research direction for machine tool intelligent control. In the end, even with the new sensors, various servo control algorithms and process control strategies are perfect, and their application is also limited by the traditional closed CNC machine tools. Therefore, open-ended research has been done! The modular CNC system architecture has become another Important topic [4].
2.2 Research status of motion control The dimensional accuracy of the workpiece is determined by the contour accuracy. This fact has inspired many researchers to focus on improving the contour accuracy. The schemes for improving contour accuracy are mainly divided into two categories: 1) Multi-axis coordinated motion The control method [5,6]; the basic problem in the motion control system is to require multi-axis linkage to achieve specific performance indicators. Koren proposed a multi-axis coordinated motion control. Kulkarini studied in detail the multi-axis coordinated motion compensation control strategy and proposed The optimal solution. Tomizuka et al. added an adaptive feedforward strategy based on a multi-axis co-ordinated controller to improve its transient performance and suppress interference. However, the limitation of the above methods lies in its multi-axis coordination index. Is linear. Keron proposed a variable-gain multi-axis coordination controller, but the stability of the system has not yet been proved, the difficulty is that the time-varying characteristics of the geometric profile is difficult to analyze together with the transfer function. Chiu is considering nonlinearity. In the coordination index, a comprehensive control algorithm is proposed for a system with a relative order of 1. Kokotovic removes the relative order restriction by using the integral backstepping method. Benxian introduced intelligent control into coordinated control, and used adaptive fuzzy control to provide additional compensation to each linkage axis to improve system robustness. However, the above algorithm did not consider the influence of model uncertainty and external interference, which is also the current One direction of research.
.2) Based on the method of improving the accuracy of single-axis motion; most researchers focus on achieving small contour errors by improving the tracking accuracy of single axes. Among them, Lee's work is representative of such methods. They proposed Integrated feed-forward friction force compensation! Disturbance observer! The position feedback controller and feed-forward controller are integrated control structures, ie a high-performance servo system based on disturbance observer (DOB) [2]. Reference [4] The control strategy is improved to incorporate the workpiece processing status and statistical information into the design of the control system. Fig. 1 reflects this important and practical control strategy model. After Ohnishi proposed the disturbance observer, it was improved by Umeno. Its role is to compensate for disturbances. And the model uncertainty makes the system more robust to model uncertainties. The disturbance observer is not limited to continuous disturbances, and the bandwidth to suppress disturbances is also easy to adjust. However, because it is designed based on linear system theory, it cannot Effectively suppresses discontinuous disturbances. For example, friction can cause large positional errors. Therefore, a feedforward friction compensator is used to improve the robustness of the system. For nonlinear friction Compensation methods are commonly used [7,8]: online compensation method based on exponential nonlinear function, inverse controller compensation method based on neural network, etc. Feedforward controller can use optimal predictive control! Zero phase error tracking control Repeat control, etc. [9]. Position feedback control usually uses PID control. However, DOB is not very appropriate when the system parameter changes greatly or the motion trajectory has discontinuous acceleration, and there is no effect on the driving saturation in motion control. Consider [10,11]. For this reason, the researchers turned their ideas to adaptive control. However, because of the lack of robustness, traditional adaptive control systems have encountered great difficulties in practical applications, and redesigned with robustness! Robust adaptive control, which is characterized by robust optimization principles and intelligent ideas, has been welcomed and attracted the attention of theoretical researchers [12,13]. In order to solve the parameter uncertainty and nonlinear model uncertainty, Yao proposed The new motion control method, namely adaptive robust control [14], combines the adaptive control and deterministic robust control design methods, avoids weaknesses, retains the advantages of both and overcomes deterministic robust control. Not guaranteed Transient performance, poor robustness of adaptive control. Through proper controller configuration, robust control can ensure both good transient performance and good tracking accuracy; parameter learning can be used in adaptive control. Asymptotic tracking is achieved without the use of discontinuous control laws or high-gain feedback methods. Al2Majed proposed a high-performance servo system design method based on linear adaptive robust control (ARC) [2], through numerical control systems and high-speed large The application of capacity hard disk control system proves that ARC has better tracking performance than DOB [2,15,16]. In addition, supervised control based on DOB or ARC [17,18]! Multi-rate sampling control [19] in motion The control system has also been applied. 
2.3 The application of multi-sensor information fusion theory in processing Multi-sensor information fusion (MSIF) is an information processing process, it will come from different ways! Different time! Different space sensor information is coordinated into a unified feature expression to complete a certain A description of objects and environmental characteristics. Modern industrial production is integrated! Complex! Large-scale! Continuous is its characteristics, using a large variety of sensors to monitor and control the production process. In this multi-sensor system, the various sensors Provide space for information! Time! Different ways of expression, credibility! Different degrees of uncertainty, different focuses and uses, this puts new requirements on information processing and management. Practice has proved that a single sensor is difficult to correctly reflect Processing status, the development of multi-sensor information fusion is the inevitable way. Multi-sensors can provide information in various aspects of the processing process, and the synthesis and knowledge extraction of these information (ie, information fusion), and then the correct prediction and control of the processing process. The application of information fusion technology in machining is mainly based on tool status monitoring! Prediction of machining accuracy! Error compensation etc. Industrial control and monitoring binding .MSIF, will bring new mechanism of a conventional industrial monitoring and control, it is expected to form a new type of industrial monitoring and control system [24].
The commonly used methods to improve machining accuracy include: error avoidance based on improved machine tool accuracy and error compensation based on error elimination [24]. As the accuracy of the machine tool increases, the required cost will increase exponentially. Therefore, Under the condition of the existing equipment, the error compensation technology will be an effective method for the processing of general workpieces. The machining error modeling and forecasting technology is to improve the machining accuracy, reduce the machining error, and the key technology for error compensation. Many domestic and foreign A lot of researches on error compensation have been done in the literature, and several methods for establishing error compensation models have been proposed, such as triangular relation method, finite element method, homogeneous coordinate transformation method and neural network method, etc. The process state information includes cutting speed, depth of cut. Feedrate! Vibration! Cutting force! Spindle motor current! Feed motor current! Acoustic emission! Tool wear, etc. These are all related to machining accuracy [3,24]. How to get error from numerous information sources or The mapping relationship between the state compensation signal and the information source, wherein the artificial neural network method has a very strong learning ability and a nonlinear mapping capability, and has a comparison with other methods. Through appropriate training, the mapping from the error source to the positioning error can be accurately implemented, and the shortcomings of other methods such as heavy workload or insufficient boundary conditions are avoided. Therefore, the information fusion technology based on the neural network will be widely used in error compensation. Application. In the actual engineering system, input and output signals are susceptible to noise pollution. The emergence of stochastic fuzzy neural networks provides ideas for solving such problems [25,26]. Wavelet analysis theory is considered to be a breakthrough in Fourier analysis.[27] The wavelet transform performs multi-scale analysis of the signal by scaling and translation, and can effectively extract the local information of the signal. The wavelet neural network inherits the advantages of both, and adaptively adjusts the shape of the wavelet base to implement the wavelet transform, and has good Function approximation capabilities and pattern classification capabilities.
Some literature uses neural network [28] or wavelet neural network [27] for tool condition monitoring. However, their work is mostly directed to a certain workpiece or processing method, and the prediction models used are mostly based on a large amount of historical processing data. With the single forecasting method, when the cutting conditions change or the historical data is small, the forecasting accuracy will be reduced. Therefore, the online intelligent forecasting model with self-learning ability combined with the processing status information and historical processing data is a strong robustness and adaptability. There is an urgent need to solve the problem.
2.4 Introduction to the Open System Architecture of Numerical Control System Introduction The numerical control system is a special computer system. It is used in industrial field control and therefore has many differences from general-purpose computers. For a long time, the development of numerical control systems has become a self-contained system and established its own software. Hardware structure, implementation of technical confidentiality and technical blockade, making it difficult for machine tool manufacturers and end users to carry out secondary development, limiting the capabilities of machine tools and CNC systems. When CNC machine tools enter the distributed control and flexible manufacturing system environment, and even require After the common information systems such as CAD/CAPP/CAM were communicated, the original CNC devices with stand-alone services seemed to be insufficient. The new environment required that CNC devices be further transformed into open CNC systems.
The open architecture generally adopts a modular, hierarchical structure, and provides a unified application program interface through various forms. It has portability, extensibility, interoperability and scalability, etc. The internal openness of the composition and the openness between the components of the system are currently open. At present, the open architecture research focuses mainly on the function partitioning of the control module based on the PC; the hardware and software implementation of the control module; the division and formulation of the interface protocol; Research on the reference model of the architecture; planning of the machine tool control system oriented to open architecture! Design and implementation and intelligent numerical control architecture [29]. In fact, from the practical point of view, the open architecture research In the initial stage, there are still many problems that need to be solved, and concretely stated in the development outlook.
3 Development prospects
Although the research on the servo controller design method for high-performance numerical control systems is so extensive, there are still many problems to really achieve high performance! Intelligent, especially practical, it boils down to the following aspects, as a reference for future research work. .
(1) How to further improve the overall performance of the controller? Effectively integrate the current research results of modern control theory and intelligent control theory to improve the performance of the controller to achieve the purpose of avoiding errors, and apply processing based on multi-sensor information fusion and other theories. Error intelligent modeling and forecasting technology to achieve the purpose of error compensation, the combination of the two is an effective research idea.
(2) The above-mentioned related design methods still have their own shortcomings, so the design of high-performance controllers relies on the further improvement of these related technologies. For example, the stability and robustness analysis in predictive control theory are in urgent need of breakthroughs. For the characteristics of a large number of nonlinear and uncertain systems such as industrial processes, nonlinear predictive control and robust predictive control will become the focus of future research; how to use adaptive robust control for time-delay systems and high-order systems, etc. Therefore, further improvement and improvement of these methods is a direction for future development. These research directions include multi-axis nonlinear dynamic system identification and modeling; robust feedback control with respect to disturbances and parameter changes; robust against performance changes. Feedforward control; Predictive control over time delay; High-performance control system design based on disturbance observer; High-performance control system design based on nonlinear adaptive robust control; Multi-axis motion coordination control; Intelligent supervisory control; Trajectory planning Wait.
(3) Processing error based on multi-sensor information fusion theory, intelligent modeling and forecasting technology is system identification, pattern recognition, multi-sensor fusion technology, prediction theory, neural network, fuzzy system, wavelet transform, fractal theory and knowledge-based decision-making. With the comprehensive application of technologies such as control, it is vital to make full use of the advantages of the above technologies and continue to pay attention to cross-disciplinary research.
(4) The implementation of control strategy is actually a study of open architecture, which is also a key technology for realizing high-performance and intelligent numerical control. From the current research results, there is no uniform and clear concept for open architecture. Connotation! The system implementation technology is still in the era of a hundred schools of thought. There are still many problems to be further studied: solving the real-time problems of Windows and other operating systems; lack of compatibility between the systems of various systems; lack of real-time transmission control protocols; The new object-oriented data expression method must be compatible with open NC system; as a completely open CNC system, its security and reliability are greatly threatened, so the study of reliability and security is also necessary.
(5) CNC system architecture is not only open, but also networked! Software will be a trend and will be the focus of research. Software NC is a new concept of CNC system, and control decisions including servo control are completely based on The user's openness 0 of the PC's software implementation 0 and/or core control strategy is the two basic features of the software NC [29]. This open structure not only supports the customization of the motion control strategy, but is completely oriented towards the intelligent realization of the numerical control system, making the CNC The system has greater performance space and better system interactivity, and can fully draw on the latest results of related disciplines, thereby promoting the rapid growth of CNC technology. To achieve software NC, the system has higher CPU computing power Requirements. Higher-performance processors! More optimized real-time scheduling capabilities are the basis for software NC to achieve satisfactory control performance.