Height vibration identification and cancellation control method and system for cutting torch of cutting machine tool

Abstract

The invention relates to the technical field of industrial thermal cutting equipment control, and particularly discloses a cutting machine cutting torch height vibration identification and cancellation control method and system. According to the method, the multi-modal vibration and false vibration characteristics of the heightening shaft in the operation process are extracted based on feature modeling and a self-adaptive parameter identification algorithm by utilizing the existing height sensor and servo encoder signals of a machine tool, a simplified general vibration model is established, a cancellation control strategy is designed according to the simplified general vibration model, and active suppression of the vibration of the heightening shaft is achieved. According to the system, modeling, parameter identification and controller adjustment are completed through an automatic process, manual intervention or additional hardware is not needed, the high-speed cutting efficiency can be kept, meanwhile, the height fluctuation of the cutting torch is remarkably reduced, the distance between the cutting torch and a workpiece is stabilized, and the flatness and consistency of a cutting section are improved. The method is suitable for performance optimization of new equipment and intelligent transformation of old machine tools, the machining quality and the production efficiency can be remarkably improved, and the service life of the equipment can be remarkably prolonged.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of industrial thermal cutting equipment control, and particularly relates to a cutting machine torch height vibration identification and cancellation control method and system. BACKGROUND

[0002] In laser cutting machines, plasma cutting machines and other industrial thermal cutting equipment, the torch height control system is the core link to ensure cutting quality and efficiency. The distance between the torch and the workpiece directly affects the energy coupling, airflow distribution and kerf shape during cutting. Once the height control is unstable, problems such as rough cutting surface, uneven kerf width or slag accumulation may occur. The influence of vibration on torch height control mainly comes from factors such as insufficient rigidity of the machine tool, large inertia of the moving parts and complex structural characteristics. In particular, in old equipment, lightweight models or high-speed operating conditions, the machine tool structure will produce significant elastic deformation and additional dynamic response under stress and motion. These vibration phenomena not only manifest as observable height fluctuations, but also may appear in the form of "false vibration", i.e. the relative motion of the machine tool body reference system causes the height feedback signal to produce undesirable periodic or random disturbances. False vibration directly interferes with the height closed-loop control, causing servo accuracy to decrease, kerf consistency and cutting surface finish to deteriorate significantly, and in severe cases, even causing processing failure.

[0003] In the prior art, the control of torch height vibration mainly relies on several means. Passive control usually reduces vibration by increasing damping, optimizing support or using vibration isolation devices, but its adaptability is poor and it is difficult to effectively respond to wideband vibration that occurs during high-speed operation. Semi-active and active control can achieve more efficient vibration suppression through feedback force, but often rely on additional sensors, actuators and special control hardware, resulting in high cost, complex integration and debugging. While software control schemes based on numerical control systems can avoid additional hardware investment, most of them rely on accurate dynamic modeling of the machine tool and its motion axes. Traditional elastic arm free bending models or second-order damped oscillation models have certain theoretical basis, but they are insufficient in the face of multi-modal superposition, time-varying and non-ideal structural characteristics, with high modeling complexity and poor robustness.

[0004] To avoid hardware dependence or efficiency loss, some torch height adjustment systems attempt to reduce the impact of vibration by reducing the closed-loop bandwidth or cutting speed. However, this method sacrifices production efficiency at the cost, which deviates from the core demand of modern efficient manufacturing. At the same time, most existing control strategies lack identification and compensation mechanisms for false vibration, making it difficult to balance control accuracy and production speed in dynamic environments. These problems result in the inability of torch height control to fully utilize the performance of the equipment under high dynamic and high precision processing conditions, restricting the overall processing quality and market competitiveness of thermal cutting machines. SUMMARY

[0005] To solve the above technical problems, the application provides a cutting machine torch height vibration identification and cancellation control method and system, which can effectively identify and inhibit the multi-modal and time-varying vibration characteristics in the torch height control without additional hardware investment, and takes into account the real-time performance and system robustness.

[0006] Specifically, the technical solutions provided by the application are as follows: A cutting machine torch height vibration identification and cancellation control method, comprising the following steps: S1, constructing a characteristic model of the cutting machine torch height vibration; S2, obtaining model parameters of the characteristic model by using a system identification method; S3, integrating the characteristic model and a servo drive system model to form a complete torch height state space description; S4, according to the torch height state space description, using a cancellation control strategy based on an LQR controller to actively inhibit the false vibration of the torch height in the machine tool follow-up cutting process.

[0007] Further, the transfer function of the characteristic model is: , The vector form of the transfer function is: , , , Wherein, represents the servo driver input in the Laplace space, represents the torch height vibration in the Laplace space, is a differential filtering time constant, , , and are model parameters to be identified.

[0008] Further, S2 comprises the following steps: S201, before starting identification, move the torch down to the machining position of the workpiece to be machined, and obtain the measurement value of the torch height at this time as a standard value; S202, input a set of excitation signals to the torch height control system as the torch height target trajectory, and obtain a plurality of sampling values of the torch height while the torch moves up and down along the target trajectory; S203, use the standard value as a reference value, the difference between the sampling value and the reference value is the vibration height, and use the LMS identification algorithm to obtain the model parameters of the characteristic model through iterative training: For , the parameter update is: Wherein, and Respectively n Round and n +1 round of iteration completed parameter vector, is called the step length, is the updated gradient.

[0009] Preferably, in S203, the fixed step size of the LMS identification algorithm is replaced with a variable step size: For the parameter vector For each parameter element in the LMS identification algorithm, a fixed step size is used. In order to meet the requirements of convergence speed and accuracy at the same time, the step size is set for each parameter element to form a step size vector ,and , ,

[0010] Among them, diag() is a function that generates a diagonal matrix. is the attenuation coefficient and , is the input vector Middle i elements, Indicates taking the maximum element value in the input vector. is the amplitude coefficient and , N is the number of elements in the parameter vector.

[0011] Preferably, the excitation signal in S202 is a linear amplitude modulated sinusoidal signal, and the sampling frequency of the cutting torch height is 1 kHz.

[0012] Furthermore, after the identification is completed and the model parameters of the characteristic model are obtained, the predicted output of the characteristic model is compared with the actual height sampling value, and the residual error is calculated. If the residual error is reduced to the set threshold compared with the initial error, the model identification is judged to be successful; otherwise, the system identification step is re-executed; if the system identification is re-executed the number of times reaching the set threshold and still cannot converge to the set standard, an alarm is triggered, prompting to check for mechanical abnormalities.

[0013] Furthermore, in S3, the transfer function of the characteristic model is first rewritten as:

[0014] in, , , , , Its observable standard form is:

[0015] wherein, , represents the cutting torch vibration height in time domain, derivative of time, represents the rate of change of vibration height; , represents the servo driver input in time domain; , , , ; Based on the observable standard form, the total controlled object including vibration characteristics and servo drive system is obtained, and a complete cutting torch height state space description is formed, and the state space equation is: , wherein, , , , , is the sampling period, , , , ; is the servo driver speed command input gain, is the reference servo driver acceleration time.

[0016] Further, in S4, the optimal control law of state feedback is obtained by using the LQR controller: solving Riccati equation , the solution of Riccati equation is obtained ; the optimal feedback gain matrix is calculated ; the optimal control law is calculated ; On the basis of full state feedback, the reference signal is introduced: , ; wherein, and are the selected parameter matrix, and and are symmetric positive definite matrix; r is the introduced reference signal, is the servo driver input at the first n time containing vibration compensation finally obtained.

[0017] ​Preferably, a dynamic re-identification process is also included: during normal cutting of the machine tool, the height feedback signal and control performance are continuously monitored, and when it is detected that the vibration modal characteristics or external disturbances exceed a preset threshold, a rapid re-identification process is automatically triggered to update the characteristic model parameters and controller settings in real time.

[0018] A torch height vibration identification and cancellation control system based on the above method, comprising a height sensor, a servo encoder, a numerical control controller and a drive module; the controller integrates a data acquisition unit, a characteristic modeling and identification module, a controller design module and an execution instruction module, the height sensor and the encoder real-time collect the torch position and dynamic response, the characteristic modeling and identification module processes data and generates a vibration characteristic model, the controller design module dynamically generates control parameters accordingly, and finally the drive module acts on the servo system to realize real-time cancellation control of the torch height vibration.

[0019] Compared with existing torch height vibration control technologies that rely on passive vibration isolation, additional hardware or complex dynamic modeling, the present application can achieve active control by relying only on the existing height sensor and servo encoder of the numerical control system without the need for additional sensors or actuators. The characteristic modeling method in the present application avoids the limitations of traditional elastic arm vibration models or second-order damping models in handling multi-modal superposition and time-varying characteristics, making the vibration identification more concise, more efficient and more adaptable. Combined with adaptive parameter identification algorithms and cancellation control strategies, the present application can effectively identify and compensate for multi-modal vibrations and false vibrations generated by the torch during dynamic operation, significantly reduce the vibration amplitude of the height axis, stabilize the distance between the torch and the workpiece, ensure the flatness and consistency of the cutting section, and thus improve the machining quality and the yield of finished products.

[0020] Through the automatic model identification and control process, the implementation of the present application can complete vibration characteristic modeling and controller adjustment without human intervention, simplifying the use and debugging process and reducing the dependence on the technical level of the operator. Unlike traditional methods that suppress vibration by reducing the closed-loop bandwidth or cutting speed, the present application achieves effective suppression of height vibration while maintaining high-speed cutting efficiency, balancing production efficiency and machining precision, and is suitable for performance improvement of new equipment and low-cost intelligent modification of old machines, thereby extending the service life of the machine tool while improving the competitiveness of the equipment, meeting the needs of modern manufacturing for efficient, stable and intelligent cutting processes. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate embodiments of the present application, and are used to explain the present application, and do not constitute a limitation of the present application.

[0022] Figure 1is a basic structure schematic diagram of a cutting torch numerical control height adjusting system provided by an embodiment of the present application; Figure 2 is a false vibration phenomenon schematic diagram provided by an embodiment of the present application; Figure 3 is an amplitude-frequency characteristic diagram of a linear amplitude modulation sinusoidal signal provided by an embodiment of the present application; Figure 4 is a height adjusting control schematic diagram with rigid compensation provided by an embodiment of the present application; Figure 5 is a flowchart of vibration identification and cancellation control provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0024] As shown in Figure 1 , a basic structure of a cutting torch numerical control height adjusting system for cutting processing includes a spindle box, a Z-axis guide rail, a nozzle (cutting torch), a grounding row and other components. In the figure, the Z-axis is the height axis of the numerical control machine tool, which is stationary relative to the Z-axis guide rail and takes the vertical downward direction as the positive direction. The spindle box is connected with a servo motor through a screw rod and other mechanisms and can slide up and down along the guide rail. The nozzle coordinate Z H is measured by an encoder, and the distance between the nozzle and the plate to be processed Z G is measured by the inherent height sensor of the thermal cutting system. According to the function, the position control of the nozzle by the height adjusting system can be divided into height control and following control. The height control is to take the nozzle coordinate Z H as the controlled quantity to perform closed-loop control, and the following control is to take the distance between the nozzle and the plate to be processed Z G as the controlled quantity to perform closed-loop control.

[0025] The core function of the height adjusting system is to keep the relative distance between the nozzle and the plate unchanged during the cutting processing, that is, to keep Z G constant. However, when the rigidity of the machine tool is insufficient, the false vibration phenomenon existing in the machine tool will interfere with the height adjusting control. Under the condition that the horizontal speed of the machine tool is zero and the plate to be processed is stationary relative to the ground reference system, theoretically Z BThe height position of the plate to be processed in the Z-axis coordinate system should be a constant. However, in fact, for a weak rigid machine tool, when the nozzle moves up and down, the machine tool reference system moves relative to the earth reference system, causing the coordinates of the plate to be processed to appear up and down fluctuations Z B , that is, Figure 1 The origin O of the height adjustment system should theoretically coincide with the origin O' of the earth reference system, but due to vibration, there is a height difference between O and O' Z V , called false vibration height. As shown in Figure 2 , it is the false vibration phenomenon in a certain control period. The weaker the rigidity of the machine tool, the greater the fluctuation Z V , the greater the fluctuation Z V will interfere with the nozzle height control and affect the cutting performance.

[0026] To reduce the interference of false vibration on height adjustment control, the embodiment provides a cutting torch height vibration identification and cancellation control method. The existing height sensor and servo encoder signals of the machine tool are used to identify the multi-modal vibration characteristics in the cutting torch height shaft operation without adding new hardware, and a cancellation control strategy is designed based on the model to realize real-time suppression of the cutting torch height vibration, so as to stabilize the distance between the nozzle and the workpiece and improve the cutting quality.

[0027] As shown in Figure 3 , in one specific embodiment, the method mainly includes the following parts: I. Constructing a feature model of a weak rigid machine tool To simplify the problem, it is assumed that the weak rigid machine tool is a linear constant system. The linear constant high-order object is as follows: (Equation 1) The partial fraction expansion of the above formula is obtained (Equation 2) If the coefficients of the discretized system are allowed to change over time, and the sampling period satisfies certain conditions (i.e. the discretized system maintains the controllability of the original system and maintains the control accuracy requirements), the controlled object (servo driver input) can be described by a second-order time-varying discrete system: (Equation 3) When the object is stable or contains an integral element, the following conclusions are obtained: ① The coefficients of each term in the above formula are slowly time-varying; ② In the dynamic process, the output of formula 3 is equivalent to that of formula 1; ③ The steady-state direct current gain is the same as that of formula 1, which is .

[0028] Observation Figure 2 When the servo drive input is zero, the spurious vibration amplitude tends to zero as time increases. This indicates that the spurious vibration is asymptotically stable, satisfying the required stability condition. Therefore, the coefficients in equation 1 are slowly time-varying, and the time-varying parameters in equation 3 can be approximated by their steady-state average values. With the servo drive input to the spurious vibration amplitude Z V The equivalent second-order transfer function is as follows: (Equation 4) where is the DC gain of the object.

[0029] In the derivation of equation 3, the differentiator used is a difference differentiator, which amplifies measurement noise and reduces the signal-to-noise ratio. Therefore, equation 4 is first described in the form of a continuous system, making it easier to discuss the discretization method of the system. Equation 4 is rewritten as: (Equation 5) It can be seen that equation 5 involves first-order and second-order differentials. Common differentiators include difference differentiators, difference differentiators with low-pass filters, and maximum speed differential trackers. In this embodiment, a difference differentiator with a low-pass filter is used, and equation 5 is written in the following practical form: (Equation 6) where is called the differential filter time constant.

[0030] The vector form of equation 6 is: (Equation 7) where , .

[0031] II. Obtain model parameters through system identification For a multiple-input single-output linear time-invariant system: (Equation 8) and are column vectors of N x 1, is the weight vector, is the input vector, is the output variable. Let the estimate of , then the estimated output is: (Formula 9) Output error: (Formula 10) The cost function of the LMS identification algorithm is the square of the output error: (Formula 11) The gradient of the cost function with respect to the estimated weight is: (Formula 12) According to the steepest descent method, the weight update is as follows: (Formula 13) where is called the learning rate, also known as the step size, which is the conventional LMS identification algorithm.

[0032] In Formula 13, the same step size is taken for each element of . In fact, the optimal value of each element can sometimes differ by a large order of magnitude, and it is difficult to balance the convergence performance of all elements with the same step size (which is one of the starting points of the NLMS algorithm). Therefore, consider a more general step size: (Formula 14) where diag() is a function that generates a diagonal matrix. At this time, the learning rate matrix is extended.

[0033] In the conventional LMS, the convergence speed and the steady-state misadjustment are contradictory, and a fixed step size is taken for the weight vector, which cannot meet the needs of convergence speed and accuracy at the same time. This embodiment derives the selection strategy of the time-varying step size from the perspective of the eigenvalues of the matrix.

[0034] Let the weight error be: (Formula 15) It can be derived that: (Formula 16) In the above formula, is called the weight error transfer matrix, and its eigenvalues are: (Formula 17) where is an element of the input vector.

[0035] It can be seen that the weight error transfer matrix has only one adjustable eigenvalue , and the remaining eigenvalues are all equal to 1, and the weight error There is no decay during the iteration in the direction of the corresponding eigenvector. This restriction imposes a requirement on the input vector, namely The system must be adequately motivated. In addition, to achieve ,exist On the basis of a full incentive system, it is also necessary to meet So, choose Each element is (Formula 18) Where, ; is the attenuation coefficient, and .So (Formula 19) when hour, Get the minimum value 0. At this point, the LMS step value problem is transformed into the attenuation coefficient The value of is, and its valid range is .when The closer it is to 0.5, tends to 0, in the weight error space The faster the dimension of the corresponding eigenvector decays, the faster the weight vector converges.

[0036] Observing formula 18, we find that if we want the algorithm’s iterative process to run normally, we need to ensure In practice, a more practical technique is (Formula 20) Where, is a positive real number, and its value can be determined based on the approximate range of each dimension input. An effective empirical formula is , (Formula 21) Where, is called the amplitude coefficient.

[0037] Formulas 18, 20, and 21 constitute the complete step size update strategy of this embodiment. There are two undetermined parameters in this update strategy, namely the attenuation coefficient and amplitude coefficient Compared with the fixed step size of conventional LMS, the two have a clear range of values, namely , Usually, when or When the value is increased, the convergence speed of the algorithm will increase, but the steady-state imbalance will also increase, and the two are in a contradictory relationship. However, since the parameters to be adjusted have a fixed reference value range, the process of adjusting the algorithm parameters in actual use is simple and fast, and has wide adaptability.

[0038] To solve the measurement noise interference problem faced by conventional LMS algorithms in system identification applications, this embodiment provides a method based on frequency domain analysis. Consider the frequency domain form of the system shown in Formula 8: (Formula 22) Construct new input and output (Formula 23) (Formula 24) Where, It is called a filter, for dimensional diagonal matrix, which is defined as follows (Formula 25) Easy to obtain (Formula 26) If we consider the input and output noise based on Formula 22 and , we can deduce accordingly (Formula 27) It can be seen that when for and Has a strong attenuation effect while retaining as much as possible and When the information is and The LMS algorithm can still converge to This suppresses the measurement noise interference to a certain extent. The design can be done in combination with the frequency distribution of the excitation signal. Considering that the actual measurement noise is mostly high-frequency noise, and in order to simplify the design, is a low-pass filter.

[0039] 3. Design of Cancellation Controller Based on Characteristic Model The Linear Quadratic Regulator (LQR) is an optimal control method based on a state-space model. It is widely used in modern control systems. Its core idea is to design an optimal control law by minimizing a quadratic performance indicator. The general steps of the cancellation control strategy based on the LQR controller include: First, the plant is modeled as a linear time-invariant system, usually in state-space form. Then, a quadratic cost function needs to be defined. Next, the optimal feedback gain matrix is obtained by solving the Riccati equation. Then, the optimal control law is calculated based on the optimal feedback gain matrix. Finally, the calculated optimal control law is applied to the actual system to form a closed-loop control system.

[0040] In this embodiment, the formula 6 is first written as an equivalent transfer function form as follows: (Formula 28) where, , , , .

[0041] Its observable canonical form implementation (Formula 29) where, , denotes the cutting torch vibration height in the time domain, is the derivative of time, indicating the rate of change of the vibration height; , denotes the servo driver input in the time domain; , , , The subscript "v" indicates that it is related to the false vibration.

[0042] Based on formula 29, the total plant containing the servo system and the characteristics of the weak rigid machine tool is obtained, and its discrete form of state space equation is shown in formula 30.

[0043] (Formula 30) where, , , , , is the system sampling period, i.e. the control period; , , , ; is the servo driver speed command input gain, is the reference servo driver acceleration time.

[0044] The optimal control law of state feedback is obtained by using the LQR controller, that is: The parameter matrix and , and is a symmetric positive definite matrix; solving Riccati equation to obtain the solution of Riccati equation ; calculating the optimal feedback gain matrix ; calculating the optimal control law (i.e. control input) .

[0045] For the online solving problem of Riccati equation in embedded devices such as DSP, a numerical solution is provided: ,

[0046] Since the height controller belongs to a tracker rather than a regulator, a reference signal needs to be introduced on the basis of full state feedback. The introduction method of the reference signal r is as follows: (Equation 31) (Equation 32) Thus, the height control system with rigid compensation can be obtained Figure 4 As shown in the figure, the Plant is the system shown in Equation 30.

[0047] Four, system identification and cancellation control process Before system identification, a series of test data of the target height control system needs to be obtained, so that the model parameters of the system can be determined through iterative training by using the above identification method. Specifically, as shown in Figure 5 the complete system identification and cancellation control process of the embodiment mainly includes the following steps: S1, cutting torch is positioned at the reference height collection Before starting identification, slowly move the cutting torch along the Z axis to a position about 3mm away from the surface of the workpiece. At this time, keep the motor stationary through the servo driver, and set a 1-second stationary delay time to ensure the stable relative position of the machine tool and the workpiece, and avoid the influence of residual inertia on measurement. Use the height sensor of the machine tool to collect the measurement value of the cutting torch height at this time as the standard value.

[0048] S2, excitation signal setting and cutting torch driving sampling To fully stimulate the dynamic characteristics of the machine tool height axis, a set of linear amplitude modulation sinusoidal signals is input to the height adjustment control system as the cutting torch height target trajectory. While the cutting torch moves up and down along the target trajectory, the actual cutting torch height response and driving state data are synchronously collected by the height sensor and the servo encoder. The sampling frequency is set to 1 kHz to ensure the capture of vibration details. As shown in FIG. 8, the linear amplitude modulation sinusoidal signal has rich frequency components. In addition to a larger frequency component at 9.8 Hz, the remaining frequency components are uniformly distributed in the 0.5-10 kHz frequency range, that is, the linear frequency modulation signal can stimulate the system in a wide frequency range. In addition, unlike signals such as step signals, linear amplitude modulation sinusoidal signals do not cause mechanical jerking, meeting the tolerance requirements of the machine tool. Figure 3

[0049] S3, feature model parameter identification The standard value obtained in S1 is taken as the reference value, and the measured value in the S2 excitation process is taken as the sampling value. The feature model (formula 6) is identified by the above identification method to obtain the corresponding model parameters (a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t, u, v, w, x, y, z). a 1, a 2, b 0, b 1).

[0050] To accelerate convergence and take into account accuracy, a variable step size update strategy based on matrix eigenvalue analysis is introduced. The step size matrix is dynamically adjusted according to the input signal energy and eigenvalue. To avoid the influence of high-frequency noise on identification, the algorithm combines a low-pass filter difference differentiator with a frequency domain weighting filter to suppress noise, ensuring the physical consistency and stability of the model parameters.

[0051] S4, residual error evaluation and model optimization After identification is completed, the modeling prediction output is compared with the actual height sampling value, and the residual error is calculated. If the residual error is reduced by more than 80% compared with the initial error, it is determined that the model identification is successful. If the threshold is not met, S2 and S3 are re-executed. At most, it is repeated three times. If the three identifications cannot converge to the set standard, the system alarm is triggered, prompting to check the mechanical abnormalities such as couplings and transmission parts.

[0052] S5, cancellation control design and height vibration suppression After the feature model identification is successful, the model is integrated with the servo drive system model to form a complete height axis state space description. The above cancellation control strategy based on the LQR controller is used to achieve active vibration control in the machine tool follow-up cutting process. To ensure follow-up performance, a reference signal compensation is introduced in the feedback control, so that the control system can maintain height following accuracy while suppressing vibration.

[0053] ​​In some embodiments, a dynamic re-identification process is also included, that is, during the normal cutting process of the machine tool, the height feedback signal and control performance are continuously monitored. When the vibration modal characteristics or external interference are detected exceeding the preset threshold, the rapid re-identification process is automatically triggered, and the model parameters and controller settings are updated in real time, so as to keep the cutting torch height control in the optimal state, ensure the smoothness and consistency of the cutting section, and maintain high-speed cutting efficiency.

[0054] Through the above implementation steps, the present invention can automatically identify and suppress multi-modal and false vibrations in cutting torch height control without hardware expansion, significantly reduce the height fluctuation amplitude, stabilize the distance between the nozzle and the workpiece, improve the cutting quality and equipment operation efficiency, and is suitable for low-cost intelligent transformation of old machine tools and performance optimization of new equipment.

[0055] Based on the aforementioned method for identifying and canceling the height vibration of a cutting machine torch, this embodiment also provides a control system for identifying and canceling the height vibration of a cutting machine torch. This system includes a height sensor, a servo encoder, a CNC controller, and a drive module. The controller integrates a data acquisition unit, a feature modeling and identification module, a controller design module, and an execution instruction module. The height sensor and encoder collect nozzle position and dynamic response in real time. The modeling and identification module processes this data and generates a vibration feature model. The controller design module dynamically generates control parameters based on this data. Ultimately, the drive module acts on the servo system to achieve real-time cancellation control of the cutting torch height vibration. The entire process can be completed with a single click through standard automated operations, without the need for human intervention.

[0056] The above-mentioned system can execute the vibration identification and cancellation control method described in Example 1, and has the corresponding functional modules and beneficial effects of the method. For technical details not described in detail in this embodiment, please refer to the vibration identification and cancellation control method provided in Example 1 of the present invention.

[0057] Through the description of the above embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a general hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the relevant technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0058] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, but not limited to them; under the idea of the present application, the technical features of the above examples or different examples can also be combined, the steps can be implemented in any order, and there are many other changes of different aspects of the present application as described above, which are not provided in details for simplicity; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying and canceling the height vibration of a cutting torch of a cutting machine tool, characterized in that: Including steps: S1. Construct a characteristic model of the vibration height of the cutting torch of a cutting machine tool; S2. Obtaining model parameters of the characteristic model using a system analysis method; S3, integrating the feature model with the servo drive system model to form a complete state space description of the cutting torch height; S4. According to the state space description of the cutting torch height, the cancellation control strategy based on the LQR controller is used to actively suppress the false vibration of the cutting torch height during the machine tool follow-up cutting process.

2. The cutting torch height vibration identification and cancellation control method according to claim 1, characterized in that: The transfer function of the characteristic model is: ， The vector form of the transfer function is: , , ， in, represents the servo drive input in Lagrangian space, represents the torch vibration height in Lagrangian space, is the differential filter time constant, 、 、 and are the model parameters to be identified.

3. The vibration identification and cancellation control method according to claim 2, wherein: S2 includes the steps: S201, before starting identification, move the cutting torch down to the processing position of the plate to be processed, and obtain the measured value of the cutting torch height at this time as a standard value; S202, inputting a set of excitation signals as a target trajectory of the cutting torch height into the cutting torch height adjustment control system, and acquiring multiple sample values ​​of the cutting torch height while the cutting torch moves up and down along the target trajectory; S203: Using the standard value as a reference value, the difference between the sample value and the reference value as the vibration height, and using the LMS identification algorithm to obtain the model parameters of the characteristic model through iterative training: for , its parameters are updated as follows: ,in, and Respectively n Round and n +1 round of iteration completed parameter vector, is called the step length, is the updated gradient.

4. The vibration identification and cancellation control method according to claim 3, wherein: In S203, the fixed step size of the LMS identification algorithm is replaced by a variable step size: For the parameter vector For each parameter element in the LMS identification algorithm, a fixed step size is used. In order to meet the requirements of convergence speed and accuracy at the same time, the step size is set for each parameter element to form a step size vector ,and ， ， Among them, diag() is a function that generates a diagonal matrix. is the attenuation coefficient and , is the input vector Middle i elements, Indicates taking the maximum element value in the input vector. is the amplitude coefficient and , N is the number of elements in the parameter vector.

5. The vibration identification and cancellation control method according to claim 3, wherein: The excitation signal in S202 is a linear amplitude modulated sinusoidal signal, and the sampling frequency of the cutting torch height is 1 kHz.

6. The vibration identification and cancellation control method according to claim 3, wherein: After the identification is completed and the model parameters of the characteristic model are obtained, the predicted output of the characteristic model is compared with the actual height sampling value, and the residual error is calculated. If the residual error is reduced to the set threshold compared with the initial error, the model identification is considered successful; otherwise, the system identification step is executed again; If the system identification re-execution times reach the set threshold and still cannot converge to the set standard, an alarm is triggered, prompting you to check for mechanical abnormalities.

7. The cutting torch height vibration identification and cancellation control method according to claim 2, characterized in that: In S3, the transfer function of the characteristic model is first rewritten as: in, , , ， ， Its observable standard form is: in, , represents the torch vibration height in the time domain, for The derivative with respect to time expresses the rate of change of vibration height; , represents the servo drive input in the time domain; , , , ; Based on the observable standard form, the total controlled object including the vibration characteristics and the servo drive system is obtained to form a complete state space description of the cutting torch height. The state space equation is: ， in, , , , , represents the identity matrix, is the sampling period; , , , ; Input gain for the servo drive speed command, It is the reference servo drive acceleration time.

8. The cutting torch height vibration identification and cancellation control method according to claim 7, characterized in that: In S4, the LQR controller is used to obtain the optimal control law of state feedback: Solving the Riccati equation , we get the solution of the Riccati equation ; Calculate the optimal feedback gain matrix ; Calculate the optimal control law ; Introducing reference signal based on full state feedback: ， ； in, and is the selected parameter matrix, and and is a symmetric positive definite matrix; r is the introduced reference signal, The final obtained n Servo drive input at that moment.

9. The cutting torch height vibration identification and cancellation control method according to claim 1, characterized in that: It also includes a dynamic re-identification process: during the normal cutting process of the machine tool, the height feedback signal and control performance are continuously monitored. When the vibration modal characteristics or external interference exceeding the preset threshold are detected, the rapid re-identification process is automatically triggered to update the feature model parameters and controller settings in real time.

10. A cutting torch height vibration identification and cancellation control system based on the method according to any one of claims 1 to 9, characterized in that: It includes a height sensor, a servo encoder, a CNC controller and a drive module; the controller integrates a data acquisition unit, a feature modeling and identification module, a controller design module and an execution instruction module. The height sensor and encoder collect the cutting torch position and dynamic response in real time. The feature modeling and identification module processes the data and generates a vibration feature model. The controller design module dynamically generates control parameters based on this. Finally, the drive module acts on the servo system to achieve real-time cancellation control of the cutting torch height vibration.

Images