Self-adaptive feeding speed control method and system of bus-type linear cutting numerical control system

Abstract

The invention provides a self-adaptive feeding speed control method and system for a bus-type linear cutting numerical control system, and the method comprises the steps: dividing the states of a machining process into a short circuit state, a normal discharge state and a no-load state through building a frequency reference benchmark in a no-load state and adopting a two-parameter fusion recognition method based on a current stability parameter and a frequency deviation parameter; based on state recognition, a staged adaptive strategy is adopted to calculate a speed adjustment coefficient: a preset parameter mode is adopted to perform adjustment and data collection in the initial stage, and a support vector machine algorithm is adopted to intelligently calculate an optimal feeding speed adjustment coefficient according to historical data in the later stage; through an improved time division method interpolation algorithm and a bus communication technology, dynamic adjustment of interpolation step length and real-time control instruction transmission are carried out; according to the invention, the machining precision, efficiency and stability are improved, and an intelligent control solution is provided for a bus type medium-speed wire cutting numerical control system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wire-cut electrical discharge machining control of numerical control machine tools, and particularly to an adaptive feed speed control method and system for a bus-type wire-cut numerical control system. BACKGROUND

[0002] With the development of manufacturing industry towards high precision, high efficiency and intelligentization, wire-cut electrical discharge machining technology, as an important means of precision machining, plays a key role in mold manufacturing and precision part machining. In particular, medium-speed wire-cut electrical discharge machining technology, with the advantages of balancing the high efficiency of fast wire-cut and the high precision of slow wire-cut, is widely used in precision manufacturing. At the same time, the application of industrial bus technologies such as EtherCAT and ProfiNet in numerical control systems is increasingly deepening, providing a technical basis for the further development of wire-cut numerical control systems.

[0003] At present, the feed speed control technology of wire-cut numerical control systems mainly has the following technical status:

[0004] Traditional pulse-type control technology mainly uses pulse frequency feedback to adjust the feed speed. This technology has the advantages of simple structure and low cost, but in actual application, the adjustment precision is restricted by the pulse resolution, and it is difficult to meet the requirements of high-precision machining. In addition, traditional pulse control mainly relies on preset parameters and experience adjustment, and lacks intelligent recognition ability for complex discharge states in the machining process.

[0005] Speed control technology based on a single parameter adjusts the feed speed by monitoring a specific process parameter. For example, some technical solutions use current feedback control to adjust the feed speed by monitoring the change of discharge current in real time, which can achieve good machining stability in high-current cutting. However, this type of technology mainly relies on a single current parameter for judgment, and has limited recognition ability for complex and variable discharge states in wire-cut machining, with relatively narrow application scope.

[0006] Another technical route is mainly to improve the overall control performance through mechanical optimization of the wire feeding system. Patent CN115528953A discloses a method for improving the stability of wire-cut wire feeding, which uses a stepper motor subdivision control system to effectively alleviate the low-frequency vibration problem of the stepper motor by smoothing the input pulse signal. This technology performs well in improving the running stability of the motor, but its technical route mainly focuses on mechanical control optimization, and has relatively insufficient recognition and intelligent response ability for discharge states in the machining process.

[0007] For special processing needs, some researches focus on developing control technology for special scenarios. Patent CN117161266A discloses a fine wire cutting wire system that can be externally attached to a unidirectional wire-cutting machine tool. It adopts a double-servo motor control scheme and uses an angular displacement sensor for tension feedback to achieve precise control of fine wire tension and wire speed. This technology is relatively advanced in fine wire tension control, but it is mainly designed for specific application scenarios of fine wire cutting and has obvious limitations in the wide applicability of conventional wire cutting processing.

[0008] In addition, bus-type control schemes have also begun to be applied in the field of wire cutting. Industrial bus technology has high-speed communication capabilities (EtherCAT communication cycle can reach below 1 ms, even achieving microsecond-level synchronization accuracy) and good device integration, providing a hardware foundation for high-performance control. However, existing bus-type control systems are mostly designed for general motion control, lacking specialized optimization algorithms for the unique discharge mechanism and discharge state recognition of electric spark wire cutting processing.

[0009] Based on the analysis of background technology, the existing medium-speed wire cutting technology still has the following shortcomings:

[0010] First, the accuracy and real-time performance of discharge state recognition are insufficient. During electric spark wire cutting processing, various discharge states such as short circuit, normal discharge, and no load may occur, and the transitions between states are frequent and complex. Existing control methods mostly rely on single parameters or operational experience for state judgment. They are easily disturbed by factors such as changes in processing environment and fluctuations in electrode wire quality, leading to high state misjudgment rates. This makes it difficult to achieve accurate and rapid recognition of complex discharge states, thereby affecting the accuracy of feed speed control and the stability of processing quality.

[0011] Second, there is a gap between the response speed of the control system and the real-time requirements of processing. The changes in discharge state during wire cutting processing usually occur within milliseconds, while traditional control systems rely on pre-set parameters and experience-based adjustments, leading to control lag and difficulty in responding to rapidly changing discharge states. Especially in high-speed cutting or complex contour processing, control lag may cause problems such as wire breakage and surface quality degradation, making it difficult to meet the strict real-time requirements of modern high-precision processing.

[0012] Third, the traditional wire cutting speed adjustment algorithm does not fully integrate with the modern bus architecture. Although bus technology has the advantage of high-speed communication, it is still a technical challenge to design an interpolation feed speed adjustment algorithm that is suitable for electric spark wire cutting characteristics and adapts to discharge states under a bus-type system architecture, achieving deep integration of real-time speed adjustment and bus communication. This prevents the full utilization of the parallel processing and real-time synchronization capabilities of the bus network.

[0013] Therefore, it is necessary to develop a self-adaptive feed speed control method which plays the advantages of bus technology and is optimized for the characteristics of wire electrical discharge machining, so as to ensure that it is in a stable discharge working state and meets the requirements of high-efficiency and high-precision machining. SUMMARY

[0014] The application provides a self-adaptive feed speed control method and system of a bus-type wire cutting numerical control system, and realizes intelligent control of a medium-speed wire electrical discharge wire (electrode wire) cutting process, so as to improve the precision, efficiency and stability of medium-speed wire cutting.

[0015] The application adopts the following technical scheme.

[0016] The self-adaptive feed speed control method of the bus-type wire cutting numerical control system comprises the following steps.

[0017] The self-adaptive feed speed control method is executed by a self-adaptive feed speed control system, and the method comprises the following steps.

[0018] Step S1: in the electrode wire idle running state, continuously collect the feedback frequency value to establish a frequency reference benchmark.

[0019] Step S2: in the machining process, the discharge current fluctuation and the feedback frequency change are synchronously monitored, a double-parameter fusion method based on the current stability parameter and the frequency deviation parameter is adopted to identify the machining state, and the machining state is classified into a short circuit state, a normal discharge state and an idle state.

[0020] Step S3: according to the machining state identification, a staged adaptive strategy is adopted to calculate a speed adjustment coefficient α, the value of α is less than 1 in the short circuit state to reduce the feed speed, the value of α is equal to 1 in the normal discharge state to maintain the feed speed, and the value of α is greater than 1 in the idle state to increase the feed speed; the feed speed value of the electrode wire in the next interpolation period is the product of the feed speed value of the electrode wire in the last interpolation period and the speed adjustment coefficient α, and a new feed speed value is calculated.

[0021] Step S4: multiply the adjusted feed speed value by the interpolation period to calculate the interpolation step; dynamically adjust the interpolation step; use the time division method interpolation algorithm to calculate the coordinate axis increments according to the adjusted interpolation step, and realize adaptive adjustment of the feed speed;

[0022] Step S5: store the coordinate axis increment data calculated by interpolation into a buffer area, and synchronously take out the corresponding data from the buffer area every interpolation period, and transmit it to the servo driver through the bus communication; the motor drives the feed transmission system to move, ensuring the stability of the discharge between the electrode wire and the workpiece.

[0023] Step S1 is the frequency reference base establishment process; specifically, before the wire cutting machine tool of the bus type wire cutting numerical control system enters the processing state, the electrode wire is started and maintained to run in the idle state. At this time, the control system collects the feedback frequency signal through the signal acquisition module and records the continuous data, with a sampling frequency of 20K; experiments show that in most systems, the frequency value usually maintains between 900Hz and 1200Hz, reflecting the typical state of high-speed movement of the electrode wire without participating in the discharge load;

[0024] The frequency interval of the frequency signal is used as the reference basis for subsequent processing state judgment, providing a stable reference basis for the adaptive feed speed control system, that is, by establishing a stable frequency reference in the idle state, the state change in the subsequent processing process is identified.

[0025] Step S2 is a double-parameter fusion discharge state judgment process, specifically: during the electrode wire discharge machining process, the system monitors the current feedback frequency value and current fluctuation through the signal acquisition module in real time; to accurately judge the trend of the discharge state change, the system dynamically compares the real-time collected frequency value with the reference frequency interval established in the idle state, and analyzes the current fluctuation characteristics;

[0026] The specific state judgment rule is:

[0027] When the feedback frequency significantly increases and the current fluctuation is severe, it is determined that the short circuit state exists, indicating that there are abnormal conditions such as short circuit, wire sticking or poor discharge in the processing area;

[0028] When the feedback frequency is stable in the 200Hz to 500Hz interval and the current fluctuation is stable, it is determined that the normal discharge state exists, that is, the discharge between the electrode wire and the workpiece is stable, the cutting load is moderate, and it is in an ideal processing state;

[0029] When the feedback frequency approaches the idle reference and the current approaches zero, it is determined that the idle state exists, indicating that the feed speed is too slow and the cutting efficiency is low;

[0030] Compared with the traditional single parameter judgment method, the double parameter fusion judgment method can more accurately identify complex discharge states and avoid misjudgment caused by single parameter fluctuations.

[0031] As shown in Figure 2 Step S2 adopts a double parameter fusion method implemented by the intelligent discharge state recognition system, in which the current stability parameter is a statistical characteristic value of current fluctuation, and the frequency deviation parameter is a relative deviation value of the current frequency and the reference frequency. The double parameter fusion judgment in this step is realized by a back propagation neural network. Specifically:

[0032] The intelligent discharge state recognition system performs 5-point sliding average filtering pretreatment on the feedback frequency signal and the discharge current signal, extracts the frequency deviation rate δ f and the current stability coefficient σ two characteristic parameters;

[0033] The frequency deviation rate δ f = (f current -f baseline ) / f baseline , wherein f current is the current frequency value, and f baseline is the no-load reference frequency; the current stability coefficient σ = ([(1 / n)Σ(I i -I mean ) 2 ]) 0.5 / I mean , wherein I i is the i-th current sampling value, I mean is the average current value, and n is 20;

[0034] The back propagation neural network adopts a 2-5-3 structure, the input layer receives two characteristic parameters, the hidden layer uses a ReLU activation function, and the output layer uses a Softmax activation function to output the probability values of short circuit, normal discharge, and no-load three states;

[0035] The system establishes state determination criteria according to the training data: the short circuit state characteristic is δ f >1.5 and σ>0.25, the normal discharge state characteristic is -0.3≤δ f ≤0.8 and 0.05≤σ≤0.20, and the no-load state characteristic is δ f close to 0 and σ<0.05; when recognizing the state, the output with the maximum probability is selected as the determination result, and when the maximum probability value is lower than 0.8, the observation time is prolonged for re-determination;

[0036] Compared with the traditional single parameter judgment, the recognition accuracy will be further improved;

[0037] The step S2 adopts a neural network algorithm to learn the mapping relationship between the double parameters and the processing state through network training;

[0038] The frequency reference is determined by multiple sampling and averaging, the processing state recognition is updated in real time, the adjustment range of the speed adjustment coefficient is set with upper and lower limit values to prevent abnormal adjustment, and a safety protection mechanism is adopted when continuous abnormal state is detected.

[0039] Step S3 includes a speed adjustment coefficient calculation process; according to the state recognition result of step S2, the adaptive feed speed control system calculates the speed adjustment coefficient α in real time; the speed adjustment coefficient α is adaptively adjusted by using a support vector machine algorithm;

[0040] The support vector machine algorithm uses a radial basis function as a kernel function, and the current, frequency and adjustment effect data collected in the initial learning stage are used to construct training samples; the effectiveness evaluation index of the speed adjustment coefficient is obtained by linear weighting of the current stability parameter and the frequency deviation parameter; according to the evaluation result, online learning is carried out regularly, and then the model parameters are updated to adapt to different working conditions;

[0041] The adaptive feed speed control system dynamically determines the specific value of the speed adjustment coefficient in the corresponding range according to the current fluctuation degree and the amplitude of the frequency deviation from the reference;

[0042] In step S3, the phased adaptive strategy includes an initial learning stage and an intelligent optimization stage, the initial learning stage uses preset parameters for speed adjustment and collects training data, and the intelligent optimization stage uses a support vector machine algorithm to predict the optimal speed adjustment coefficient according to historical data.

[0043] In step S3, the first 100 interpolation periods use an empirical adjustment mode, and the speed adjustment coefficient α is set according to the identified state; the specific calculation rule is:

[0044] In the short circuit state, α is set to a value less than 1, the feed speed needs to be reduced to avoid wire breakage and processing quality decline, and the value range of α is [0.9, 1.0);

[0045] In the normal discharge state, α is set to 1, the current feed speed is maintained to maintain stable processing state;

[0046] In the no-load state, α is set to a value greater than 1, the feed speed needs to be increased to improve processing efficiency, and the value range of α is (1.0, 1.1];

[0047] The feature vector [δ f ,σ,η] is collected as training data, wherein the adjustment effect evaluation index η=0.6×(1-|f target -f actual | / ftarget )+0.4 x (1 - s), f target is the target frequency value, f actual is the actual frequency value; from the 101st interpolation period, the support vector machine mode is switched to, the radial basis function kernel is adopted, the parameter setting is regularized parameter C = 100, kernel parameter g = 0.1, loss function parameter e = 0.01, and the model performance is ensured by Z-score standardization, grid search optimization and 5-fold cross validation;

[0048] The adaptive feed speed control system is configured with multiple safety protection mechanisms: the speed adjustment coefficient a is limited in the range of 0.9-1.1 to prevent excessive adjustment, the feed speed is forcibly set to zero and an alarm is reported when short-circuit state is detected for 10 consecutive interpolation periods, and the model parameters are updated online every 500 interpolation periods to adapt to different working conditions. In actual application, the standard deviation of the adjustment coefficient a is less than 0.05, and the system response time is less than 5 milliseconds.

[0049] In step S4, the time division interpolation algorithm includes a straight line interpolation algorithm and a circular arc interpolation algorithm, the straight line interpolation algorithm adjusts the displacement increment through the speed adjustment coefficient, and the circular arc interpolation algorithm adjusts the angle increment through the speed adjustment coefficient;

[0050] In step S4, the EtherCAT protocol is adopted, the communication period and the interpolation period are synchronized to 1ms, it is ensured that the control command can be timely and accurately transmitted to the actuator, and the improved time division interpolation algorithm dynamically adjusts the interpolation step length in the interpolation process by introducing the speed adjustment coefficient a;

[0051] As shown in Figure 4 , the straight line interpolation algorithm is based on the geometric parameter calculation of the starting point coordinates (x s ,y s ) and the end point coordinates (x e ,y e ) of the given straight line segment, and the geometric relationship of the straight line is shown in the figure, including the angle θ and the influence of the contour speed f on the interpolation process. The calculation formula of the straight line length is L = sqrt((y e -y s ) 2 +(x e -x s ) 2 ), and the calculation of the straight line angle θ uses the atan2 function: θ = atan2(y e -y s ,x e -x s ). The calculation rule of the reference axis pulse number N is: when |x e -x s |≥|y e -ys When |x| e -x s | / δ x Otherwise N = |y e -y s | / δ y , where δ x and δ y These are the pulse equivalents for the X and Y axes, respectively. The formula for calculating the feed rate is V. i =V i-1 ×α×1000 / T, where V i and V i-1 Let be the feed rates for the i-th and (i-1)-th interpolation cycles, respectively, and T be the interpolation cycle. Therefore, the interpolation step size is ΔL. i =V i ×T. Figure 4 The displacement increment ΔX i and ΔY i Corresponding to the unit period displacement increment ΔX i =ΔL i ×cosθ / δ x ΔY i =ΔL i ×sinθ / δ y ;where δ x and δ y These are the pulse equivalents for the X and Y axes, respectively.

[0052] By introducing a speed adjustment coefficient α, the system can dynamically adjust the interpolation step size according to the real-time machining status, thereby achieving adaptive linear interpolation control.

[0053] like Figure 5 As shown, the circular interpolation algorithm is based on the coordinates of the starting point of the circular arc (x... s ,y s ), endpoint coordinates (x) e ,y e ) and center coordinates (x c ,y c The geometric parameters of the circular arc are calculated. The figure shows the geometric relationship of the circular arc interpolation, including the calculation process of radius R, angle increment Δθ and polar angle step size.

[0054] The formula for calculating the radius R is R = sqrt((x) s -x c ) 2 +(y s -y c ) 2 ),

[0055] Starting angle θ s and termination angle θ eThe calculation formulas of the angles θ s and θ s are respectively θ c =atan2(y s -y c ,x e -x e ) and θ c =atan2(y e -y c ,x s -x s );

[0056] The calculation rule of the included angle θ total is: when (θ e -θ s )<0, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )-2π for a forward circular arc; when 0<(θ e -θ s )<2π, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )+2π for a reverse circular arc.

[0057] The calculation formula of the feed speed is V i =V i-1 ×α×1000 / T, wherein V i and V i-1 are respectively the feed speeds of the i-th and the i-1-th interpolation periods, and T is the interpolation period.

[0058] The interpolation step is ΔL i =V i ×T, and the center angle increment Δθ i in the current interpolation period is Δθ i =2×arcos(ΔL i / 2 / R).

[0059] Figure 5 The displacement increments ΔX i and ΔY i in the interpolation period correspond to the i-th interpolation period displacement increments: ΔX i =R×cos(θ s +ΣΔθ i ) / δ x , and ΔY i =R×sin(θ s +ΣΔθ i ) / δ ywhere δ x and δ y are the X-axis and Y-axis pulse equivalent, respectively. By introducing the velocity adjustment factor α, the system can dynamically adjust the interpolation step size according to the real-time processing state, realizing adaptive circular arc interpolation control.

[0060] As shown in Figure 6 , step S5 is the implementation process of the circular interpolation buffer mechanism. The complete algorithm logic from the start of interpolation type judgment to real-time loop processing is shown in the figure, including key steps such as event existence judgment, data generation, and buffer management;

[0061] The buffer is a circular structure, and its size is set to 5 columns. It is initialized before interpolation starts, and 5 groups of interpolation data are filled in advance;

[0062] In each interpolation period, the system follows the process shown in Figure 6 to first perform interpolation type judgment, then check if the event exists, and if it exists, pop the first column of data and call the corresponding interpolation algorithm. According to the different interpolation types, the system calls the linear interpolation algorithm or the circular arc interpolation algorithm to generate single-column data and update real-time variables;

[0063] Figure 6 The real-time loop processing part in shows the dynamic management process of buffer data. Specifically, for circular arc interpolation, the system accumulates the angle increment and performs the corresponding coordinate conversion. When all the data in the interpolation buffer are taken out, or the accumulated pulses / angles meet the continuation condition, the system continues to supplement new data columns; otherwise, it stops supplementing and triggers the completion event.

[0064] The adaptive feed speed control system of the bus-type wire cutting CNC system includes a CNC host, a bus communication module, a signal acquisition module, and a servo driver.

[0065] The CNC host is built-in with an adaptive feed speed control algorithm, including a state recognition module, a speed calculation module, and an interpolation control module.

[0066] The state recognition module uses a dual-parameter fusion algorithm to realize processing state classification,

[0067] The speed calculation module calculates the velocity adjustment factor α in real time according to the recognition result,

[0068] Figure 6 The interpolation control module manages the interpolation buffer and executes the improved time division method interpolation algorithm according to the algorithm flow shown in

[0069] The bus communication module uses EtherCAT and Profinet protocols to realize data exchange between the CNC host and the servo driver, and the communication period is synchronized with the interpolation period.

[0070] The signal acquisition module is used for acquiring discharge current and feedback frequency signals, and provides reliable data basis for state judgment and adaptive control of the system.

[0071] The adaptive feed speed control system has an abnormality detection function, that is, when a continuous short circuit state is detected, the feed speed is automatically reduced and an alarm is issued.

[0072] The machining size precision control of the adaptive feed speed control system is within the range of ±0.002mm, the feedback frequency is stably in the interval of 200-500Hz when normal discharging, and the system response time is less than 5ms.

[0073] In the application, the adaptive feed speed control system realizes adaptive control of the feed speed in the wire cutting machining process through the organic combination of the double-parameter fusion judgment mechanism, the improved time segmentation method interpolation algorithm and the ring buffer mechanism, and significantly improves the machining precision, efficiency and stability.

[0074] The double-parameter fusion identification method based on current stability and feedback frequency established in the application solves the problem of insufficient accuracy of traditional single parameter judgment. The interpolation algorithm using a dynamic speed adjustment coefficient realizes the deep fusion of bus communication and control strategy. The interpolation buffer management mechanism designed ensures the continuity and precision of the motion trajectory in the dynamic adjustment process. The control method has been successfully applied in production testing, verifying its effectiveness, stability and practicability in different complexity machining tasks, and providing technical support for the intelligent development of the bus type wire cutting numerical control system.

[0075] The application discloses an adaptive feed speed control method and system of a bus type wire cutting numerical control system. The method realizes intelligent classification of machining states by establishing a frequency reference benchmark in an idle state and adopting a double-parameter fusion identification method based on current stability parameters and frequency deviation parameters, and divides the machining process into three states of short circuit, normal discharging and idle state. Based on the state identification result, a phased adaptive strategy is adopted to calculate a speed adjustment coefficient. The application improves machining precision, efficiency and stability, and provides an intelligent control solution for the bus type wire cutting numerical control system.

[0076] Compared with the prior art, the application has the following beneficial effects:

[0077] (1) The accuracy of the processing state recognition is improved. The double-parameter fusion recognition method based on the current stability parameter and the frequency deviation parameter is established, compared with the single parameter judgment method of the prior art, the short circuit state, the normal discharge state and the no-load state can be accurately distinguished, the state misjudgment caused by the processing environment change, the electrode wire quality fluctuation and other factors is reduced, and the occurrence of the wire breaking, the surface quality decline and other processing problems is reduced.

[0078] (2) The intelligent adaptive speed regulation is realized. The support vector machine algorithm is adopted, the training data is established by the preset parameters in the early stage, and the support vector machine algorithm is adopted to intelligently calculate the optimal speed regulation coefficient in the later stage. The strategy can automatically adapt to different working conditions and material characteristics, avoids the dependence on operation experience and the subjectivity of manual parameter adjustment of the traditional method, and improves the stability and consistency of the processing quality.

[0079] (3) The special control algorithm system under the bus architecture is constructed. The special control algorithm integrating the state recognition, the intelligent regulation and the track management is designed, the deep fusion of the wire cutting control and the bus communication is realized through the improved time segmentation method interpolation algorithm and the ring buffer management mechanism, the high-speed communication and the real-time synchronization characteristics of the bus network are fully utilized, and the problem that the traditional control method has poor application effect under the bus architecture is solved. BRIEF DESCRIPTION OF DRAWINGS

[0080] The application will be further described in detail below in combination with the drawings and specific embodiments:

[0081] The accompanying drawings are as follows: Figure 1 Fig. 1 is a schematic diagram of the overall process of the control method of the application (showing the complete control process from recording the no-load state feedback frequency value to adopting the interpolation buffer mechanism);

[0082] The accompanying drawings are as follows: Figure 2 Fig. 2 is a schematic diagram of intelligent discharge state recognition (showing how to use the back propagation neural network (BP neural network) to recognize the discharge state according to the current and the frequency);

[0083] The accompanying drawings are as follows: Figure 3 Fig. 3 is a schematic diagram of adaptive regulation of the speed regulation coefficient (showing how to adaptively regulate the speed regulation coefficient by the support vector machine method in combination with the previous data);

[0084] The accompanying drawings are as follows: Figure 4 Fig. 4 is a schematic diagram of the principle of the linear interpolation time segmentation method of the application (showing the geometric relationship of the starting point (x s ,y s ), the terminal point (x e ,y e ), the included angle θ, the feed speed V and the displacement increments ΔX i and ΔY i in the linear interpolation);

[0085] Appendix Figure 5 This is a schematic diagram illustrating the principle of the circular interpolation time division method of the present invention (showing the starting point (x) in circular interpolation). s ,y s ), endpoint (x) e ,y e ), center coordinates (x) c ,y c ), radius R, angle increment Δθ, feed rate V, and displacement increment ΔX i and ΔY i (geometric relationship);

[0086] Appendix Figure 6 This is a schematic diagram of the adaptive control algorithm for feed speed of the present invention (showing the algorithm control logic of interpolation type judgment, event existence judgment, data generation and real-time loop processing);

[0087] Appendix Figure 7 This is a schematic diagram showing the changes in sampling points and frequency values ​​when the speed is too high;

[0088] Appendix Figure 8 This is a schematic diagram showing the changes in sampling points and frequency values ​​when the speed is too slow;

[0089] Appendix Figure 9 This is a schematic diagram showing the changes in sampling points and frequency values ​​during normal discharge;

[0090] Appendix Figure 10 It is a photographic illustration of a typical geometric part that has been cut and repaired. Detailed Implementation

[0091] Example 1:

[0092] like Figure 1 As shown, the adaptive feed speed control method for a bus-type wire EDM CNC system proposed in this invention includes key steps such as frequency reference benchmark establishment, dual-parameter fusion state judgment, speed adjustment coefficient calculation, bus communication transmission, and interpolation buffer management. The entire control process starts from signal acquisition, goes through state recognition and parameter calculation, and finally achieves adaptive feed speed control.

[0093] As shown in the figure, the adaptive feed speed control method of the bus type linear cutting numerical control system, the method is to establish the frequency reference benchmark in the idle state, using the dual parameter fusion identification method based on the current stability parameter and the frequency deviation parameter, the processing state is divided into short circuit, normal discharge and idle; Based on state recognition, using the adaptive strategy to calculate the speed adjustment coefficient: the initial stage uses the preset parameter mode for adjustment and data collection, the later stage uses the support vector machine algorithm to calculate the optimal feed speed adjustment coefficient according to the historical data; Through the improved time segmentation method interpolation algorithm and bus communication technology, the interpolation step is dynamically adjusted and the real-time control command is transmitted.

[0094] As shown in the figure, the adaptive feed speed control method is executed by the adaptive feed speed control system, the method comprises the following steps: Figure 1

[0095] Step S1: In the idle running state of the electrode wire, the feedback frequency value is continuously collected to establish the frequency reference benchmark;

[0096] Step S2: In the processing process, the discharge current fluctuation and the feedback frequency change are monitored synchronously, and the dual parameter fusion method based on the current stability parameter and the frequency deviation parameter is used to identify the processing state, and the processing state is classified into short circuit state, normal discharge state and idle state;

[0097] Step S3: According to the processing state recognition, the adaptive strategy is used to calculate the speed adjustment coefficient α, the short circuit state α takes a value less than 1 to reduce the feed speed, the normal discharge state α takes 1 to maintain the feed speed, and the idle state α takes a value greater than 1 to improve the feed speed; The feed speed value of the electrode wire in the next interpolation period is the feed speed value of the electrode wire in the last interpolation period multiplied by the speed adjustment coefficient α, and the new feed speed value is calculated;

[0098] Step S4: The adjusted feed speed value is multiplied by the interpolation period to calculate the interpolation step; The interpolation step is dynamically adjusted; Using the time segmentation method interpolation algorithm, the coordinate axis increment is calculated according to the adjusted interpolation step, and the adaptive adjustment of the feed speed is realized;

[0099] Step S5: The coordinate axis increment data after interpolation calculation is stored in the buffer area, and the corresponding data is taken out from the buffer area every interpolation period, and is transmitted to the servo driver through bus communication; The motor drives the feed transmission system to move, so as to ensure the discharge stability between the electrode wire and the workpiece.

[0100] ​Step S1 is a frequency reference establishment process; specifically, before the wire cutting machine tool of the bus type wire cutting numerical control system enters the processing state, the electrode wire is started and maintained to run in the idle state. At this time, the control system collects the feedback frequency signal through the signal acquisition module and records the continuous data, with a sampling frequency of 20K; experiments show that in most systems, the frequency value is usually maintained between 900Hz and 1200Hz, reflecting the typical state of high-speed movement of the electrode wire but no participation in the discharge load;

[0101] The frequency interval of the frequency signal serves as the reference basis for the subsequent processing state judgment, providing a stable reference basis for the adaptive feed speed control system, that is, by establishing a stable frequency reference in the idle state, the state change in the subsequent processing process is identified.

[0102] Step S2 is a double-parameter fusion discharge state judgment process, specifically, during the electrode wire discharge machining process, the system monitors the current feedback frequency value and current fluctuation through the signal acquisition module; to accurately judge the trend of the discharge state change, the system dynamically compares the real-time collected frequency value with the reference frequency interval established in the idle state, and analyzes the current fluctuation characteristics;

[0103] The specific state judgment rule is:

[0104] When the feedback frequency significantly increases and the current fluctuation is severe, it is determined to be a short circuit state, indicating that there is an abnormal situation such as short circuit, wire sticking or poor discharge in the processing area;

[0105] When the feedback frequency is stable in the 200Hz to 500Hz interval and the current fluctuation is stable, it is determined to be a normal discharge state, that is, the discharge between the electrode wire and the workpiece is stable, the cutting load is moderate, and it is in an ideal processing state;

[0106] When the feedback frequency is close to the idle reference and the current is close to zero, it is determined to be an idle state, indicating that the feed speed is too slow and the cutting efficiency is low;

[0107] Compared with the traditional single-parameter judgment method, this double-parameter fusion judgment method can more accurately identify complex discharge states and avoid misjudgment caused by single-parameter fluctuation.

[0108] As shown in Figure 2 , step S2 adopts a double-parameter fusion method implemented by an intelligent discharge state recognition system, wherein the current stability parameter is a statistical characteristic value of the current fluctuation, and the frequency deviation parameter is the relative deviation value of the current frequency and the reference frequency, and the double-parameter fusion judgment of this step is realized through a back propagation neural network; specifically,

[0109] The intelligent discharge state recognition system performs 5-point sliding average filtering preprocessing on the feedback frequency signal and the discharge current signal, extracts the frequency deviation rate δf and current stability coefficient σ two characteristic parameters;

[0110] frequency deviation rate δ f = (f current -f baseline ) / f baseline , where f current is the current frequency value, f baseline is the idle reference frequency; current stability coefficient σ = ([(1 / n) Σ (I i -I mean ) 2 ]) 0.5 / I mean , where I i is the i-th current sampling value, I mean is the average current value, n is 20;

[0111] The back propagation neural network adopts a 2-5-3 structure, the input layer receives two characteristic parameters, the hidden layer uses a ReLU activation function, and the output layer uses a Softmax activation function to output probability values of short circuit, normal discharge and idle states;

[0112] The system establishes state determination criteria according to the training data: the short circuit state characteristics are δ f >1.5 and σ>0.25, the normal discharge state characteristics are -0.3≤δ f ≤0.8 and 0.05≤σ≤0.20, and the idle state characteristics are δ f close to 0 and σ<0.05; when identifying the state, the output with the maximum probability is selected as the determination result, and when the maximum probability value is lower than 0.8, the observation time is prolonged for re-determination;

[0113] Compared with the traditional single parameter judgment, the recognition accuracy is further improved;

[0114] In step S2, the neural network algorithm is used for processing state recognition to learn the mapping relationship between the double parameters and the processing state through network training;

[0115] The frequency reference benchmark is determined by multiple sampling and averaging, the processing state recognition is updated in real time, and the upper and lower limit values are set for the adjustment range of the speed adjustment coefficient to prevent abnormal adjustment, and a safety protection mechanism is used when continuous abnormal states are detected.

[0116] Step S3 includes a speed adjustment coefficient calculation process; according to the state recognition result of step S2, the speed adjustment coefficient α is calculated in real time by the adaptive feed speed control system; the speed adjustment coefficient α is adaptively adjusted by using a support vector machine algorithm;

[0117] The support vector machine algorithm adopts a radial basis function as a kernel function, and constructs training samples by using current, frequency and adjustment effect data collected in an initial learning stage; an effectiveness evaluation index of the speed adjustment coefficient is obtained by linear weighting of a current stability parameter and a frequency deviation parameter; according to an evaluation result, online learning is regularly performed, and model parameters are updated to adapt to different working conditions;

[0118] The adaptive feed speed control system dynamically determines the specific value of the speed adjustment coefficient in a corresponding range according to the current fluctuation degree and the amplitude of the frequency deviation from the reference;

[0119] In step S3, the phased adaptive strategy includes an initial learning stage and an intelligent optimization stage, the initial learning stage uses preset parameters for speed adjustment and collects training data, and the intelligent optimization stage uses a support vector machine algorithm to predict the optimal speed adjustment coefficient according to historical data.

[0120] In step S3, the first 100 interpolation periods use an empirical adjustment mode, and the speed adjustment coefficient α is set according to the identified state; the specific calculation rule is:

[0121] In the short circuit state, α is set to a value less than 1, the feed speed needs to be reduced to avoid wire breakage and processing quality decline, and the value range of α is [0.9, 1.0);

[0122] In the normal discharge state, α is set to 1, and the current feed speed is maintained to maintain a stable processing state;

[0123] In the no-load state, α is set to a value greater than 1, the feed speed needs to be increased to improve processing efficiency, and the value range of α is (1.0, 1.1];

[0124] The characteristic vector [δ f ,σ,η] is collected as training data, wherein the adjustment effect evaluation index η = 0.6 × (1-|f target -f actual | / f target )+0.4×(1-σ), f target is the target frequency value, and f actual is the actual frequency value; from the 101th interpolation period, the support vector machine mode is switched to, a radial basis function kernel is used, the parameter setting is regularized parameter C = 100, kernel parameter γ = 0.1, loss function parameter ε = 0.01, and the model performance is ensured by Z-score standardization, grid search optimization and 5-fold cross-validation;

[0125] The adaptive feed speed control system is configured with multiple safety protection mechanisms: the speed adjustment coefficient a is limited in the range of 0.9-1.1 to prevent excessive adjustment, the feed speed is forcibly set to zero and an alarm is given when short circuit state is detected for 10 consecutive interpolation periods, and the model parameters are updated online every 500 interpolation periods to adapt to different working conditions. In actual application, the standard deviation of the adjustment coefficient a is less than 0.05, and the system response time is less than 5 milliseconds.

[0126] In step S4, the time division interpolation algorithm includes a straight line interpolation algorithm and a circular arc interpolation algorithm, the straight line interpolation algorithm adjusts the displacement increment through the speed adjustment coefficient, and the circular arc interpolation algorithm adjusts the angle increment through the speed adjustment coefficient;

[0127] In step S4, the EtherCAT protocol is used, the communication period and the interpolation period are synchronized to 1ms, the control command can be transmitted to the actuator in time and accurately, and the improved time division interpolation algorithm dynamically adjusts the interpolation step by introducing the speed adjustment coefficient a in the interpolation process;

[0128] As shown in Figure 4 , the straight line interpolation algorithm is based on the geometric parameter calculation of the starting point coordinates (x s ,y s ) and the end point coordinates (x e ,y e ) of the given straight line segment, and the geometric relationship of the straight line is shown in the figure, including the influence of the included angle θ and the contour speed f on the interpolation process. The calculation formula of the straight line length is L=sqrt((y e -y s ) 2 +(x e -x s ) 2 ), and the calculation of the straight line included angle θ uses the atan2 function: θ=atan2(y e -y s ,x e -x s ). The calculation rule of the reference axis pulse number N is: when |x e -x s |≥|y e -y s |, N=|x e -x s | / δ x , otherwise N=|y e -y s | / δ y , where δ x and δ y are the pulse equivalents of the X-axis and Y-axis respectively. The calculation formula of the feed speed is V i =V i-1×α×1000 / T, where V i and V i-1 Let be the feed rates for the i-th and (i-1)-th interpolation cycles, respectively, and T be the interpolation cycle. Therefore, the interpolation step size is ΔL. i =V i ×T. Figure 4 The displacement increment ΔX i and ΔY i Corresponding to the unit period displacement increment ΔX i =ΔL i ×cosθ / δ x ΔY i =ΔL i ×sinθ / δ y ;where δ x and δ y These are the pulse equivalents for the X and Y axes, respectively.

[0129] By introducing a speed adjustment coefficient α, the system can dynamically adjust the interpolation step size according to the real-time machining status, thereby achieving adaptive linear interpolation control.

[0130] like Figure 5 As shown, the circular interpolation algorithm is based on the coordinates of the starting point of the circular arc (x... s ,y s ), endpoint coordinates (x) e ,y e ) and center coordinates (x c ,y c The geometric parameters of the circular arc are calculated. The figure shows the geometric relationship of the circular arc interpolation, including the calculation process of radius R, angle increment Δθ and polar angle step size.

[0131] The formula for calculating the radius R is R = sqrt((x) s -x c ) 2 +(y s -y c ) 2 ),

[0132] Starting angle θ s and termination angle θ e The calculation formulas are θ s =atan2(y s -y c ,x s -x c ) and θ e =atan2(y e -y c ,x e -x c );

[0133] included angle θtotal The calculation rule is: when (θ e -θ s )<0, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )-2π; when 0<(θ e -θ s )<2π, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )+2π.

[0134] The calculation formula of the feed speed is V i =V i-1 ×α×1000 / T, wherein V i and V i-1 are the feed speeds of the i-th and the i-1-th interpolation periods respectively, and T is the interpolation period.

[0135] The interpolation step is ΔL i =V i ×T, and the center angle increment Δθ i in the current interpolation period is Δθ i =2×arcos(ΔL i / 2 / R).

[0136] Figure 5 The displacement increments ΔX i and ΔY i in the formula correspond to the displacement increments of the i-th interpolation period: ΔX i =R×cos(θ s +ΣΔθ i ) / δ x , and ΔY i =R×sin(θ s +ΣΔθ i ) / δ y , wherein δ x and δ y are the pulse equivalents of the X-axis and the Y-axis respectively. By introducing the speed adjustment coefficient α, the system can dynamically adjust the interpolation step according to the real-time processing state, and realize adaptive circular interpolation control.

[0137] As Figure 6As shown, step S5 is the implementation process of the ring-shaped interpolation buffer mechanism, the complete algorithm logic from the start of interpolation type judgment to real-time loop processing is shown, including key steps such as event existence judgment, data generation and buffer management;

[0138] The buffer is in a ring structure, and is set to have a size of 5 columns. Before interpolation starts, 5 groups of interpolation data are filled in advance;

[0139] In each interpolation period, the system performs Figure 6 As shown in the flowchart, first, the interpolation type is judged, then it is checked whether the event exists, if it exists, the first column of data is popped out and the corresponding interpolation algorithm is called. According to the different interpolation types, the system calls the straight line interpolation algorithm or the circular arc interpolation algorithm to generate single column data and update real-time variables;

[0140] Figure 6 The real-time loop processing part in the figure shows the dynamic management process of the buffer data, specifically, for circular arc interpolation, the system accumulates the angle increment and performs corresponding coordinate conversion. When all the data in the interpolation buffer are taken out, or the accumulated pulse / angle meets the continuation condition, the system continues to supplement new data columns; otherwise, it stops supplementing and triggers the completion event.

[0141] The adaptive feed speed control system of the bus type wire cutting numerical control system, the hardware architecture of adaptive feed speed control includes numerical control host, bus communication module, signal acquisition module and servo driver;

[0142] The numerical control host is built-in adaptive feed speed control algorithm, including state recognition module, speed calculation module and interpolation control module;

[0143] The state recognition module realizes the classification of machining states by using a double-parameter fusion algorithm,

[0144] The speed calculation module calculates the speed adjustment coefficient α in real time according to the recognition result,

[0145] The interpolation control module manages the interpolation buffer and executes the improved time division method interpolation algorithm according to the algorithm flow shown in the figure; Figure 6

[0146] The bus communication module uses EtherCAT and Profinet protocols to realize data exchange between the numerical control host and the servo driver, and the communication period is synchronized with the interpolation period;

[0147] The signal acquisition module is used to collect the discharge current and feedback frequency signals to provide a reliable data basis for the state judgment and adaptive control of the system. The servo driver receives and executes the feed control instructions sent by the numerical control host, and drives the motor to move accurately according to the adjusted speed;

[0148] ​The adaptive feed speed control system has an abnormality detection function, that is, when a continuous short circuit state is detected, the feed speed is automatically reduced and an alarm is issued;

[0149] The machining size precision control of the adaptive feed speed control system is within the range of ±0.002mm, the feedback frequency is stably in the interval of 200-500Hz when discharging normally, and the system response time is less than 5ms.

[0150] Example 2:

[0151] In order to verify the effectiveness of the control algorithm, comparative experiments of three typical machining states are carried out in this example.

[0152] The experiment is based on a bus type medium-speed wire cutting numerical control system, and an EtherCAT bus communication protocol is used, and the communication period and the interpolation period are synchronized to 5ms.

[0153] As shown in Figure 7 , in the state of too fast speed, the feed speed is set to 150mm / min, the feedback frequency fluctuates sharply between 100-300Hz, the current fluctuation amplitude is more than 30%, frequent short circuit phenomenon occurs in the machining process, and the machining surface roughness Ra>3.2μm.

[0154] As shown in Figure 8 , in the state of too slow speed, the feed speed is set to 20mm / min, the feedback frequency is maintained in the interval of 800-1000Hz, close to the idle state, the current change is less than 5%, and the machining efficiency is only 40% of the normal state.

[0155] As shown in Figure 9 , after using the algorithm of the application, the system automatically adjusts the feed speed in the range of 80-120mm / min, the feedback frequency is stably in the interval of 200-500Hz, the current fluctuation is controlled within 10%, and the machining surface roughness Ra≤1.6μm.

[0156] Example 3:

[0157] In this example, a typical geometric figure machining verification experiment is carried out, a circular part is selected, the workpiece material is 45# steel, the thickness is 30mm, and a cut-one-repair-two process is used. As shown in Figure 10 , the machined parts are 4 cylindrical bodies with a diameter of 10mm.

[0158] A screw micrometer (resolution 0.001mm) is used to measure the size of the finished product, and the specific measurement results are shown in Table 1.

[0159] Table 1 Measurement results of actual machining cylindrical body samples

[0160]

[0161]

[0162] From table 1, it can be seen that the diameter deviation of the cylindrical sample is within the range of ±0.002mm, wherein the No.3 sample achieves the ideal machining effect of zero deviation.

[0163] The experimental results show that, by using the control method of the application, the machining size precision is controlled within the range of ±0.002mm, the feedback frequency is stabilized in the interval of 200-500Hz during normal discharge, and the machining stability and surface quality are significantly improved.

[0164] The double-parameter fusion recognition method based on current stability and feedback frequency solves the problem of insufficient accuracy of traditional single parameter judgment. The interpolation algorithm with dynamic speed adjustment coefficient realizes the deep fusion of bus communication and control strategy. The designed interpolation buffer management mechanism guarantees the continuity and precision of the motion trajectory in the dynamic adjustment process. The control method has been successfully applied in production test, verifying its effectiveness, stability and practicability in different complexity machining tasks, and providing technical support for the intelligent development of bus type wire cutting numerical control system.

[0165] The above only describes the preferred embodiments of the application and is not intended to limit the application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A method for adaptive feed rate control of a bus-type numerically controlled linear cutting system, characterized in that: The method divides the processing state into short circuit, normal discharge and no load by establishing the frequency reference benchmark in no load state and using the double parameter fusion identification method based on current stability parameter and frequency deviation parameter; Based on the state identification, a phased adaptive strategy is used to calculate the speed adjustment coefficient: in the early stage, a preset parameter mode is used for adjustment and data collection, and in the late stage, a support vector machine algorithm is used to intelligently calculate the optimal feed speed adjustment coefficient according to historical data; Through the improved time segmentation interpolation algorithm and bus communication technology, the interpolation step is dynamically adjusted and the real-time control command is transmitted.

2. The adaptive feed rate control method for a bus-type NC linear cutting system according to claim 1, wherein: The adaptive feed speed control method is executed by an adaptive feed speed control system, and the method comprises the following steps; Step S1: In the electrode wire no load running state, the feedback frequency value is continuously collected to establish the frequency reference benchmark; Step S2: In the processing process, the discharge current fluctuation and the feedback frequency change are monitored synchronously, and the double parameter fusion method based on the current stability parameter and the frequency deviation parameter is used to identify the processing state, so that the processing state is classified into short circuit state, normal discharge state and no load state; Step S3: According to the processing state identification, a phased adaptive strategy is used to calculate the speed adjustment coefficient α, and the feed speed is adjusted; The feed speed value of the electrode wire in the next interpolation period is the feed speed value of the electrode wire in the last interpolation period multiplied by the speed adjustment coefficient α, and the new feed speed value is calculated; Step S4: The adjusted feed speed value is multiplied by the interpolation period to calculate the interpolation step; The interpolation step is dynamically adjusted; The time segmentation interpolation algorithm is used to calculate the increments of each coordinate axis according to the adjusted interpolation step, so as to realize the adaptive adjustment of the feed speed; Step S5: The coordinate axis increment data after interpolation calculation is stored in the buffer area, and the corresponding data is taken out from the buffer area in each interpolation period, and is transmitted to the servo driver through bus communication; The motor drives the feed transmission system to move, so as to ensure the discharge stability between the electrode wire and the workpiece.

3. The adaptive feed rate control method for a bus-type NC linear cutting system according to claim 2, wherein: Step S1 is the frequency reference benchmark establishment process; Specifically, before the wire cutting machine of the bus type wire cutting numerical control system enters the processing state, the electrode wire is started and maintained in the no load state. At this time, the control system collects the feedback frequency signal through the signal acquisition module and records the continuous data, so that the frequency value of the frequency signal reflects the typical state that the electrode wire moves at high speed but does not participate in the discharge load; The frequency interval of the frequency signal is used as the benchmark reference for subsequent processing state judgment, and a stable reference basis is provided for the adaptive feed speed control system, that is, the state change in the subsequent processing process is identified by establishing a stable frequency benchmark in the no load state.

4. The adaptive feed rate control method for a bus-type NC linear cutting system according to claim 2, wherein: Step S2 is the double parameter fusion discharge state judgment process, specifically: in the electrode wire discharge processing process, the system monitors the current feedback frequency value and current fluctuation through the signal acquisition module; In order to accurately judge the change trend of the discharge state, the system dynamically compares the real-time collected frequency value with the benchmark frequency interval established in the no load state, and analyzes the current fluctuation characteristics; The specific state judgment rule is: When the feedback frequency is significantly increased and the current fluctuates sharply, it is determined to be a short-circuit state, indicating that there is an abnormal situation of short circuit, sticky wire or poor discharge in the processing area; When the feedback frequency is stable in the interval of 200Hz to 500Hz and the current fluctuation is stable, it is determined to be a normal discharge state, that is, the discharge between the electrode wire and the workpiece is stable, the cutting load is moderate, and it is in an ideal processing state; When the feedback frequency is close to the no-load reference and the current is close to zero, it is determined to be a no-load state, indicating that the feed speed is too slow and the cutting efficiency is low.

5. The adaptive feed rate control method for a bus-type linear CNC system according to claim 4, wherein: In step S2, a double-parameter fusion method implemented by an intelligent discharge state recognition system is used, wherein the current stability parameter is a statistical characteristic value of current fluctuation, and the frequency deviation parameter is a relative deviation value of the current frequency and the reference frequency. The double-parameter fusion judgment in this step is realized by a back propagation neural network. Specifically: The intelligent discharge state recognition system carries out multi-point sliding average filtering pretreatment on the feedback frequency signal and the discharge current signal, extracts two characteristic parameters of frequency deviation rate δ f and current stability coefficient σ Frequency deviation rate δ f = (f current - f baseline ) / f baseline , where f current is the current frequency value, f baseline is the empty reference frequency; current stability coefficient σ = ([(1 / n)Σ(I i - I mean ) 2 ]) 0.5 / I mean , where I i is the i-th current sampling value, I mean is the average current value, and n is a preset constant value; The back propagation neural network adopts a 2-5-3 structure, the input layer receives two characteristic parameters, the hidden layer uses a ReLU activation function, and the output layer uses a Softmax activation function to output probability values of short-circuit, normal discharge and no-load states; The system establishes state determination criteria according to the training data: short-circuit state feature is δ f >1.5 and σ>0.25, normal discharge state feature is -0.3≤δ f ≤0.8 and 0.05≤σ≤0.20, no-load state feature is δ f close to 0 and σ<0.05; the output with the largest probability is selected as the determination result in state recognition, and when the largest probability value is lower than 0.8, the observation time is prolonged for re-determination; In step S2, the processing state recognition uses a neural network algorithm to learn the mapping relationship between the double parameters and the processing state through network training; The frequency reference is determined by averaging multiple samples, the processing state recognition is updated in real time, the adjustment range of the speed adjustment coefficient is set with upper and lower limit values to prevent abnormal adjustment, and a safety protection mechanism is enabled when continuous abnormal states are detected.

6. The adaptive feed rate control method for a bus-type linear CNC system according to claim 2, wherein: Step S3 includes a speed adjustment coefficient calculation process; According to the state recognition result of step S2, the adaptive feed speed control system calculates the speed adjustment coefficient α in real time; the speed adjustment coefficient α is adaptively adjusted by using a support vector machine algorithm; The support vector machine algorithm uses a radial basis function as a kernel function, and constructs training samples from the current, frequency and adjustment effect data collected in the initial learning stage; the effectiveness evaluation index of the speed adjustment coefficient is obtained by linear weighting of the current stability parameter and the frequency deviation parameter; The adaptive feed speed control system dynamically determines the specific value of the speed adjustment coefficient within the corresponding range according to the current fluctuation degree and the frequency deviation from the reference; In step S3, the phased adaptive strategy includes an initial learning stage and an intelligent optimization stage. In the initial learning stage, preset parameters are used for speed adjustment and training data is collected. In the intelligent optimization stage, the support vector machine algorithm is used to predict the optimal speed adjustment coefficient according to historical data.

7. The adaptive feed rate control method for a bus-type linear CNC system according to claim 6, wherein: In step S3, the first preset number of interpolation periods use an empirical adjustment mode to set the speed adjustment coefficient α according to the recognized state; the specific calculation rule is: In the short-circuit state, α is set to a value less than 1, the feed speed needs to be reduced to avoid wire breakage and processing quality degradation, and the value range of α is [0.9, 1.0); In the normal discharge state, α is set to 1 to maintain the current feed speed to keep the processing state stable; In the no-load state, α is set to a value greater than 1, the feed speed needs to be increased to improve processing efficiency, and the value range of α is (1.0, 1.1]. The feature vector [δ f , σ, η] is collected as training data, wherein the effect evaluation index η = 0.6 × (1 - |f target -f actual | / f target ) + 0.4 × (1 - σ), f target is a target frequency value, and f actual is an actual frequency value. After the preset number of interpolation periods, switch to the support vector machine mode from the next interpolation period, adopt the radial basis function kernel, and set the parameters as the regularization parameter C=100, the kernel parameter γ=0.1, and the loss function parameter ε=0.01, and ensure the model performance through Z-score standardization, grid search optimization and 5-fold cross-validation; The adaptive feed speed control system is configured with multiple safety protection mechanisms: the speed adjustment coefficient α is limited in the range of 0.9-1.1 to prevent excessive adjustment, and when short-circuit state is detected for a plurality of continuous interpolation periods, the feed speed is forcibly set to zero and an alarm is triggered.

8. The adaptive feed rate control method for a bus-type linear CNC system according to claim 7, wherein: In step S4, the time division interpolation algorithm includes a straight line interpolation algorithm and a circular arc interpolation algorithm, the straight line interpolation algorithm adjusts the displacement increment through the speed adjustment coefficient, and the circular arc interpolation algorithm adjusts the angle increment through the speed adjustment coefficient; In step S4, the EtherCAT protocol is adopted, the communication period and the interpolation period are synchronized to a preset time length, and the improved time division interpolation algorithm dynamically adjusts the interpolation step through the introduction of the speed adjustment coefficient α during the interpolation process; The straight line interpolation algorithm is based on the given straight line segment starting point coordinates (x s ,y s ) and end point coordinates (x e ,y e ) to perform geometric parameter calculation, the geometric relationship of the straight line, including the angle θ and the contour speed f on the influence of the interpolation process; The formula for calculating the straight line length is L = sqrt((y e -y s ) 2 +(x e -x s 2 ),​ The calculation of the straight line angle θ uses the atan2 function: θ = atan2(y e -y s ,x e -x s ); The calculation rule of the reference axis pulse number N is: when |x e -x s |≥|y e -y s |, N=|x e -x s | / δ x , otherwise N=|y e -y s | / δ y , wherein δ x and δ y are the pulse equivalent of the X axis and the Y axis respectively. The calculation formula of the feed speed is V i = V i-1 × α × 1000 / T, wherein V i and V i-1 are the feed speeds of the i-th and the i-1-th interpolation periods, T is the interpolation period, and the interpolation step is ΔL i = V i × T; displacement increment ΔX i and ΔY i corresponding to a unit period displacement increment ΔX i = ΔL i × cos θ / δ x , ΔY i = ΔL i × sin θ / δ y ; where δ x and δ y are the X-axis and Y-axis pulse equivalents, respectively. By introducing the speed adjustment coefficient α, the adaptive feed speed control system dynamically adjusts the interpolation step according to the real-time machining state, and adaptive straight line interpolation control is realized. The circular arc interpolation algorithm calculates the geometric parameters of the circular arc based on the starting point coordinates (x s ,y s ), the ending point coordinates (x e ,y e ) and the center coordinates (x c ,y c ) of the circular arc, and the geometric relationship of the circular arc interpolation, including the calculation process of the radius R, the angle increment Δθ and the polar angle step. The formula for calculating the radius R is R = sqrt((x s -x c ) 2 +(y s -y c ) 2 ), Starting angle θ s and ending angle θ e are calculated as θ s = atan2(y s - y c , x s - x c ) and θ e = atan2(y e - y c , x e - x c ); angle θ total The calculation rule is: when (θ e -θ s )<0, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )-2π; when 0<(θ e -θ s )<2π, θ total =θ e -θ s , otherwise θ total =(θ e -θ s )+2π. The calculation formula of the feed speed is V i = V i-1 × α × 1000 / T, wherein V i and V i-1 are the feed speeds of the i-th and the i-1-th interpolation periods, and T is the interpolation period. The interpolation step size is ΔL i = V i × T, the center angle increment Δθ in the current interpolation period i is Δθ i = 2 × arcos(ΔL i / 2 / R); displacement increment ΔX i and ΔY i corresponding to the i-th interpolation period displacement increment: ΔX i = R x cos(θ s + ΣΔθ i ) / δ x , ΔY i = R x sin(θ s + ΣΔθ i ) / δ y ; where δ x and δ y are the X-axis and Y-axis pulse equivalents, respectively; By introducing the speed adjustment coefficient α, the adaptive feed speed control system dynamically adjusts the interpolation step according to the real-time machining state, and adaptive circular arc interpolation control is realized.

9. The adaptive feed rate control method for a bus-type linear CNC system of claim 6, wherein: In step S5, the implementation process of the ring-shaped interpolation buffer mechanism, the complete algorithm logic from the start of interpolation type judgment to real-time loop processing, includes the steps of event existence judgment, data generation and buffer management; The buffer is in a ring structure and is set to 5 columns in size, and is initialized before interpolation starts and is filled with 5 groups of interpolation data in advance; In each interpolation period, the adaptive feed speed control system first performs interpolation type judgment, then checks whether an event exists, and if an event exists, pops out the first column of data and calls the corresponding interpolation algorithm; According to the different interpolation types, the adaptive feed speed control system calls the straight line interpolation algorithm or the circular arc interpolation algorithm, generates a single column of data and updates the real-time variables; The dynamic management process of the buffer data is as follows: for circular arc interpolation, the system accumulates the angle increment and performs corresponding coordinate conversion. When all the data in the interpolation buffer are taken out, or the accumulated pulses / angles meet the continuation condition, the system continues to supplement new data columns; otherwise, stop supplementing and trigger the completion event.

10. Adaptive feed rate control system for a bus type numerically controlled linear cutting system, characterized in that: The hardware architecture of the adaptive feed speed control includes a numerical control host, a bus communication module, a signal acquisition module and a servo driver; The numerical control host is built-in with an adaptive feed speed control algorithm, including a state recognition module, a speed calculation module and an interpolation control module; The state recognition module realizes machining state classification through a double-parameter fusion algorithm, The speed calculation module calculates the speed adjustment coefficient α in real time according to the recognition result, The interpolation control module manages the interpolation buffer and executes the improved time division interpolation algorithm according to the algorithm process shown in FIG. 6; By introducing the speed adjustment coefficient α, the adaptive feed speed control system dynamically adjusts the interpolation step according to the real-time machining state, and adaptive straight line interpolation control is realized. By introducing the speed adjustment coefficient α, the adaptive feed speed control system dynamically adjusts the interpolation step according to the real-time machining state, and adaptive circular arc interpolation control is realized. The bus communication module adopts EtherCAT and Profinet protocols to realize data exchange between the numerical control host and the servo driver, and the communication cycle is synchronized with the interpolation cycle; The signal acquisition module is used to collect discharge current and feedback frequency signals, providing reliable data basis for system state judgment and adaptive control. The servo driver receives and executes the feeding control instructions sent by the numerical control host, and drives the motor to perform accurate motion according to the adjusted speed; The adaptive feed speed control system has an abnormality detection function, that is, when a continuous short circuit state is detected, the feed speed is automatically reduced and an alarm is issued; The adaptive feed speed control system performs adaptive control of the feed speed of the wire cutting machining process through a double-parameter fusion judgment mechanism, an improved time division method interpolation algorithm and a ring buffer mechanism.

Images