Cutter state real-time monitoring and compensating method, system and equipment and medium

Abstract

The invention discloses a cutter state real-time monitoring and compensating method, system and equipment and a medium, and relates to numerical control precision machining. In the method, a cutting edge optical image, an acoustic emission signal time domain waveform, a vibration acceleration parameter and real-time temperature data of a cutting area of a cutter in the machining process are collected; measuring the maximum width and depth of a wear belt through an image to form a wear characteristic quantized value, and judging a wear over-limit state; judging the damage state based on the continuous ultralimit time of the instantaneous mutation rate of the acoustic emission signal; judging the abnormal vibration grade and state through vibration spectrum analysis; and inputting the vibration spectrum and the temperature data into a prediction model to obtain a position offset prediction quantity. And calculating a size deviation value according to a wear / damage state, calculating a coordinate offset according to vibration abnormity, generating a cutter geometric compensation instruction or a motion trail compensation instruction, and completing real-time compensation operation. According to the scheme, the fine abnormal change of the cutter in the machining process can be captured in time, and compensation is conducted according to the fine abnormal change.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of numerical control precision machining, and in particular to a tool state real-time monitoring and compensation method, system, device and medium. BACKGROUND

[0002] When machining precision parts such as molds in a numerical control center, the existing technology relies on a traditional mode for tool state management. On the one hand, an operator stops and inspects at regular intervals, and judges the tool service life by observing the tool edge wear with the naked eye or relying on machining experience, and the tool is replaced only when obvious problems are found or the preset machining time is reached. On the other hand, although some devices are equipped with single parameter monitoring devices (such as vibration sensors or temperature sensors), they can only set fixed thresholds for vibration amplitude or temperature value, and trigger an alarm when the threshold is exceeded, and the monitoring data is not associated with other state analysis such as tool wear and thermal deformation.

[0003] However, this traditional mode has significant technical limitations: it cannot capture subtle abnormal changes in the tool during machining, and when the tool has subtle abnormal changes, it will affect the machining precision, resulting in unqualified parts. SUMMARY

[0004] To solve the above technical problems, the present application provides a tool state real-time monitoring and compensation method, system, device and medium.

[0005] In a first aspect of the present application, a tool state real-time monitoring and compensation method is provided: Obtaining an edge optical image of a tool during machining, a time-domain waveform of a sound emission signal generated by machining, a vibration acceleration parameter of the tool, and real-time temperature data of a cutting area; Measuring the maximum width value and the maximum depth value of the wear zone in the edge optical image, combining the maximum width value and the maximum depth value into a wear feature quantization value, and if the wear feature quantization value is greater than a preset wear threshold, determining that the tool is in a wear overrun state; Calculating an instantaneous mutation rate according to the time-domain waveform of the sound emission signal, and if the instantaneous mutation rate continuously exceeds the upper limit of a preset sound emission fluctuation range for a set time length threshold, determining that the tool is in a broken state; Calculating a vibration spectrum from the vibration acceleration parameter, determining a vibration anomaly level through the vibration spectrum, and if the vibration anomaly level exceeds a preset safety level, determining that the tool is in a vibration anomaly state; Inputting the vibration spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction value; When the tool is in the wear overrun state and / or the breakage state, a size deviation value is calculated based on the wear feature quantification value, and when the tool is in the vibration abnormality state, a coordinate offset amount is calculated based on the position offset prediction value; A tool geometry compensation instruction is generated according to the size deviation value, and / or a motion trajectory compensation instruction is generated according to the coordinate offset amount, and a compensation operation is completed based on the tool geometry compensation instruction or the motion trajectory compensation instruction.

[0006] By adopting the above technical solutions, all-round and real-time monitoring and accurate compensation of size deviation and position offset caused by tool wear, breakage, vibration and thermal deformation are realized, effectively solving the monitoring lag problem of traditional manual inspection and single-parameter threshold alarm. Through multi-source sensor data fusion and correlation analysis, the method can timely capture and determine the state at the initial stage of slight abnormal change of the tool, and then automatically trigger the corresponding geometric compensation or motion trajectory compensation instruction to dynamically correct the error in the machining process, thereby significantly reducing the machining precision deviation and part unqualified rate caused by abnormal tool state, and ensuring the stable and high-quality machining of precision parts such as molds.

[0007] Optionally, the maximum width value and the maximum depth value of the wear zone in the cutting edge optical image are measured, and the maximum width value and the maximum depth value are combined into a wear feature quantification value, including: The cutting edge optical image is subjected to noise reduction and contrast enhancement processing, and the contour boundary of the target region is extracted; The maximum distance along the preset direction of the contour boundary is measured to obtain the maximum width value of the target region, and the maximum distance along the direction perpendicular to the preset direction is measured to obtain the maximum depth value of the target region; The maximum width value and the maximum depth value are weighted and summed according to a preset wear weight to obtain the wear feature quantification value.

[0008] By adopting the above technical solutions, the quantification accuracy and reliability of the tool cutting edge wear state can be effectively improved. By performing noise reduction and contrast enhancement processing on the optical image, the recognizability of the wear region is significantly improved, facilitating accurate extraction of the contour boundary. Then, by measuring the maximum distance along the preset direction and the direction perpendicular thereto, the maximum width and depth values are obtained, respectively, ensuring comprehensive capture of the wear feature in two dimensions. Finally, the multi-dimensional wear information is fused into a single quantification index by weighted summation according to the preset weight, which not only simplifies the judgment logic but also enhances the adaptability to different wear modes, providing accurate and stable data basis for subsequent tool state determination and compensation.

[0009] Optionally, the calculation of the instantaneous mutation rate according to the time-domain waveform of the acoustic emission signal includes: Filtering and denoising the time-domain waveform of the acoustic emission signal to extract a target signal segment; Selecting a first sampling point and a second sampling point in the target signal segment, calculating a first difference value between the amplitude of the second sampling point and the amplitude of the first sampling point and taking the absolute value as a first absolute value, and calculating the sampling time interval of the first sampling point and the second sampling point, and taking the ratio of the first absolute value to the sampling time interval as the instantaneous mutation rate, the first sampling point and the second sampling point being sampling points of any two consecutive time points in the target signal segment.

[0010] By adopting the above technical solution, early, sensitive and reliable identification of the tool damage state can be realized. The original acoustic emission signal is filtered and denoised to effectively eliminate environmental noise and mechanical background interference, thereby ensuring the signal quality for subsequent analysis. The instantaneous change characteristics of the signal in the time domain are converted into a quantifiable index by extracting the target signal segment and calculating the amplitude mutation rate between consecutive sampling points, thereby greatly improving the ability to capture instantaneous burst events such as the generation of micro cracks and material peeling. The index has clear physical meaning and efficient calculation, and provides key data basis for real-time and accurate determination of the tool damage state, thereby effectively avoiding processing quality defects and equipment damage caused by sudden tool failure.

[0011] Optionally, the vibration frequency spectrum is calculated according to the vibration acceleration parameter, the vibration abnormality level is determined through the vibration frequency spectrum, and if the vibration abnormality level exceeds a preset safety level, the tool is determined to be in a vibration abnormality state, which includes: Performing Fourier transform on the vibration acceleration parameter to obtain a vibration frequency spectrum, identifying a dominant peak frequency in the vibration frequency spectrum, the dominant peak frequency being a frequency with an amplitude exceeding a preset amplitude threshold, extracting a vibration acceleration amplitude corresponding to the dominant peak frequency from the vibration frequency spectrum, and calculating an energy proportion of the vibration energy of the dominant peak frequency in the total energy of the vibration frequency spectrum; Calculating an amplitude deviation degree of the vibration acceleration amplitude from a preset vibration amplitude range, calculating an energy deviation degree of the energy proportion from a preset energy distribution threshold, and weighting and summing the amplitude deviation degree and the energy deviation degree according to a preset weight to obtain a vibration abnormality index; Matching the vibration abnormality index with a preset vibration abnormality level division interval to obtain a vibration abnormality level, and determining that the tool is in the vibration abnormality state when the vibration abnormality level is higher than the preset safety level.

[0012] By adopting the technical scheme, multi-dimensional and refined comprehensive evaluation and early warning of the tool vibration state are realized. The method can effectively reveal the characteristic frequency components hidden in complex vibration by converting the time domain vibration signal into a frequency domain spectrum through Fourier transform; by simultaneously considering the amplitude deviation degree and energy concentration degree (energy proportion) of the dominant frequency, the abnormality degree is quantified from two key dimensions of vibration intensity and energy distribution, respectively, overcoming the shortcomings of high false positive rate and poor sensitivity of the single threshold method; finally, a comprehensive vibration abnormality index is generated by weighted fusion, and a preset grade interval is matched, so that the state determination result is more objective, accurate and reliable, which provides a core decision basis for early detection, risk classification and precise compensation of vibration abnormalities.

[0013] Optionally, the amplitude deviation degree of the vibration acceleration amplitude and a preset vibration amplitude range is calculated, the energy deviation degree of the energy proportion and a preset energy distribution threshold is calculated, and the amplitude deviation degree and the energy deviation degree are weighted and summed according to a preset weight group to obtain a vibration abnormality index: When the vibration acceleration amplitude is greater than a maximum value of the preset vibration amplitude range, an excess value of the vibration acceleration amplitude minus the maximum value is calculated, a target difference value of the maximum value and a minimum value of the preset vibration amplitude range is calculated, the excess value is divided by the target difference value to generate the amplitude deviation degree; When the vibration acceleration amplitude is less than the minimum value, an insufficient value of the minimum value minus the vibration acceleration amplitude is calculated, the insufficient value is divided by the target difference value to generate the amplitude deviation degree; A second difference value of the energy proportion and the preset energy distribution threshold is calculated, and an absolute value thereof is taken as a second absolute value, the second absolute value is divided by the preset energy distribution threshold to generate the energy deviation degree; The amplitude deviation degree is multiplied by an amplitude weight coefficient in the preset weight group, the energy deviation degree is multiplied by an energy weight coefficient in the preset weight group, and the two product results are added to obtain the vibration abnormality index.

[0014] By adopting the technical scheme, accurate, standardized and adjustable quantitative evaluation of the vibration abnormality degree is realized. The method can sensitively capture the two key fault signs of vibration intensity abnormality and energy distribution abnormality by calculating the amplitude deviation degree and the energy deviation degree respectively, and overcomes the limitation of single parameter criterion. The exceeding and insufficient conditions are processed respectively in the calculation process, and the target difference is used for standardization, so that the deviation degree obtained is a dimensionless relative value, which has good comparability and universality. Finally, the two-dimensional indexes are fused into a comprehensive vibration abnormality index by a pre-set weight coefficient, which not only reflects the importance difference of different indexes and enhances the pertinence and accuracy of state discrimination, but also provides a flexible adjustment means for system adaptation to different tools, materials and process parameters, and significantly improves the reliability and practicality of vibration state monitoring.

[0015] Optionally, the inputting the vibration frequency spectrum and the real-time temperature data into a pre-set prediction model to obtain a position offset prediction quantity comprises: Based on the vibration frequency spectrum, a pre-set number of peak frequencies arranged in descending order of amplitude and amplitude values corresponding to each of the peak frequencies are extracted; The pre-set number of peak frequencies, the pre-set number of amplitude values and the real-time temperature data are combined according to a pre-set rule to generate a multi-dimensional feature vector; The multi-dimensional feature vector is input into the prediction model to output the position offset prediction quantity, and the prediction model is trained by multi-dimensional feature vectors and position offset quantities in historical machining data.

[0016] By adopting the technical scheme, high-precision and forward-looking prediction of the position offset quantity under the coupling effect of tool thermal deformation and vibration is realized. The method innovatively multi-modal fuses vibration frequency spectrum features (peak frequency and amplitude) with real-time temperature data to construct a multi-dimensional feature vector that can comprehensively reflect the change of machining state, effectively capturing the influence mechanism of thermal-mechanical coupling effect; the prediction model trained by historical data is used for processing, so that the prediction result is not based on a single instantaneous parameter, but is derived from deep learning of a large number of historical process rules, thereby significantly improving the accuracy and robustness of the prediction. Finally, the scheme overcomes the limitations of traditional single signal source or lag compensation methods, and provides a key and reliable front-end input for realizing online and active compensation of machining precision.

[0017] Optionally, when the tool is in the wear overrun state and / or the breakage state, a size deviation value is calculated based on the wear feature quantitative value, and when the tool is in the vibration abnormality state, a coordinate offset quantity is calculated based on the position offset prediction quantity. When the tool is in the wear overrun state or the breakage state, the maximum width value in the wear feature quantization value is multiplied by a preset width deviation coefficient to obtain a first product, the maximum depth value in the wear feature quantization value is multiplied by a preset depth deviation coefficient to obtain a second product, and the first product and the second product are added to obtain the size deviation value; When the tool is in the vibration abnormal state, each axis coordinate correction value is calculated based on each axis offset component of the position offset prediction quantity in a preset coordinate and a corresponding preset compensation coefficient, and the coordinate offset quantity is formed by combining the axis coordinate correction values.

[0018] By adopting the technical solutions, for the wear overrun or breakage state, the size deviation value is calculated by weighting the maximum width and depth in the wear feature quantization value based on the physical properties, and the influence of tool wear on the machining size is directly associated; for the vibration abnormal state, the coordinate correction value is calculated according to each axis component of the position offset prediction quantity and the corresponding compensation coefficient, and the trajectory offset caused by vibration is accurately mapped. This state-specific and targeted deviation calculation method provides accurate data support for subsequent compensation instruction generation, avoids the compensation misalignment problem caused by the general deviation estimation in the traditional mode, effectively guarantees the accuracy of the compensation operation, and further improves the stability of the machining precision.

[0019] In a second aspect of the present application, a tool state real-time monitoring and compensation system is provided, specifically comprising: A data acquisition module is configured to acquire an optical image of a cutting edge of a tool, a time-domain waveform of an acoustic emission signal generated during machining, a vibration acceleration parameter of the tool, and real-time temperature data of a cutting region. A wear overrun judgment module is configured to measure a maximum width value and a maximum depth value of a wear zone in the optical image of the cutting edge, combine the maximum width value and the maximum depth value into a wear feature quantization value, and determine that the tool is in a wear overrun state if the wear feature quantization value is greater than a preset wear threshold value. A breakage state judgment module is configured to calculate an instantaneous mutation rate based on the time-domain waveform of the acoustic emission signal, and determine that the tool is in a breakage state if the instantaneous mutation rate continuously exceeds an upper limit of a preset acoustic emission fluctuation range for a set time length threshold. A vibration abnormality judgment module is configured to calculate a vibration frequency spectrum based on the vibration acceleration parameter, determine a vibration abnormality level based on the vibration frequency spectrum, and determine that the tool is in a vibration abnormal state if the vibration abnormality level exceeds a preset safety level. A data processing module is configured to input the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction quantity. The data processing module is further configured to calculate a size deviation value based on the wear characteristic quantization value when the tool is in the wear overrun state and / or the breakage state, and calculate a coordinate offset based on the position offset prediction value when the tool is in the vibration anomaly state. The compensation module is configured to generate a tool geometry compensation instruction according to the size deviation value, and / or generate a motion trajectory compensation instruction according to the coordinate offset, and perform a compensation operation based on the tool geometry compensation instruction or the motion trajectory compensation instruction.

[0020] In a third aspect of the present application, an electronic device is provided, which includes a processor, a memory, a user interface, and a network interface, the memory is configured to store instructions, the user interface and the network interface are configured to communicate with other devices, and the processor is configured to execute the instructions stored in the memory to enable the electronic device to perform the method according to any one of the preceding aspects.

[0021] In a fourth aspect of the present application, a computer-readable storage medium is provided, which stores instructions, when the instructions are executed, the method according to any one of the preceding aspects is performed.

[0022] In summary, the one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: By fusing multi-source sensing information (optical image, acoustic emission signal, vibration acceleration, and temperature data), all-round and high-sensitivity real-time monitoring of tool wear, breakage, vibration anomaly, and thermal deformation induced offset is achieved; innovatively, a multi-dimensional feature quantization, instantaneous mutation capture, frequency spectrum energy analysis, and machine learning prediction combination method is adopted, which significantly improves the accuracy of state determination and early warning capability; and on this basis, state-adaptive and accurate compensation is achieved, which fundamentally solves the lag and misjudgment problems of traditional manual inspection and single-parameter threshold method, and finally effectively guarantees the stability and continuity of the machining process, greatly improves the machining quality, qualification rate, and production efficiency of precision parts. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a system architecture schematic diagram of an embodiment of a tool state real-time monitoring and compensation method or a tool state real-time monitoring and compensation system according to the present application; Figure 2 is a flowchart of a tool state real-time monitoring and compensation method according to an embodiment of the present application; Figure 3 is a module schematic diagram of a tool state real-time monitoring and compensation system according to an embodiment of the present application; Figure 4 is a structural schematic diagram of an electronic device according to an embodiment of the present application.

[0024] Reference signs: 301, data acquisition module; 302, wear overrun judgment module; 303, breakage state judgment module; 304, vibration anomaly judgment module; 305, data processing module; 306, compensation module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION

[0025] In order for those skilled in the art to better understand the technical solutions in the specification, the technical solutions in the specification will be clearly and completely described below in combination with the drawings in the specification. Obviously, the described embodiments are only some of the embodiments of the present application, not all.

[0026] As shown in Figure 1 The system architecture 100 can include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a communication link medium between the terminal devices 101, 102, 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0027] A user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 101, 102, 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0028] The terminal devices 101, 102, 103 can be hardware or software. When the terminal devices 101, 102, 103 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, and desktop computers, etc. When the terminal devices 101, 102, 103 are software, they can be installed in the above-mentioned electronic devices. They can be implemented as multiple software or software modules (such as multiple software or software modules used to provide distributed services), or as a single software or software module. No specific limitation is made here.

[0029] The embodiment discloses a tool state real-time monitoring and compensation method, Figure 2 is a flowchart of a tool state real-time monitoring and compensation method disclosed by the embodiment of the application, as Figure 2 shown, the method comprises the following steps: S201, acquiring an edge optical image of a tool in a machining process, a time-domain waveform of an acoustic emission signal generated by machining, a vibration acceleration parameter of the tool and real-time temperature data of a cutting region.

[0030] Specifically, an industrial high-definition camera installed on a numerical control center tool tower or a spindle is used to collect the optical image of the tool edge in real time; an acoustic emission sensor installed on the tool clamping part or the spindle box is used to capture the stress wave signal generated in the machining process, which is converted into time-domain waveform data after amplification and filtering by a signal conditioner; a three-axis vibration acceleration sensor installed on the tool handle or the spindle is used to measure the vibration acceleration parameters of the tool in the X, Y and Z directions; an infrared temperature sensor installed near the tool tip or the cutting fluid spray pipe is used to non-contact measure the real-time temperature of the cutting region. All sensor data are synchronously collected by a data acquisition card and transmitted to an industrial computer for processing with a unified time scale, ensuring the spatiotemporal consistency of multi-source data and providing a comprehensive, synchronous and reliable raw data basis for subsequent state determination and compensation. Before implementation, all types of sensors need to be calibrated and the installation positions need to be optimized to ensure data accuracy.

[0031] S202, measuring the maximum width value and the maximum depth value of the wear zone in the edge optical image, combining the maximum width value and the maximum depth value into a wear feature quantization value, and determining that the tool is in a wear overrun state if the wear feature quantization value is greater than a preset wear threshold.

[0032] Specifically, first, the collected edge optical image is denoised by using a Gaussian filter (such as a 5x5 kernel), and then the image contrast is enhanced by using a limited contrast adaptive histogram equalization; then, the accurate contour boundary of the wear region is extracted by Canny edge detection (threshold value 50 / 150); the direction of the tool cutting edge is preset as the reference direction, the maximum distance of the contour point set in the X-axis projection along this direction is calculated as the maximum width value (VB_max) of the wear zone, and the maximum distance of the Y-axis projection along the vertical direction is calculated as the maximum depth value (KT_max); finally, the VB_max and the KT_max are weighted and summed according to a preset weight (such as α=0.7, β=0.3) to obtain a wear feature quantization value W, and when W exceeds a threshold value W_threshold preset according to the tool material and process parameters, it is determined that the tool is in a wear overrun state.

[0033] Optionally, the combining the maximum width value and the maximum depth value into a wear feature quantization value includes: performing noise reduction and contrast enhancement processing on the cutting edge optical image to extract a contour boundary of the target region; measuring a maximum distance along a preset direction of the contour boundary to obtain a maximum width value of the target region, and measuring a maximum distance along a direction perpendicular to the preset direction to obtain a maximum depth value of the target region; performing weighted summation on the maximum width value and the maximum depth value according to a preset wear weight to obtain the wear feature quantization value.

[0034] Specifically, first, a Gaussian filter (kernel size 5x5, sigma=1.5) is used to perform convolution operation on the original optical image to suppress noise; then a limited contrast adaptive histogram equalization algorithm is applied, with a clipping amplitude coefficient of 2.5 and a grid size of 8x8, to enhance the contrast between the wear region and the tool substrate; finally, a Canny edge detection operator (low threshold 50, high threshold 150, Sobel operator aperture 3x3) is used to extract the image edge, and a contour finding algorithm is used to search for the outermost connected domain, combined with a morphological closing operation (3x3 rectangular kernel) to repair the broken contour, to finally obtain a continuous and complete wear band contour boundary.

[0035] Further, first, based on the preprocessed cutting edge image, a Hough transform is used to detect the straight line segment of the cutting edge, and the direction angle θ of the straight line segment is determined as the preset direction (i.e. the width direction) of measurement. Then, a rotation matrix is constructed to rotate the entire contour point set by-θ angle, so that the preset direction is parallel to the X-axis of the rotated coordinate system. In this rotated coordinate system, the difference between the maximum and minimum values of all contour point X coordinates is directly calculated to obtain the accurate maximum width value (VB_max) in the direction parallel to the cutting edge; the difference between the maximum and minimum values of all contour point Y coordinates is directly calculated to obtain the maximum depth value (KT_max) in the direction perpendicular to the cutting edge. This method converts the direction of the physical feature (cutting edge) into a mathematical coordinate system through image processing technology, ensuring that the measurement result strictly corresponds to the actual wear geometry of the tool.

[0036] Further, the wear weight coefficients obtained from the process experiment calibration are pre-stored in the system parameters, including a width weight coefficient a (such as 0.6) and a depth weight coefficient β (such as 0.4), and a+β=1 is satisfied; the maximum width value VB_max measured is multiplied by the coefficient a through an arithmetic operation unit, while the maximum depth value KT_max is multiplied by the coefficient β, then the two product results are added to obtain a comprehensive wear feature quantization value W (the calculation formula is: W=α×VB_max+β×KT_max); finally, the quantization value is output to the state judgment link and compared with a preset threshold W_threshold to complete the quantitative evaluation of the tool wear state.

[0037] S203, calculate the instantaneous mutation rate according to the acoustic emission signal time domain waveform, if the instantaneous mutation rate continuously exceeds the upper limit of the preset acoustic emission fluctuation range for a time reaching a set time length threshold, it is determined that the tool is in a damaged state.

[0038] Specifically, the original acoustic emission signal time domain waveform collected is pre-processed, a high-pass digital filter with a cutoff frequency of 100 kHz is used to filter out low-frequency mechanical noise, and then an effective acoustic emission signal segment is extracted through a band-pass filter (passband range 100 kHz-1 MHz); then the filtered signal is digitized at a sampling rate of 2 MHz per second, and a target signal segment with a length of 1 ms is selected for analysis; in the signal segment, the absolute value of the amplitude difference |ΔA| of two consecutive sampling points (such as time interval Δt=0.5 μs) is calculated in time sequence, and the ratio of |ΔA| to Δt is taken as the instantaneous mutation rate R_i; the value of R_i is monitored in real time, and when it continuously exceeds the upper limit of the preset acoustic emission fluctuation range (such as 500 mV / μs) for a duration reaching a set time length threshold (such as more than 10 μs), it is immediately determined that the tool is in a damaged state.

[0039] Optionally, the instantaneous mutation rate calculated according to the acoustic emission signal time domain waveform comprises: filtering and denoising the acoustic emission signal time domain waveform to extract a target signal segment; selecting a first sampling point and a second sampling point in the target signal segment, calculating a first difference between the amplitude of the second sampling point and the amplitude of the first sampling point and taking the absolute value as a first absolute value, calculating the sampling time interval of the first sampling point and the second sampling point, and taking the ratio of the first absolute value to the sampling time interval as the instantaneous mutation rate, the first sampling point and the second sampling point being any two consecutive time points in the target signal segment.

[0040] Specifically, first, a fourth-order Butterworth high-pass filter (with a cutoff frequency set to 100 kHz) is used to filter out low-frequency noise interference introduced by machine tool vibration and the like; then a band-stop filter with a passband frequency of 100 kHz to 1 MHz is used to suppress electromagnetic interference at specific frequencies; finally, a moving average filter (with a window width of 5 sampling points) is used to further smooth high-frequency noise. The processed signal is used to identify valid signal segments by setting an amplitude threshold (e.g., 3 times the root mean square value of the baseline noise), and the continuous signal segments with an amplitude exceeding the threshold are extracted as target signal segments for subsequent analysis.

[0041] Further, within the extracted target signal segment, in the time sequence of signal acquisition, the current sampling point is marked as the second sampling point (amplitude A2), and the previous adjacent sampling point is marked as the first sampling point (amplitude A1); the absolute difference (|ΔA|=|A2-A1|) of the amplitudes of the two consecutive sampling points is calculated, and the fixed sampling time interval Δt (e.g., 0.5 μs corresponding to a 2 MHz sampling rate) set by the data acquisition card is obtained; finally, the arithmetic unit calculates the instantaneous mutation rate R=|ΔA| / Δt, and the calculation result is output as a quantitative index representing the instantaneous change characteristics of the signal to the subsequent judgment logic.

[0042] S204, a vibration frequency spectrum is calculated according to the vibration acceleration parameter, a vibration abnormality level is determined through the vibration frequency spectrum, and if the vibration abnormality level exceeds a preset safety level, the tool is determined to be in a vibration abnormal state.

[0043] Specifically, first, the collected three-axis vibration acceleration signals are preprocessed, the time domain signals are windowed using a Hanning window function, and a 4096-point fast Fourier transform (FFT) is performed to obtain a vibration frequency spectrum; all peak frequencies with an amplitude exceeding a preset threshold (e.g., 0.5g) are identified within the frequency range of 0-5000 Hz, and the top 3 peak frequencies with the largest amplitudes are selected as the dominant peak frequencies; the ratio of the sum of the amplitudes of these dominant peak frequencies to the total energy of the spectrum (the sum of the amplitudes of all frequency points) is calculated to obtain the energy proportion; then the deviation of the amplitudes of the dominant peak frequencies from a preset safe amplitude range (e.g., 0.2g-0.8g) and the deviation of the energy proportion from a preset energy threshold (e.g., 30%) are calculated; finally, the amplitude deviation and energy deviation of each dominant peak frequency are weighted and summed according to weights of 0.6 and 0.4 to obtain a vibration abnormality index, and when the index exceeds the threshold range corresponding to the preset safety level, the tool is immediately determined to be in a vibration abnormal state.

[0044] Optionally, the vibration frequency spectrum is calculated according to the vibration acceleration parameter, the vibration abnormality level is determined through the vibration frequency spectrum, and if the vibration abnormality level exceeds a preset safety level, the tool is determined to be in a vibration abnormal state, which includes: performing Fourier transform on the vibration acceleration parameter to obtain a vibration frequency spectrum, identifying a dominant peak frequency in the vibration frequency spectrum, the dominant peak frequency being a frequency with an amplitude exceeding a preset amplitude threshold, extracting a vibration acceleration amplitude corresponding to the dominant peak frequency from the vibration frequency spectrum, and calculating an energy proportion of vibration energy of the dominant peak frequency in total energy of the vibration frequency spectrum; calculating an amplitude deviation degree of the vibration acceleration amplitude from a preset vibration amplitude range, calculating an energy deviation degree of the energy proportion from a preset energy distribution threshold, and weighting and summing the amplitude deviation degree and the energy deviation degree according to a preset weight to obtain a vibration anomaly index; matching the vibration anomaly index with a preset vibration anomaly level division interval to obtain a vibration anomaly level, and determining that the tool is in the vibration anomaly state when the vibration anomaly level is higher than the preset safety level.

[0045] Specifically, first, a Hanning window function is applied to the collected vibration acceleration time domain signal, and then a 4096-point fast Fourier transform (FFT) is performed to obtain a vibration frequency spectrum in the range of 0-5000 Hz; in the frequency spectrum, all peak frequencies with an amplitude exceeding a preset threshold (0.5g) are identified, and the top 3 are selected as dominant peak frequencies in descending order of amplitude value; the vibration acceleration amplitudes corresponding to these dominant peak frequencies are extracted respectively, the ratio of the vibration energy (amplitude square sum) in a ±5 Hz frequency band at each peak frequency to the total vibration energy of the entire frequency spectrum is calculated, and the energy proportion of each dominant peak frequency is obtained for subsequent vibration state evaluation.

[0046] Further, first, it is determined whether the vibration acceleration amplitude A of each dominant peak frequency is within a preset vibration amplitude range [0.2g, 0.8g], if A>0.8g, the amplitude deviation degree D_a=(A-0.8) / (0.8-0.2) is calculated, if A<0.2g, D_a=(0.2-A) / (0.8-0.2) is calculated; at the same time, the energy deviation degree D_e=|E-30%| / 30% of the energy proportion E from the preset energy distribution threshold 30% is calculated; finally, the amplitude deviation degree D_a is multiplied by a weight coefficient 0.6, the energy deviation degree D_e is multiplied by a weight coefficient 0.4, and the sum is obtained to obtain a vibration anomaly index V_i=0.6×D_a+0.4×D_e, thereby completing the quantitative evaluation of the vibration state.

[0047] Further, the vibration abnormality level division interval is set in the system in advance, wherein 0≤V_i<0.5 is a safety level (level 1), 0.5≤V_i<1.0 is an attention level (level 2), 1.0≤V_i<1.5 is a warning level (level 3), and V_i≥1.5 is a danger level (level 4), and the preset safety level is set as level 2; the vibration abnormality index V_i calculated in real time is matched with the intervals through a comparator, and when the matching result shows that the current vibration abnormality level reaches level 3 or level 4, it is determined that the preset safety level is exceeded, and the tool is immediately determined to be in a vibration abnormality state.

[0048] Optionally, the amplitude deviation degree of the vibration acceleration amplitude from the preset vibration amplitude range and the energy deviation degree of the energy proportion from the preset energy distribution threshold are calculated, and the amplitude deviation degree and the energy deviation degree are weighted and summed according to a preset weight group to obtain a vibration abnormality index: When the vibration acceleration amplitude is greater than the maximum value of the preset vibration amplitude range, an excess value of the vibration acceleration amplitude minus the maximum value is calculated, a target difference value of the maximum value and the minimum value of the preset vibration amplitude range is calculated, and the excess value is divided by the target difference value to generate the amplitude deviation degree; When the vibration acceleration amplitude is less than the minimum value, an insufficient value of the minimum value minus the vibration acceleration amplitude is calculated, and the insufficient value is divided by the target difference value to generate the amplitude deviation degree; A second difference value of the energy proportion and the preset energy distribution threshold is calculated, and an absolute value thereof is taken as a second absolute value, the second absolute value is divided by the preset energy distribution threshold to generate the energy deviation degree; The amplitude deviation degree is multiplied by an amplitude weight coefficient in the preset weight group, the energy deviation degree is multiplied by an energy weight coefficient in the preset weight group, and the two product results are added to obtain the vibration abnormality index.

[0049] Specifically, first, it is judged whether the vibration acceleration amplitude A corresponding to the dominant peak frequency is in the preset vibration amplitude range [0.2g, 0.8g]: if A is greater than the maximum value 0.8g of the range, the excess value is A-0.8g, the target difference is the difference between the maximum value and the minimum value 0.8g-0.2g=0.6g, and the amplitude deviation degree D_a is the excess value divided by the target difference, that is, D_a=(A-0.8) / 0.6; if A is less than the minimum value 0.2g of the range, the deficiency value is 0.2g-A, and the amplitude deviation degree D_a is the deficiency value divided by the target difference, that is, D_a=(0.2-A) / 0.6. At the same time, the absolute difference between the energy proportion E of the dominant peak frequency and the preset energy distribution threshold 30% is calculated, and the energy deviation degree D_e is obtained by dividing the absolute difference by 30%, that is, D_e=|E-30%| / 30%. Finally, the amplitude deviation degree D_a is multiplied by 0.6 and the energy deviation degree D_e is multiplied by 0.4 according to the preset weight, and the sum of the two is the vibration anomaly index V_i=0.6×D_a+0.4×D_e, which completes the quantitative calculation of the vibration anomaly index.

[0050] Further, the vibration acceleration amplitude A corresponding to the dominant peak frequency is first extracted from the vibration spectrum, the minimum value of the preset vibration amplitude range is 0.2g, and the maximum value is 0.8g. The target difference is calculated as the difference between the maximum value and the minimum value, that is, 0.8g-0.2g=0.6g. When it is detected that A is less than the minimum value 0.2g, the deficiency value is calculated as the difference between the minimum value and the current amplitude, that is, the deficiency value=0.2g-A. Then, the deficiency value is divided by the target difference 0.6g to obtain the amplitude deviation degree, and the calculation formula is: amplitude deviation degree D_a=(0.2g-A) / 0.6g, which completes the quantitative calculation of the amplitude deviation when the vibration acceleration amplitude is less than the minimum value.

[0051] Further, the vibration energy proportion E of the dominant peak frequency in the total energy is first obtained from the vibration spectrum analysis, which is denoted as energy proportion E. The preset energy distribution threshold is 30% (i.e. 0.3). The second difference between the energy proportion E and the preset energy distribution threshold is calculated, that is, the second difference=E-30%; the absolute value of the second difference is taken to obtain the second absolute value=|E-30%|; then the second absolute value is divided by the preset energy distribution threshold 30%, and the quantitative calculation of the energy deviation degree is completed through the formula "energy deviation degree D_e=|E-30%| / 30%", wherein the value of E is substituted into the operation in the form of percentage.

[0052] Further, the preset weight group has an amplitude weight coefficient of 0.6 and an energy weight coefficient of 0.4; the amplitude deviation degree D_a calculated in the foregoing is multiplied by the amplitude weight coefficient 0.6 to obtain a first product result of 0.6*D_a; the energy deviation degree D_e is multiplied by the energy weight coefficient 0.4 to obtain a second product result of 0.4*D_e; finally, the two product results are added to complete the calculation of the vibration anomaly index through the formula "V_i=0.6*D_a+0.4*D_e", wherein D_a and D_e are the deviation degree values quantified in the foregoing steps, and the specific result of the vibration anomaly index can be obtained by directly substituting the formula.

[0053] S205, input the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction value.

[0054] Specifically, first, feature parameters are extracted from the vibration frequency spectrum, including three dominant peak frequencies and corresponding vibration acceleration amplitudes; real-time temperature data (with a sampling frequency of 1 Hz) are analyzed in a 5-second period to calculate a temperature change rate (a difference between a current temperature and a temperature 5 seconds ago divided by 5), a temperature fluctuation range (a difference between a maximum value and a minimum value in the 5 seconds), and a cumulative temperature change value (a sum of absolute values of temperature changes in the 5 seconds). The preset prediction model adopts a three-layer fully connected neural network architecture. An input layer receives a standardized feature vector spliced from the above nine features (three frequencies + three amplitudes + three temperature features). A hidden layer contains 16 neurons and adopts a ReLU activation function to learn the non-linear relationship between the features. An output layer contains one neuron and adopts a linear activation function to directly output a position offset prediction value (unit: μm). The model has been trained with historical processing data to have a mean square error (MSE) less than 0.02 μm².

[0055] Optionally, the inputting the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction value includes: extracting a preset number of peak frequencies arranged in descending order of amplitude and amplitudes corresponding to the peak frequencies based on the vibration frequency spectrum; combining the preset number of peak frequencies, the preset number of amplitudes, and the real-time temperature data according to a preset rule to generate a multi-dimensional feature vector; inputting the multi-dimensional feature vector into the prediction model to output the position offset prediction value, the prediction model being trained with multi-dimensional feature vectors and position offset values in historical processing data.

[0056] Specifically, the vibration spectrum obtained by Fourier transform (the frequency range is set to 0-5000 Hz, and the frequency interval is 0.5 Hz) is obtained first, the number of peak frequencies to be extracted is set to 3, and the amplitude filtering threshold is set to 0.3 g (for removing noise signals). All frequency points in the spectrum are traversed, for each frequency point, if its amplitude value exceeds 0.3 g and is greater than the amplitude values of the left and right adjacent frequency points (determined as an effective peak value), the frequency and the corresponding amplitude are recorded; all effective peak values are arranged in descending order according to the amplitude values, and the first 3 frequencies in the sorted order are taken as the target peak frequencies, and the corresponding amplitude values are recorded, thereby completing the extraction of the preset number of peak frequencies and the corresponding amplitudes.

[0057] Further, the preset number is set to 3, the real-time temperature data needs to extract 3 features (temperature change rate within 5 seconds, temperature fluctuation range, cumulative temperature change value), and the preset combination rule is to splice in the order of “peak frequency → amplitude value → temperature feature”. The 3 peak frequencies arranged in descending order of amplitude are taken as the 1st to 3rd dimensions (unit: Hz) of the feature vector in turn; then the amplitude values corresponding to the 3 peak frequencies are taken as the 4th to 6th dimensions (unit: g) in the same order; finally, the 3 features of the real-time temperature data are taken as the 7th to 9th dimensions in the order of “temperature change rate (℃ / s) → temperature fluctuation range (℃) → cumulative temperature change value (℃)”, thereby generating a 9-dimensional multi-dimensional feature vector, and the numerical values in each dimension of the vector are kept to two decimal places for uniform precision.

[0058] Further, the prediction model adopts a three-layer fully connected neural network architecture: the input layer contains 9 neurons, responsible for receiving the 9-dimensional feature vector (including 3 peak frequencies, 3 corresponding amplitude values and 3 temperature features) after standardization, and completely transmitting the original features to the hidden layer in numerical form as the initial input for model calculation; the hidden layer contains 16 neurons, adopts a linear rectifier unit as the activation function, performs nonlinear conversion on the input features through the weight matrix operation between neurons, extracts the coupling correlation information between vibration and temperature features (such as the synergistic effect of high-frequency vibration and temperature sudden rise), and generates high-order features; the output layer contains 1 neuron, adopts a linear activation function, maps the high-order features output by the hidden layer to specific numerical values, and directly outputs the position offset prediction value (unit: μm). In the training stage, the mean squared error (MSE) is used as the loss function, and its calculation formula is “ The formula is denoted as "", where n is the number of samples, ŷ_i is the predicted position offset by the model, and y_i is the actual position offset measured by a laser interferometer. This function quantifies the model prediction error by calculating the average of the squares of the differences between the predicted and actual values. During training, based on 500 sets of historical data (divided into training and validation sets in an 8:2 ratio), the Adam optimizer (learning rate 0.001, batch size 32) is used to iteratively adjust the weights of each layer to minimize the MSE until the validation set MSE stabilizes below 0.02μm². During real-time operation, the new feature vector is standardized and input into the model, then passed through the input layer, the hidden layer for feature transformation, and the output layer for mapping, finally outputting the predicted position offset (retaining three decimal places).

[0059] S206. When the tool is in the wear-over-limit state and / or the breakage state, calculate the dimensional deviation value based on the wear characteristic quantification value; when the tool is in the vibration-abnormal state, calculate the coordinate offset based on the position offset prediction value.

[0060] Specifically, if the condition is determined to be wear exceeding limits and / or damage, the wear feature quantification value (including the maximum width value W and the maximum depth value D) is called. For example, the preset width deviation coefficient is 0.8 and the depth deviation coefficient is 1.2. The size deviation value (unit: μm) is calculated by the formula "size deviation value = W × 0.8 + D × 1.2". If the condition is determined to be vibration abnormality, the X-axis offset component ΔX, Y-axis offset component ΔY, and Z-axis offset component ΔZ in the position offset prediction are obtained. For example, the preset X-axis compensation coefficient is 0.9, Y-axis compensation coefficient is 0.95, and Z-axis compensation coefficient is 1.0. The coordinate correction value of each axis is calculated respectively (X correction = ΔX × 0.9, Y correction = ΔY × 0.95, Z correction = ΔZ × 1.0). The correction values ​​of the three axes are combined in the format "[X correction, Y correction, Z correction]" to form the coordinate offset (unit: μm).

[0061] Optionally, when the tool is in the wear-exceeding state and / or the breakage state, calculating the dimensional deviation value based on the wear characteristic quantification value, and when the tool is in an abnormal vibration state, calculating the coordinate offset based on the position offset prediction value includes: When the tool is in the wear-over-limit state or the breakage state, the maximum width value in the wear feature quantification value is multiplied by a preset width deviation coefficient to obtain a first product, the maximum depth value in the wear feature quantification value is multiplied by a preset depth deviation coefficient to obtain a second product, and the first product and the second product are added to obtain the size deviation value; When the tool is in the vibration abnormal state, based on each axis offset component in the preset coordinate of the position offset prediction value and the corresponding preset compensation coefficient, each axis coordinate correction value is calculated, and the coordinate offset value is formed by combining the each axis coordinate correction value.

[0062] Further, when the system determines that the tool is in the wear overrun state or the broken state, the maximum width value W (unit: μm, accurate to one decimal place) and the maximum depth value D (unit: μm, accurate to one decimal place) are extracted from the wear feature quantitative value; the preset width deviation coefficient is 0.7, and the depth deviation coefficient is 1.3, wherein the width deviation coefficient corresponds to the influence weight of tool width wear on the machining size, and the depth deviation coefficient corresponds to the influence weight of depth wear; a first product is calculated as the product of the maximum width value and the width deviation coefficient, that is, the first product W*0.7, and a second product is calculated as the product of the maximum depth value and the depth deviation coefficient, that is, the second product D*1.3; finally, the first product and the second product are added to obtain a size deviation value (unit: μm, two decimal places are reserved), and the formula is: size deviation value = W*0.7 + D*1.3.

[0063] Further, when it is determined that the tool is in the vibration abnormal state, each axis offset component in the preset three-dimensional rectangular coordinate system (X axis, Y axis, Z axis) is extracted from the position offset prediction value, and is respectively denoted as ΔX (X axis offset value, unit: μm, accurate to two decimal places), ΔY (Y axis offset value, unit: μm, accurate to two decimal places), and ΔZ (Z axis offset value, unit: μm, accurate to two decimal places); preset compensation coefficients corresponding to each axis are provided, wherein the X axis compensation coefficient is 0.85 (representing the influence weight of X axis offset on machining precision), the Y axis compensation coefficient is 0.9 (representing the influence weight of Y axis offset), and the Z axis compensation coefficient is 1.0 (representing the influence weight of Z axis offset); each axis coordinate correction value is calculated: X axis correction value = ΔX*0.85, Y axis correction value = ΔY*0.9, and Z axis correction value = ΔZ*1.0 (all results are accurate to two decimal places, and the unit is μm); finally, the coordinate correction values of the three axes are combined in the order of [X axis correction value, Y axis correction value, Z axis correction value] to form a coordinate offset value containing three-axis offset information, which is used for subsequent generation of motion trajectory compensation instructions.

[0064] S207, generating tool geometry compensation instructions according to the size deviation value, and / or generating motion trajectory compensation instructions according to the coordinate offset value, and completing compensation operation based on the tool geometry compensation instructions or the motion trajectory compensation instructions.

[0065] Specifically, first determine the compensation type to be executed: if there is a size deviation value (such as 7.5 μm calculated), generate a tool geometry compensation instruction, and use the general format of the numerical control system "G41 D03 X7.5" (where G41 is a tool radius compensation instruction, D03 corresponds to a preset No. 3 compensation register, which pre-stores the association mapping of the current tool number and the size deviation value, and X7.5 indicates that a reverse compensation amount of -7.5 μm is applied in the X-axis direction to offset the size deviation); if there is a coordinate offset "[4.2 μm, 3.6 μm, 2.1 μm]", generate a motion trajectory compensation instruction, and the format is "G52 X4.2 Y3.6 Z2.1" (G52 is a local coordinate system offset instruction, and the numerical values after X, Y and Z correspond to the coordinate offset amounts of each axis that need to be positively compensated, which directly corrects the tool motion trajectory); if both types of deviations exist, generate a combined instruction in the order of "geometric compensation first, then trajectory compensation". The generated instruction is sent to the numerical control center controller through the Ethernet interface, and the controller updates the tool parameter register and the coordinate system offset register within the next interpolation period (≤1 ms) after parsing the instruction, and adjusts the tool edge position in real time. After compensation is completed, the system immediately collects the optical image of the machined surface, detects the actual size deviation through the image measurement tool, and determines that the compensation is qualified when the deviation is ≤±0.3 μm, otherwise triggers the secondary compensation process.

[0066] The embodiment also discloses a tool state real-time monitoring and compensation system, Figure 3 is a module schematic diagram of the tool state real-time monitoring and compensation system disclosed by the embodiment of the application, as Figure 3 shown, the system comprises: a data acquisition module 301 configured to acquire an edge optical image of a tool in a machining process, a time-domain waveform of an acoustic emission signal generated in machining, a vibration acceleration parameter of the tool, and real-time temperature data of a cutting region; a wear overrun judgment module 302 configured to measure a maximum width value and a maximum depth value of a wear zone in the edge optical image, combine the maximum width value and the maximum depth value into a wear feature quantization value, and determine that the tool is in a wear overrun state if the wear feature quantization value is greater than a preset wear threshold value; a breakage state judgment module 303 configured to calculate an instantaneous mutation rate according to the time-domain waveform of the acoustic emission signal, and determine that the tool is in a breakage state if the instantaneous mutation rate continuously exceeds an upper limit of a preset acoustic emission fluctuation range for a set time length threshold value; a vibration anomaly judgment module 304 configured to calculate a vibration frequency spectrum according to the vibration acceleration parameter, determine a vibration anomaly level through the vibration frequency spectrum, and determine that the tool is in a vibration anomaly state if the vibration anomaly level exceeds a preset safety level; The data processing module 305 is configured to input the vibration spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction value. The data processing module 305 is further configured to calculate a size deviation value based on the wear characteristic quantitative value when the tool is in the wear overrun state and / or the breakage state, and calculate a coordinate offset value based on the position offset prediction value when the tool is in the vibration abnormal state. The compensation module 306 is configured to generate a tool geometry compensation instruction according to the size deviation value, and / or generate a motion trajectory compensation instruction according to the coordinate offset value, and complete a compensation operation based on the tool geometry compensation instruction or the motion trajectory compensation instruction.

[0067] Optionally, the wear overrun judgment module 302 is specifically configured to: perform noise reduction and contrast enhancement processing on the cutting edge optical image to extract a contour boundary of a target region; measure a maximum distance along a preset direction of the contour boundary to obtain a maximum width value of the target region, and measure a maximum distance along a direction perpendicular to the preset direction to obtain a maximum depth value of the target region; weight and sum the maximum width value and the maximum depth value according to a preset wear weight to obtain the wear characteristic quantitative value.

[0068] Optionally, the breakage state judgment module 303 is specifically configured to: perform filtering and noise reduction on the acoustic emission signal time domain waveform to extract a target signal segment; select a first sampling point and a second sampling point in the target signal segment, calculate a first difference value between an amplitude of the second sampling point and an amplitude of the first sampling point and take an absolute value of the first difference value as a first absolute value, calculate a sampling time interval of the first sampling point and the second sampling point, and take a ratio of the first absolute value to the sampling time interval as the instantaneous mutation rate, the first sampling point and the second sampling point being sampling points of any two continuous time points in the target signal segment.

[0069] Optionally, the vibration abnormality judgment module 304 is specifically configured to: perform Fourier transform on the vibration acceleration parameter to obtain a vibration spectrum, identify a dominant peak frequency in the vibration spectrum, the dominant peak frequency being a frequency with an amplitude exceeding a preset amplitude threshold, extract a vibration acceleration amplitude corresponding to the dominant peak frequency from the vibration spectrum, and calculate an energy proportion of vibration energy of the dominant peak frequency in total energy of the vibration spectrum; calculate a magnitude deviation degree of the vibration acceleration magnitude and a preset vibration magnitude range, calculate an energy deviation degree of the energy proportion and a preset energy distribution threshold, and weight and sum the magnitude deviation degree and the energy deviation degree according to a preset weight to obtain a vibration anomaly index; match the vibration anomaly index with a preset vibration anomaly level division interval to obtain a vibration anomaly level, and determine that the tool is in the vibration anomaly state when the vibration anomaly level is higher than the preset safety level.

[0070] Optionally, the vibration anomaly judgment module 304 is further configured to: When the vibration acceleration magnitude is greater than a maximum value of the preset vibration magnitude range, calculate an excess value of the vibration acceleration magnitude minus the maximum value, calculate a target difference value of the maximum value and a minimum value of the preset vibration magnitude range, divide the excess value by the target difference value, and generate the magnitude deviation degree; When the vibration acceleration magnitude is less than the minimum value, calculate an insufficient value of the minimum value minus the vibration acceleration magnitude, divide the insufficient value by the target difference value, and generate the magnitude deviation degree; Calculate a second difference value of the energy proportion and the preset energy distribution threshold, take an absolute value of the second difference value as a second absolute value, divide the second absolute value by the preset energy distribution threshold, and generate the energy deviation degree; multiply the magnitude deviation degree by a magnitude weight coefficient in the preset weight group, multiply the energy deviation degree by an energy weight coefficient in the preset weight group, and add the two product results to obtain the vibration anomaly index.

[0071] Optionally, the data processing module 305 is specifically configured to: extract a preset number of peak frequencies arranged in descending order of amplitude and amplitude values corresponding to each peak frequency based on the vibration frequency spectrum; combine the preset number of peak frequencies, the preset number of amplitude values, and the real-time temperature data according to a preset rule to generate a multi-dimensional feature vector; input the multi-dimensional feature vector into the prediction model to output the position offset prediction, and the prediction model is trained based on multi-dimensional feature vectors and position offset amounts in historical processing data.

[0072] Optionally, the data processing module 305 is further configured to: When the tool is in the wear-out over-limit state or the breakage state, the maximum width value in the wear-out feature quantization value is multiplied by a preset width deviation coefficient to obtain a first product, the maximum depth value in the wear-out feature quantization value is multiplied by a preset depth deviation coefficient to obtain a second product, and the first product and the second product are added to obtain the size deviation value. When the tool is in the vibration abnormality state, each axis coordinate correction value is calculated based on each axis offset component in a preset coordinate in the position offset prediction value and a corresponding preset compensation coefficient, and the coordinate offset value is formed by combining the axis coordinate correction values.

[0073] It should be noted that the apparatus provided in the above embodiments is only used as an example for dividing the above functional modules to achieve the functions thereof, and in actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the above-described functions. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is described in the method embodiments, which will not be described here.

[0074] The embodiment further discloses an electronic device, which refers to Figure 4 The electronic device can include at least one processor 401, at least one communication bus 402, a user interface 403, a network interface 404, and at least one memory 405.

[0075] The communication bus 402 is used to realize the connection and communication between the components.

[0076] The user interface 403 can include a display screen (Display) and a camera (Camera), and the optional user interface 403 can further include a standard wired interface and a wireless interface.

[0077] The network interface 404 can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0078] The processor 401 can include one or more processing cores. The processor 401 connects various parts within the server through various interfaces and lines, performs various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 405, and calling data stored in the memory 405. Alternatively, the processor 401 can be implemented in at least one of a hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The processor 401 can integrate a combination of one or more of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes operating systems, user interfaces, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 401, but can be realized by a separate chip.

[0079] The memory 405 can include a random access memory (RAM) and a read-only memory (ROM). Alternatively, the memory 405 includes a non-transitory computer-readable storage medium. The memory 405 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 405 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 405 can alternatively be at least one storage device located away from the aforementioned processor 401. As shown in the figure, the memory 405 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an application program of a tool state real-time monitoring and compensation method. Figure 4 As shown in the figure, the memory 405 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an application program of a tool state real-time monitoring and compensation method.

[0080] In Figure 4In the electronic device shown, the user interface 403 is mainly used to provide an interface for the user to input, and obtain data input by the user; and the processor 401 can be used to call an application program stored in the memory 405 and storing a kind of tool state real-time monitoring and compensation method, which, when executed by one or more processors 401, causes the electronic device to perform the method of one or more of the above embodiments.

[0081] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited to the action sequence described, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0082] The above is only an exemplary embodiment of the present disclosure, and cannot limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Those skilled in the art will easily think of other embodiments of the present disclosure after considering the disclosure of the specification. The present application is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the art not disclosed in the present disclosure. The specification and examples are only considered as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method of real-time monitoring and compensation of a tool state, characterized by, The method is applied to a numerical control center, and the method comprises the following steps: Obtaining an optical image of a cutting edge of a tool in a machining process, a time-domain waveform of an acoustic emission signal generated in the machining, a vibration acceleration parameter of the tool, and real-time temperature data of a cutting region; Measuring a maximum width value and a maximum depth value of a wear zone in the optical image of the cutting edge, combining the maximum width value and the maximum depth value into a wear feature quantization value, and determining that the tool is in a wear overrun state if the wear feature quantization value is greater than a preset wear threshold value; Calculating an instantaneous mutation rate according to the time-domain waveform of the acoustic emission signal, and determining that the tool is in a damage state if the instantaneous mutation rate continuously exceeds an upper limit of a preset acoustic emission fluctuation range for a set time length threshold value; Calculating a vibration frequency spectrum according to the vibration acceleration parameter, determining a vibration abnormality level through the vibration frequency spectrum, and determining that the tool is in a vibration abnormality state if the vibration abnormality level exceeds a preset safety level; Inputting the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction value; When the tool is in the wear overrun state and / or the damage state, calculating a size deviation value based on the wear feature quantization value, and when the tool is in the vibration abnormality state, calculating a coordinate offset value based on the position offset prediction value; Generating a tool geometry compensation instruction according to the size deviation value and / or generating a motion trajectory compensation instruction according to the coordinate offset value, and completing a compensation operation based on the tool geometry compensation instruction or the motion trajectory compensation instruction.

2. The method of claim 1, wherein, The measurement of the maximum width value and the maximum depth value of the wear zone in the optical image of the cutting edge, and the combination of the maximum width value and the maximum depth value into the wear feature quantization value comprise the following steps: Performing noise reduction and contrast enhancement processing on the optical image of the cutting edge to extract the contour boundary of a target region; Measuring a maximum distance along a preset direction of the contour boundary to obtain the maximum width value of the target region, and measuring a maximum distance along a direction perpendicular to the preset direction to obtain the maximum depth value of the target region; Weighted sum of the maximum width value and the maximum depth value according to a preset wear weight to obtain the wear feature quantization value.

3. The method of claim 1, wherein, The calculation of the instantaneous mutation rate according to the time-domain waveform of the acoustic emission signal comprises the following steps: Filtering and denoising the time-domain waveform of the acoustic emission signal to extract a target signal segment; Selecting a first sampling point and a second sampling point in the target signal segment, calculating a first difference value between the amplitude of the second sampling point and the amplitude of the first sampling point and taking the absolute value as a first absolute value, and calculating the sampling time interval of the first sampling point and the second sampling point, and taking the ratio of the first absolute value to the sampling time interval as the instantaneous mutation rate, wherein the first sampling point and the second sampling point are sampling points of any two consecutive time points in the target signal segment.

4. The method of claim 1, wherein, The calculation of the vibration frequency spectrum according to the vibration acceleration parameter, the determination of the vibration abnormality level through the vibration frequency spectrum, and the determination that the tool is in the vibration abnormality state if the vibration abnormality level exceeds the preset safety level comprise the following steps: performing Fourier transform on the vibration acceleration parameter to obtain a vibration frequency spectrum, identifying a dominant peak frequency in the vibration frequency spectrum, the dominant peak frequency being a frequency with an amplitude exceeding a preset amplitude threshold, extracting a vibration acceleration amplitude corresponding to the dominant peak frequency from the vibration frequency spectrum, and calculating an energy proportion of vibration energy of the dominant peak frequency in total energy of the vibration frequency spectrum; calculating an amplitude deviation degree of the vibration acceleration amplitude from a preset vibration amplitude range, calculating an energy deviation degree of the energy proportion from a preset energy distribution threshold, and performing weighted summation on the amplitude deviation degree and the energy deviation degree according to a preset weight to obtain a vibration anomaly index; matching the vibration anomaly index with a preset vibration anomaly level division interval to obtain a vibration anomaly level, and determining that the tool is in the vibration anomaly state when the vibration anomaly level is higher than the preset safety level.

5. The method of claim 4, wherein, The calculation of the amplitude deviation degree of the vibration acceleration amplitude from the preset vibration amplitude range, the calculation of the energy deviation degree of the energy proportion from the preset energy distribution threshold, and the weighted summation of the amplitude deviation degree and the energy deviation degree according to the preset weight group to obtain the vibration anomaly index include: When the vibration acceleration amplitude is greater than a maximum value of the preset vibration amplitude range, an excess value of the vibration acceleration amplitude minus the maximum value is calculated, a target difference value of the maximum value and a minimum value of the preset vibration amplitude range is calculated, and the excess value is divided by the target difference value to generate the amplitude deviation degree; When the vibration acceleration amplitude is less than the minimum value, an insufficient value of the minimum value minus the vibration acceleration amplitude is calculated, and the insufficient value is divided by the target difference value to generate the amplitude deviation degree; A second difference value of the energy proportion and the preset energy distribution threshold is calculated, and an absolute value of the second difference value is taken as a second absolute value, the second absolute value is divided by the preset energy distribution threshold to generate the energy deviation degree; The amplitude deviation degree is multiplied by an amplitude weight coefficient in the preset weight group, the energy deviation degree is multiplied by an energy weight coefficient in the preset weight group, and the two product results are added to obtain the vibration anomaly index.

6. The method of claim 1, wherein, The input of the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction quantity includes: Based on the vibration frequency spectrum, a preset number of peak frequencies arranged in descending order of amplitude and amplitude values corresponding to each of the peak frequencies are extracted; The preset number of peak frequencies, the preset number of amplitude values and the real-time temperature data are combined according to a preset rule to generate a multi-dimensional feature vector; The multi-dimensional feature vector is input into the prediction model, and the position offset prediction quantity is output, and the prediction model is trained based on multi-dimensional feature vectors and position offset quantities in historical processing data.

7. The method of claim 1, wherein, When the tool is in the wear overrun state and / or the damage state, a size deviation value is calculated based on the wear feature quantization value, and when the tool is in the vibration anomaly state, a coordinate offset quantity is calculated based on the position offset prediction quantity. When the tool is in the wear overrun state or the breakage state, the maximum width value in the wear feature quantization value is multiplied by a preset width deviation coefficient to obtain a first product, the maximum depth value in the wear feature quantization value is multiplied by a preset depth deviation coefficient to obtain a second product, and the first product and the second product are added to obtain the size deviation value; When the tool is in the vibration abnormal state, each axis coordinate correction value is calculated based on each axis offset component of the position offset prediction in a preset coordinate and a corresponding preset compensation coefficient, and the coordinate offset is formed by combining the axis coordinate correction values.

8. A tool condition monitoring and compensation system, characterized by Specifically comprising: A data acquisition module is configured to acquire an optical image of a cutting edge of a tool, a time-domain waveform of an acoustic emission signal generated during machining, a vibration acceleration parameter of the tool, and real-time temperature data of a cutting region. A wear overrun judgment module is configured to measure a maximum width value and a maximum depth value of a wear zone in the optical image of the cutting edge, combine the maximum width value and the maximum depth value into a wear feature quantization value, and determine that the tool is in a wear overrun state if the wear feature quantization value is greater than a preset wear threshold. A breakage state judgment module is configured to calculate an instantaneous mutation rate based on the time-domain waveform of the acoustic emission signal, and determine that the tool is in a breakage state if the instantaneous mutation rate continuously exceeds an upper limit of a preset acoustic emission fluctuation range for a set time length threshold. A vibration abnormality judgment module is configured to calculate a vibration frequency spectrum based on the vibration acceleration parameter, determine a vibration abnormality level based on the vibration frequency spectrum, and determine that the tool is in a vibration abnormal state if the vibration abnormality level exceeds a preset safety level. A data processing module is configured to input the vibration frequency spectrum and the real-time temperature data into a preset prediction model to obtain a position offset prediction. The data processing module is further configured to calculate a size deviation value based on the wear feature quantization value when the tool is in the wear overrun state and / or the breakage state, and calculate a coordinate offset based on the position offset prediction when the tool is in the vibration abnormal state. A compensation module is configured to generate a tool geometry compensation instruction based on the size deviation value, and / or generate a motion trajectory compensation instruction based on the coordinate offset, and complete a compensation operation based on the tool geometry compensation instruction or the motion trajectory compensation instruction.

9. An electronic device, comprising: The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is configured to store instructions. The user interface and the network interface are configured to communicate with other devices. The processor is configured to execute the instructions stored in the memory to enable the electronic device to perform the method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method of any one of claims 1-7.

Images