Machine tool machining shaft anomaly detection method and system based on dual-order excitation

Abstract

The invention provides a machine tool machining shaft anomaly detection method and system based on double-order excitation, and relates to the technical field of numerical control machine tools. The method comprises the steps that a key machining shaft of a numerical control machine tool is determined; constructing a shaft transmission anomaly detection model; a double-order reverse test sequence is injected into the key machining shaft through the controller, and a double-order pulse response sequence corresponding to the double-order reverse test sequence is recorded through the encoder; a pulse arrival time sequence is analyzed, and reverse response delay, static lag section length and speed transient characteristics are calculated; and inputting into a shaft transmission anomaly detection model for analysis, and outputting a reverse gap detection value of the key processing shaft. The technical problem that in the prior art, due to the fact that tiny anomalies of the numerical control machine tool are difficult to accurately detect, the machining anomaly detection precision is insufficient is solved, and the machining anomaly detection accuracy of the numerical control machine tool is improved through the double-order reverse test sequence and the shaft transmission anomaly detection model.

Description

Technical Field

[0001] This application relates to the field of CNC machine tool technology, and in particular to a method and system for detecting abnormalities in machine tool machining axes based on two-stage excitation. Background Technology

[0002] Currently, existing CNC machine tool machining anomaly detection mainly relies on passive sensing signals such as vibration, current, or sound during the machining process, which can provide monitoring of the machining process to a certain extent. However, the signal characteristics generated by early, minute anomalies are extremely weak and easily drowned out by strong interference signals such as main cutting forces and environmental noise, resulting in a low signal-to-noise ratio. This leads to insufficient sensitivity in detecting minute anomalies, a high false negative rate, and difficulty in achieving pre-emptive assurance of machining accuracy and predictive maintenance. Although some online monitoring technologies improve detection capabilities through algorithm optimization, they are still limited by the inherent shortcomings of passive sensing modes. They usually only issue alarms when anomalies develop into obvious faults, failing to provide accurate quantification and early warning in the early stages. The inability to provide early warnings for key performance parameters of CNC machine tools leads to problems such as degraded machining quality, increased scrap rates, and sudden downtime, further impacting production efficiency.

[0003] In summary, existing technologies suffer from the technical problem that the difficulty in accurately detecting minute anomalies in CNC machine tools leads to insufficient accuracy in detecting machining anomalies, which further affects the production efficiency of CNC machine tools. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for detecting machine tool machining axis anomalies based on two-stage excitation, in order to solve the technical problem in the prior art that the difficulty in accurately detecting minute anomalies of CNC machine tools leads to insufficient detection accuracy of machining anomalies, which further affects the production efficiency of CNC machine tools.

[0005] In view of the above problems, this application provides a method and system for detecting abnormalities in machine tool machining axes based on two-stage excitation.

[0006] In a first aspect, this application provides a machine tool machining axis anomaly detection method based on two-order excitation. This method is implemented through a machine tool machining axis anomaly detection system based on two-order excitation. The method includes: identifying the critical machining axis of a CNC machine tool; constructing a shaft transmission anomaly detection model for the critical machining axis, which is obtained through training on multiple historical abnormal transmission data samples of the critical machining axis; injecting a two-order reverse test sequence into the critical machining axis via a controller, and recording the corresponding two-order impulse response sequence via an encoder, wherein the two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, the first amplitude being smaller than the second amplitude; analyzing the pulse arrival time sequence of the two-order impulse response sequence, calculating the reverse response delay, hysteresis length, and speed transient characteristics; inputting the reverse response delay, hysteresis length, and speed transient characteristics into the shaft transmission anomaly detection model for analysis, and outputting the backlash detection value of the critical machining axis.

[0007] Optionally, a shaft transmission anomaly detection model for the key machining axis is constructed. The shaft transmission anomaly detection model is trained by identifying multiple historical abnormal transmission data samples of the key machining axis. The multiple historical abnormal transmission data samples include normal transmission data samples, abnormal transmission data samples corresponding to known backlash samples, and double-order impulse response sequence samples corresponding to the same double-order reverse test sequence. The reverse response delay feature vector, hysteresis length feature vector, and velocity transient feature vector of the double-order impulse response sequence samples are extracted as a feature vector set for regression training, and the trained shaft transmission anomaly detection model is output.

[0008] Optionally, a two-stage reverse test sequence is injected into the critical machining axis via a controller connected to a SEPIC chopper circuit; the duty cycle of the SEPIC chopper circuit is adjusted with the first amplitude to obtain a pulse test sequence for outputting the first amplitude, and the duty cycle of the SEPIC chopper circuit is adjusted with the second amplitude to obtain a pulse test sequence for outputting the second amplitude; the controller injects the pulse test sequence of the first amplitude or the pulse test sequence of the second amplitude into the critical machining axis.

[0009] Optionally, the first pulse response sequence corresponding to the pulse test sequence of the first amplitude is recorded by an encoder, and anomaly identification is performed on the first pulse response sequence to obtain an anomaly identification result; if the anomaly identification result returns empty, the controller is not activated to inject the pulse test sequence of the second amplitude into the key machining axis; if the anomaly identification result returns not empty, the controller is activated to inject the pulse test sequence of the second amplitude into the key machining axis, and the second pulse response sequence corresponding to the pulse test sequence of the second amplitude is recorded by an encoder.

[0010] Optionally, the SEPIC chopper circuit is connected to a conduction mode controller; historical transmission sample data of the key machining axis is identified, and the temporal continuity of the historical transmission sample data is analyzed. The conduction mode controller determines whether the key machining axis is in continuous transmission mode based on the temporal continuity. If it is in continuous transmission mode, the SEPIC chopper circuit is set to continuous conduction mode, and the duty cycle of the SEPIC chopper circuit is adjusted with a first amplitude or a second amplitude to construct a two-order reverse test sequence.

[0011] Optionally, a pulse test sequence template is determined, the pulse test sequence template including a plurality of pulse units, wherein each pulse unit includes a forward injection segment, a forward injection observation segment, a reverse injection segment and a recovery stabilization segment; the first amplitude and the second amplitude are converted into a first duty cycle and a second duty cycle; the controller outputs a pulse test signal based on the pulse test sequence template by controlling the duty cycle of the first amplitude or the second amplitude, thereby generating a two-order reverse test sequence.

[0012] Optionally, the first amplitude is smaller than the second amplitude, and the ratio of the first amplitude to the second amplitude is in the range of [0.2, 0.5].

[0013] Optionally, the key machining axis is calibrated to obtain multiple feature positions; a double-order reverse test sequence is injected into the key machining axis at the multiple feature positions by a controller to obtain multiple double-order impulse response sequences corresponding to the multiple feature positions; the pulse arrival time sequence of the multiple double-order impulse response sequences is analyzed to update the backlash detection value of the key machining axis.

[0014] Optionally, the pulse arrival time series of the multiple two-order impulse response sequences are analyzed to obtain multiple backlash detection values; the weighted minimum variance fusion of the multiple backlash detection values ​​is calculated to update the backlash detection value of the key machining axis.

[0015] Secondly, this application also provides a machine tool machining axis anomaly detection system based on two-order excitation, used to execute the machine tool machining axis anomaly detection method based on two-order excitation as described in the first aspect, wherein the machine tool machining axis anomaly detection system based on two-order excitation includes: a key machining axis determination module, used to determine the key machining axes of a CNC machine tool; a model building module, used to build a shaft transmission anomaly detection model for the key machining axis, the shaft transmission anomaly detection model being obtained by training through identifying multiple historical anomaly transmission data samples of the key machining axis; and a two-order impulse response module, used to inject two-order impulses into the key machining axis through a controller. The reverse test sequence is recorded by an encoder, which includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, wherein the first amplitude is smaller than the second amplitude. An arrival time analysis module is used to analyze the pulse arrival time sequence of the double-order pulse response sequence and calculate the reverse response delay, hysteresis length, and transient speed characteristics. A detection result output module is used to input the reverse response delay, hysteresis length, and transient speed characteristics into the shaft transmission anomaly detection model for analysis and output the backlash detection value of the key machining shaft.

[0016] One or more technical solutions provided in this application have at least the following beneficial effects: By identifying the critical machining axes of a CNC machine tool, constructing a shaft transmission anomaly detection model for these critical axes (trained by identifying multiple historical anomaly transmission data samples), injecting a two-order reverse test sequence into the critical machining axis via a controller, and recording the corresponding two-order impulse response sequence via an encoder (the two-order reverse test sequence includes a first amplitude impulse test sequence and a second amplitude impulse test sequence, where the first amplitude is smaller than the second amplitude), analyzing the pulse arrival time sequence of the two-order impulse response sequence, and calculating the reverse response delay, lag length, and speed transient characteristics, and inputting these characteristics into the shaft transmission anomaly detection model for analysis, the backlash detection value of the critical machining axis is output. In other words, by constructing a shaft transmission anomaly detection model for the critical machining axes of a CNC machine tool, combined with a two-order reverse test sequence, and outputting the backlash detection value of the critical machining axis, the accuracy of machining anomaly detection is improved, thereby enhancing the production efficiency of the CNC machine tool.

[0017] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the machine tool machining axis anomaly detection method based on two-stage excitation proposed in this application.

[0020] Figure 2 This is a schematic diagram of the pulse signal for the two-stage reverse test sequence of this application.

[0021] Figure 3 This is a schematic diagram of the machine tool machining axis anomaly detection system based on two-stage excitation in this application.

[0022] Figure labeling: Module 11 for determining critical machining axis, Module 12 for model building, Module 13 for two-order impulse response, Module 14 for arrival time analysis, and Module 15 for detection result output. Detailed Implementation

[0023] This application provides a machine tool machining axis anomaly detection method and system based on two-stage excitation, solving the technical problem in existing technologies where the difficulty in accurately detecting minute anomalies in CNC machine tools leads to insufficient detection accuracy, further affecting the production efficiency of CNC machine tools. By constructing an axis transmission anomaly detection model for key machining axes of CNC machine tools and combining it with a two-stage reverse test sequence, the backlash detection value of key machining axes is output, improving the accuracy of machining anomaly detection and thus enhancing the production efficiency of CNC machine tools.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides a method for detecting abnormalities in machine tool machining axes based on two-order excitation. The method is executed by a machine tool machining axis abnormality detection system based on two-order excitation, and specifically includes the following steps: Identify the critical machining axes of the CNC machine tool.

[0026] Specifically, CNC machine tools typically have multiple machining axes responsible for feed motion in different directions. To improve machining accuracy and ensure that each machining axis operates stably, it is necessary to identify the most critical machining axis, which is usually the axis that has the greatest impact on machining accuracy, product quality, and production efficiency. These are typically the main drive axes in the CNC machine tool, such as the spindle and feed axes.

[0027] A preliminary performance evaluation is conducted on all axes of the CNC machine tool to analyze the role and importance of each axis in machining. For example, in a vertical machining center, the spindle is likely the most critical because it directly affects the contact between the tool and the workpiece and the machining accuracy. By monitoring and analyzing the load, speed, vibration, and other operating data of each axis using sensor data or machine tool historical operation records, the axes with the greatest impact on overall machining quality during normal machining and when abnormalities occur are identified. If the machine tool has a long operating history, historical machining data is obtained and analyzed, and combined with the axes most prone to failure during machining, critical machining axes are further identified. Typically, abnormalities in critical machining axes can easily lead to machining accuracy problems or more serious machine tool failures. By correctly identifying critical machining axes, focused monitoring and protection of those axes can be implemented, and potential faults or performance degradation can be identified in advance.

[0028] A shaft transmission anomaly detection model for the key machining axis is constructed. The shaft transmission anomaly detection model is obtained by training and identifying multiple historical abnormal transmission data samples of the key machining axis.

[0029] Furthermore, this application also includes the following steps: constructing a shaft transmission anomaly detection model for the key machining axis, wherein the shaft transmission anomaly detection model is obtained by training by identifying multiple historical abnormal transmission data samples of the key machining axis; wherein, the multiple historical abnormal transmission data samples include normal transmission data samples, abnormal transmission data samples corresponding to known backlash samples, and double-order impulse response sequence samples corresponding to the same double-order reverse test sequence; extracting the reverse response delay feature vector, hysteresis length feature vector, and velocity transient feature vector of the double-order impulse response sequence samples as a feature vector set for regression training, and outputting the trained shaft transmission anomaly detection model.

[0030] Specifically, multiple historical abnormal transmission data samples of critical machining axes are acquired, i.e., abnormal data of critical machining axes caused by certain factors during past machining processes. These multiple historical abnormal transmission data samples include normal transmission data samples, abnormal transmission data samples corresponding to known backlash samples, and double-order impulse response sequence samples corresponding to the same double-order reverse test sequence. For example, suppose 200 sets of historical data samples are collected, including 80 normal samples and 120 abnormal samples. The normal samples come from 3 new machine tools, and their backlash is confirmed to be between 2-4µm using a laser interferometer, with the label set to an average value of 3µm. The abnormal samples simulate faults by replacing ball screw nut pairs with different degrees of wear, and their backlash is accurately measured using a dial indicator. The label values ​​are distributed between 10µm and 30µm, such as 12.5µm, 15.0µm, 18.3µm, 22.1µm, and 28.7µm. Normal transmission data samples are state data of the CNC machine tool's transmission system during normal machining, used as comparative data samples to understand the working characteristics under normal conditions; known backlash samples contain backlash detection values ​​that have been determined in historical data; abnormal transmission data samples corresponding to known backlash samples refer to abnormal samples of the CNC machine tool's transmission system caused by backlash problems during machining; double-order impulse response sequence samples corresponding to the same double-order backlash test sequence are impulse response sequences obtained by injecting double-order backlash test sequences into key machining axes using a controller.

[0031] By analyzing two-order impulse response sequence samples, we extracted the inverse response delay feature vector, the lag length feature vector, and the speed transient feature vector. The inverse response delay feature refers to the time delay in the CNC machine tool's transmission system response after the pulse signal input. The inverse response delay reflects the dynamic performance of the CNC machine tool's transmission system, especially its sensitivity; a longer delay usually indicates significant friction or backlash in the transmission system. The lag length feature refers to the time period during the response process where there is no significant change or the system remains stationary; a longer lag usually indicates significant mechanical backlash or friction within the CNC machine tool's transmission system. The speed transient feature vector is the rate of speed change after the pulse signal input, reflecting the response capability under sudden load changes or external disturbances. The extracted feature vectors are combined into a feature vector set for training the shaft transmission anomaly detection model. For example, suppose a sample's feature vectors are: inverse response delay 0.03 seconds, lag length 0.02 seconds, and speed transient feature 18 rad / s. 2 By performing similar feature extraction on multiple samples, a set of feature vectors is obtained.

[0032] Regression training is performed using feature vectors combined with known shaft drive anomalies (such as backlash detection values). The goal of training the model is to enable the input feature vectors to accurately predict abnormal shaft drive states. Multiple historical abnormal drive data samples are divided into training and validation sets; 80% of the data is used for training, and the remaining data is used to validate the model's accuracy. The regression model is trained using feature vectors from the training set and known backlash detection values. Taking linear regression as an example, the training process is as follows: A regression model is established so that the input feature vectors can predict backlash detection values. Assume the linear regression prediction model is: y = w1*x1 + w2*x2 + w3*x3 + b, where y is the backlash detection value, i.e., the prediction target; x1, x2, and x3 are the backlash response delay, lag length, and transient speed features, respectively; w1, w2, and w3 are the model weight coefficients; and b is the bias term. By minimizing the error (the difference between the actual and predicted values), the model is trained to determine the optimal weight coefficients w1, w2, and w3, and the bias b. After training, the model is validated using samples from the validation set. By inputting the feature vectors from the validation set into the trained regression model, the predicted backlash detection value is obtained and compared with known backlash samples. For example, suppose the model predicts a backlash detection value of 0.13 for an abnormal sample, while the actual backlash detection value for the abnormal sample is 0.12. The model's performance is evaluated by calculating the mean squared error (MSE) by comparing the difference between the predicted and actual values. If the error on the validation set is small, it indicates that the model training is effective. Otherwise, a more complex regression algorithm should be used, or more training samples should be added, especially those containing different types of faults; alternatively, the features should be further optimized or selected, such as adding or deleting some unimportant features. If the regression model's prediction error on the validation set is small (e.g., MSE of 0.01), it indicates that the model can predict the backlash detection value of the shaft drive system well. Training is then stopped, resulting in the shaft drive anomaly detection model.

[0033] The controller injects a two-order reverse test sequence into the key machining axis, and the encoder records the two-order pulse response sequence corresponding to the two-order reverse test sequence. The two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, wherein the first amplitude is smaller than the second amplitude.

[0034] The first amplitude is smaller than the second amplitude, and the ratio of the first amplitude to the second amplitude is in the range of [0.2, 0.5].

[0035] Furthermore, this application also includes the following steps: injecting a two-order reverse test sequence into the critical machining axis via a controller connected to a SEPIC chopper circuit; adjusting the duty cycle of the SEPIC chopper circuit with the first amplitude to obtain a pulse test sequence for outputting the first amplitude, and adjusting the duty cycle of the SEPIC chopper circuit with the second amplitude to obtain a pulse test sequence for outputting the second amplitude; the controller injecting the pulse test sequence of the first amplitude or the pulse test sequence of the second amplitude into the critical machining axis.

[0036] Furthermore, this application also includes the following steps: wherein the SEPIC chopper circuit is connected to a conduction mode controller; historical transmission sample data of the key machining axis is identified, the temporal continuity of the historical transmission sample data is analyzed, and the conduction mode controller determines whether the key machining axis is in continuous transmission mode based on the temporal continuity; if it is in continuous transmission mode, the SEPIC chopper circuit is set to continuous conduction mode, and the duty cycle of the SEPIC chopper circuit is adjusted with a first amplitude or a second amplitude to construct a two-order reverse test sequence.

[0037] Furthermore, this application also includes the following steps: determining a pulse test sequence template, wherein the pulse test sequence template includes a plurality of pulse units, wherein each pulse unit includes a forward injection segment, a forward injection observation segment, a reverse injection segment, and a recovery stabilization segment; converting the first amplitude and the second amplitude into a first duty cycle and a second duty cycle; the controller outputs a pulse test signal based on the pulse test sequence template by controlling the duty cycle of the first amplitude or the second amplitude, thereby generating a two-order reverse test sequence.

[0038] Specifically, the controller is the core device used to control the testing process. It is responsible for injecting a two-stage reverse test sequence into the critical machining axes of the CNC machine tool and coordinating the work of other hardware devices. The SEPIC chopper circuit is a power electronic circuit used to adjust the duty cycle of the pulse signal to generate pulse test sequences of different amplitudes. The controller first generates and sends the pulse signal, which is produced by adjusting the duty cycle of the SEPIC chopper circuit. The duty cycle controls the amplitude of the pulse signal; by adjusting the strength of the amplitude, the controller can output pulse sequences of first and second amplitudes respectively. The SEPIC chopper circuit converts the input DC signal into a pulse signal with a specific amplitude by adjusting the duty cycle, and then injects it into the critical machining axis. The first amplitude pulse signal has a smaller duty cycle, and the second amplitude pulse signal has a larger duty cycle. The duty cycle refers to the proportion of time occupied by the high-level portion of the pulse signal. Controlling the duty cycle can adjust the signal amplitude, affecting the test signal strength of the CNC machine tool's transmission system. Adjusting the duty cycle is used to generate pulse signals of different amplitudes.

[0039] The controller sets a pulse signal with a first amplitude and adjusts the duty cycle of the SEPIC chopper circuit to produce a smaller pulse signal amplitude, suitable for detecting slight response characteristics. Assuming the first amplitude duty cycle is 20%, the percentage is 0.2. The controller then adjusts the duty cycle of the SEPIC chopper circuit to output a larger pulse signal, suitable for detecting larger response characteristics. Assuming the second amplitude duty cycle is 50%, the percentage is 0.5.

[0040] The controller injects signals into the critical machining axes of the CNC machine tool according to the set pulse signals. These pulse signals excite the CNC machine tool's transmission system to generate a response. The response data of the CNC machine tool's transmission system is used to analyze the transmission performance of the axes, such as backlash delay, lag length, and speed transient characteristics. The controller adjusts the duty cycle to ensure that the amplitude ratio of the first amplitude to the second amplitude is always kept within the range of [0.2, 0.5]. This helps maintain a sensitive response under smaller amplitude stimuli, while testing the extreme performance of the CNC machine tool's transmission system under larger amplitude stimuli. To protect the CNC machine tool and the workpiece, the second amplitude can be selected as no more than 50% to 80% of the rated value; to ensure that the small backlash response can be distinguished, the first amplitude can be selected as 20% to 50% of the second amplitude. For example, if the rated voltage of the machine tool is 10V, then the pulse signal amplitude of the second amplitude should be set between 5V and 8V to ensure that the rated load of the machine tool is not exceeded during the test, thereby protecting the machine tool and the workpiece. The first amplitude should be set between 1.2V and 3V to help detect minute backlash in the axis without causing excessive impact on the system. The ratio of the first amplitude to the second amplitude should be between [0.2, 0.5] to ensure a reasonable ratio between small and large amplitude signals. This ensures that the input strength of the test signal is neither too strong to affect the normal operation of the machine tool, nor too weak to effectively detect minute backlash responses.

[0041] A conduction mode controller is a control device responsible for managing the operating mode of a circuit. It monitors the timing data of critical machining axes on a CNC machine tool and analyzes whether the machining axis is in continuous drive mode. Based on the analysis results, the controller decides whether to put the SEPIC chopper circuit in continuous conduction mode to output pulse test signals.

[0042] To determine whether a critical machining axis of a CNC machine tool is in continuous drive mode, it is first necessary to collect historical transmission sample data for that axis, including parameters such as axis speed, load, current, and temperature. Based on this historical transmission sample data, the controller can analyze it to examine the axis's operating status over a period of time, checking for stable and continuous load and speed variations. The conduction mode controller analyzes the temporal continuity of the historical transmission sample data, i.e., whether a stable operating state exists. Temporal continuity refers to whether the motion of the transmission system remains continuous and stable, without large fluctuations or intermittent pauses. If the temporal continuity indicates that the axis's operating state is stable and the load does not fluctuate significantly over a period of time, it can be inferred that the CNC machine tool's transmission system is in continuous drive mode.

[0043] Based on the timing continuity analysis, the conduction mode controller determines whether the critical machining axis is in continuous drive mode. If so, the controller will decide to enter continuous conduction mode. In continuous drive mode, the SEPIC chopper circuit will be set to continuous conduction mode, and the controller will control the amplitude of the pulse test sequence by adjusting the duty cycle. In continuous conduction mode, the controller can generate two pulse signals with different amplitudes by adjusting the duty cycle to perform a two-stage reverse test. A series of precise test signals are generated using a pulse test sequence template to evaluate the dynamic response of the critical machining axis of the CNC machine tool. Pulse signals with different amplitudes are generated using pulse units defined by the pulse test sequence template, thereby constructing a two-stage reverse test sequence to detect the axis performance.

[0044] The pulse test sequence template is a test mode composed of multiple pulse units. Each pulse unit consists of different test segments and is used to detect the dynamic response of a CNC machine tool transmission system under different conditions. Each pulse unit includes a forward injection segment, a forward injection observation segment, a reverse injection segment, and a recovery and stabilization segment. The pulse unit is the smallest unit in the pulse test sequence. Each pulse unit consists of four stages used to detect the response of the CNC machine tool transmission system under different operating conditions: the forward injection segment injects a positive signal to test the response of the CNC machine tool transmission system under a positive load; the forward injection observation segment observes the response of the CNC machine tool transmission system to a positive load; the reverse injection segment injects a reverse signal to test the response of the CNC machine tool transmission system under a reverse load; and the recovery and stabilization segment observes the process of the CNC machine tool transmission system recovering to a stable state and measures the recovery capability of the CNC machine tool transmission system.

[0045] As attached Figure 2As shown, the first and second amplitudes represent two different intensities of the test signal. When generating the test signal, the first and second amplitudes need to be converted into corresponding duty cycles. Adjusting the duty cycle determines the magnitude of the pulse signal amplitude. The controller adjusts the duty cycle based on the pulse signal amplitude to ensure the pulse signal strength is appropriate. The first amplitude is usually smaller, so the controller converts it to a lower duty cycle, such as 30%; the second amplitude is usually larger, so the controller converts it to a higher duty cycle, such as 60%. For example, assuming the first amplitude is 0.5V, the controller sets the duty cycle to 30%, meaning the high-level time of the pulse signal occupies 30% of the total cycle. If the second amplitude is 1V, the controller sets the duty cycle to 60%, meaning the high-level time of the pulse signal occupies 60% of the total cycle.

[0046] Based on the duty cycles of the first and second amplitude values, corresponding pulse test signals are output. The first amplitude pulse signal outputs a pulse signal with an amplitude of 0.5V and a duty cycle of 30%; the second amplitude pulse signal outputs a pulse signal with an amplitude of 1V and a duty cycle of 60%. The controller injects these two pulse signals into the critical machining axes of the CNC machine tool according to the pulse test sequence template, resulting in a two-stage reverse test sequence, including the first and second amplitude pulse signals, used to test the response of the CNC machine tool transmission system under light and heavy loads, respectively. The two-stage reverse test sequence is a test sequence containing two pulse signals with different amplitudes. The first amplitude is smaller, and the second amplitude is larger, used to test the dynamic response of the CNC machine tool transmission system under different loads.

[0047] The controller injects either a pre-adjusted pulse signal (either a first-amplitude pulse test sequence or a second-amplitude pulse test sequence) into the critical machining axis. By injecting these two pulse signals, the dynamic response of the CNC machine tool's transmission system under different load conditions is tested, particularly for characteristics such as backlash, system hysteresis, and speed transients. For example, in one scenario, assuming a two-stage backlash test is performed on the Z-axis of the CNC machine tool, the first amplitude of the first-amplitude pulse test sequence is set to 0.5V with a duty cycle of 30%. The controller adjusts the pulse signal amplitude using the SEPIC chopper circuit, resulting in an output pulse signal with an amplitude of 0.5V and a duty cycle of 30%. The second amplitude of the second-amplitude pulse test sequence is set to 1V with a duty cycle of 60%. The controller adjusts the SEPIC chopper circuit to output a pulse signal with an amplitude of 1V and a duty cycle of 60%. The controller then sequentially injects the first and second amplitude pulse signals into the Z-axis for testing. The first amplitude pulse output test data includes: inverse response delay of 0.03 seconds, stasis length of 0.01 seconds, and velocity transient characteristic of 12 rad / s. 2 The second amplitude pulse test data includes: inverse response delay of 0.05 seconds, stasis length of 0.02 seconds, and velocity transient characteristic of 18 rad / s.2 .

[0048] An encoder is a sensor used to detect and record rotational position, speed, or other related parameters. It records the dynamic response of the shaft after a pulse signal is input, particularly the response data corresponding to the first and second amplitude pulse test sequences. The encoder records the response data after receiving the first amplitude pulse signal. For the first pulse response sequence, an anomaly detection algorithm is used for analysis. By comparing the difference between the response data and the normal response data, it is determined whether there is an anomaly in the CNC machine tool's transmission system.

[0049] If the anomaly identification result returns empty, meaning no anomaly was detected, it indicates that the CNC machine tool transmission system is working normally when receiving the first amplitude pulse signal. In this case, the controller does not need to continue executing the second amplitude pulse injection process. If the anomaly identification result returns not empty, meaning an anomaly was detected, it indicates that there is a problem with the CNC machine tool transmission system when receiving the first amplitude pulse signal, such as backlash or friction. In this case, the controller will continue executing the second amplitude pulse test sequence injection. In other words, if the anomaly identification result is empty, indicating normal operation, the controller will not continue injecting the second amplitude pulse test sequence; the test stops at the first amplitude pulse signal injection stage. If the anomaly identification result is not empty, indicating an anomaly was detected during the first amplitude test, the controller will inject the second amplitude pulse test sequence to further test the performance of the CNC machine tool transmission system under a larger load. The controller will inject a second pulse signal with a larger amplitude, i.e., the second amplitude pulse test sequence, and record the second pulse response sequence through the encoder.

[0050] In one example, suppose a two-stage reverse test is performed on the Y-axis of a CNC machine tool. The specific operation is as follows: the controller injects a pulse signal of 0.5V with a first amplitude and a duty cycle of 30%. The encoder records the first pulse response sequence: the reverse response delay is 0.03 seconds, the lag length is 0.01 seconds, and the speed transient characteristic is 12 rad / s. 2 The anomaly detection algorithm analyzes the first pulse response sequence and finds no obvious anomalies, returning an empty result. Therefore, the controller does not continue injecting the second amplitude pulse test sequence. Suppose that an anomaly occurs in the CNC machine tool transmission system during the first amplitude test, and the anomaly detection result is not empty. The controller continues to inject a second amplitude of 1V with a duty cycle of 60%. The encoder records the second pulse response sequence: inverse response delay of 0.05 seconds, static lag length of 0.03 seconds, and speed transient characteristic of 18 rad / s. 2By using pulse test sequences with first and second amplitudes, the dynamic response of the CNC machine tool transmission system under light and heavy loads is tested, providing complete transmission system performance data. By injecting pulse signals of smaller amplitudes and recording the system response, minor anomalies such as backlash and hysteresis are detected; pulse signals of larger amplitudes allow for further testing of performance under heavier loads. Injecting a two-stage reverse test sequence enhances the response detection capability under different load conditions, ensuring the detection of potential mechanical faults or transmission problems.

[0051] Furthermore, this application also includes the following steps: wherein, a first pulse response sequence corresponding to the pulse test sequence of the first amplitude is recorded by an encoder, anomaly identification is performed on the first pulse response sequence, and anomaly identification result is obtained; if the anomaly identification result returns empty, the controller is not activated to inject a pulse test sequence of the second amplitude into the key machining axis; if the anomaly identification result returns not empty, the controller is activated to inject a pulse test sequence of the second amplitude into the key machining axis, and a second pulse response sequence corresponding to the pulse test sequence of the second amplitude is recorded by an encoder.

[0052] Analyze the pulse arrival time sequence of the two-order impulse response sequence, and calculate the inverse response delay, quiescent length, and velocity transient characteristics.

[0053] The reverse response delay, static lag length, and transient speed characteristics are input into the shaft transmission anomaly detection model for analysis, and the backlash detection value of the key machining shaft is output.

[0054] Specifically, the controller injects a two-stage pulse test sequence, consisting of a first-amplitude pulse signal and a second-amplitude pulse signal. The encoder records the response data, forming a two-stage pulse response sequence. This sequence encompasses the CNC machine tool's transmission system's response to each pulse signal, including output responses such as vibration, velocity, and acceleration. The pulse arrival time sequence refers to the time data from the input pulse signal to the start of the CNC machine tool's transmission system's response. By measuring the response delay time after the pulse signal input, the dynamic response characteristics can be obtained.

[0055] Reverse response delay refers to the time delay between the input of a pulse signal and the start of the system's response in a CNC machine tool transmission system. It reflects the transmission response time; a longer delay usually indicates significant mechanical backlash, friction, or transmission problems. Reverse response delay = Pulse response start time - Pulse input time. In other words, it involves accurately identifying key timestamps in the pulse arrival time sequence, including the electrical start time of the reverse pulse and the moment when the axis actually begins to move, determined by finding the first derivative of the position data first exceeding a preset threshold. Based on the pulse arrival time sequence, the response delay is calculated, which is the time interval from the issuance of the electrical command pulse in the reverse injection segment to the moment when the physical transmission system overcomes backlash and friction and begins to generate a small but detectable displacement by the encoder; that is, the difference between the displacement moment and the pulse issuance moment. For example, assuming the input time of the first amplitude pulse signal is 0 seconds and the response time is 0.02 seconds, then the reverse response delay is 0.02 seconds. Stasis length refers to the time period during the pulse response process when the system is stationary or has no significant response, i.e., the width of the platform where the position reading remains unchanged or changes very little. Stasis length = Response start time - End time of the non-response period. For example, suppose that under a pulse signal of the first amplitude, the response begins at 0.02 seconds, but a significant change only begins at 0.03 seconds, with a lag length of 0.01 seconds. The velocity transient characteristic describes the rate of velocity change after the pulse signal input. Velocity transient characteristic = velocity change in the pulse response / response time. For example, suppose that under a pulse signal of the second amplitude, the velocity transient change is 18 rad / s. 2 This indicates that the CNC machine tool transmission system has strong response capability and good rigidity.

[0056] The extracted backlash delay, stagnation length, and transient speed characteristics are used as input data and fed into the shaft transmission anomaly detection model for analysis. Based on the input characteristics, the shaft transmission anomaly detection model calculates the backlash detection value, reflecting the degree of transmission anomaly, especially anomalies related to backlash. A large backlash detection value indicates a significant transmission problem in the CNC machine tool transmission system, such as wear or increased bearing clearance, which may lead to decreased machining accuracy. The backlash detection value is a key output value of the shaft transmission anomaly detection model, used to quantify potential backlash or similar transmission problems in the CNC machine tool transmission system. A large backlash detection value usually means that the CNC machine tool transmission system has significant mechanical backlash, which may lead to accuracy problems or excessive wear. For example, the backlash delay of the first amplitude impulse response is 0.02 seconds, the stagnation length is 0.01 seconds, and the transient speed characteristic is 12 rad / s. 2 The second amplitude impulse response has a back-current delay of 0.04 seconds, a stasis period of 0.03 seconds, and a velocity transient characteristic of 18 rad / s. 2The data is input into the shaft transmission anomaly detection model for analysis, and the output backlash detection value is 0.15, indicating that there is a certain degree of backlash or transmission problem.

[0057] By measuring backlash, the degree of abnormality in the transmission system, especially backlash issues, can be quantified. Large backlash indicates that the transmission accuracy and stability of the CNC machine tool's transmission system may be affected. When the backlash reading is high, an alarm is triggered, prompting the operator to perform timely maintenance or repair to prevent potential faults from causing more serious system problems. Accurate detection of the backlash in the CNC machine tool's transmission system allows for timely detection and repair of problems, avoiding decreased machining accuracy and production downtime due to faults.

[0058] Furthermore, this application also includes the following steps: calibrating the position of the key machining axis to obtain multiple feature positions; injecting a double-order reverse test sequence into the key machining axis at the multiple feature positions via a controller to obtain multiple double-order impulse response sequences corresponding to the multiple feature positions; analyzing the pulse arrival time sequence of the multiple double-order impulse response sequences to update the backlash detection value of the key machining axis.

[0059] Furthermore, this application also includes the following steps: analyzing the pulse arrival time series of the plurality of two-order impulse response sequences to obtain a plurality of backlash detection values; calculating the weighted minimum variance fusion of the plurality of backlash detection values ​​to update the backlash detection value of the key machining axis.

[0060] Specifically, the key machining axes of a CNC machine tool are calibrated to identify multiple characteristic positions, typically located at different points on the axis or in critical areas affecting machining quality. Calibration ensures consistent and accurate testing at different positions, such as the front, middle, and rear ends of the axis. A double-order reverse test sequence is injected into the key machining axis at each characteristic position via a controller. Each characteristic position corresponds to a double-order pulse signal sequence, including a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude. At each characteristic position, the encoder records the pulse response sequence after receiving the double-order reverse test sequence. Each response sequence includes dynamic characteristics such as reverse response delay, hysteresis length, and transient speed characteristics, used to analyze the transmission performance of the CNC machine tool's transmission system. For example, recording the pulse response data at each position yields multiple double-order pulse response sequences corresponding to multiple characteristic positions. The double-order reverse test sequence is a test sequence composed of two pulse signals with different amplitudes. The first amplitude is typically smaller, and the second amplitude is larger, the purpose of which is to test the axis's response under different loads. The two-order impulse response sequence is the response data generated by the CNC machine tool transmission system after the injection of a two-order reverse test sequence. It records the dynamic response characteristics of the CNC machine tool transmission system to pulse signals of different amplitudes.

[0061] The pulse arrival times in each two-order impulse response sequence are analyzed, as the pulse arrival time sequence reflects the speed of the system response. By analyzing the pulse arrival time sequences, the backlash delay and other dynamic characteristics at each characteristic location are calculated. For each characteristic location, a backlash detection value is calculated using a shaft transmission anomaly detection model, resulting in multiple backlash detection values. These backlash detection values ​​quantify the transmission system problems at each location, particularly backlash, friction, and hysteresis. For example, the backlash detection value at the front end of a shaft is 0.1, at the middle it is 0.2, and at the rear end it is 0.15; the transmission system state differs at different locations.

[0062] Multiple backlash detection values ​​are weighted and least variance fused to obtain a comprehensive backlash detection value. The weighted least variance fusion method typically assigns weights to each detection value based on the importance of each feature location or the reliability of the signal. Depending on the actual machining process of the machine tool, the importance of different travel segments of the axis may vary. For example, in a vertical machining center performing a critical face milling operation, the middle travel segment of the Z-axis is most frequently used; therefore, the measurement value of this segment should have a higher weight in the final evaluation of overall accuracy. The more important the accuracy, stability, and anti-interference capability of the sensor itself under specific operating conditions, the higher its reliability and the higher its weight. A sensor operating in harsh environments with high vibration and high temperature may have a lower reliability rating. The backlash detection value at each location is calculated, and the weight of each detection value is determined based on the sensor's reliability, the importance of the measurement location, or the standard deviation of the measurement error. The comprehensive backlash detection value is then calculated using the weighted least variance fusion formula. For example, suppose a two-stage backlash test is performed on the Z-axis of a CNC machine tool, and the backlash detection values ​​at three different positions are obtained: front end 0.1, middle end 0.2, and rear end 0.15. Assume the weights are as follows: front end 0.4, middle end 0.35, and rear end 0.25. According to the weighted minimum variance fusion formula, the backlash detection value is calculated to be 0.1475.

[0063] By using weighted minimum variance fusion, the backlash detection values ​​of critical machining axes are updated to obtain the current state, reflecting the overall backlash status of the critical machining axes of the CNC machine tool. Testing at multiple feature locations provides a comprehensive understanding, enabling early detection of potential problems, timely maintenance or repair, prevention of escalation of faults, and improvement of the stability and machining accuracy of the CNC machine tool.

[0064] In summary, the machine tool machining axis anomaly detection method based on two-order excitation provided in this application has the following beneficial effects: It identifies the key machining axes of a CNC machine tool; constructs an axis transmission anomaly detection model for the key machining axes, which is trained by identifying multiple historical abnormal transmission data samples of the key machining axes; injects a two-order reverse test sequence into the key machining axes through a controller, and records the corresponding two-order impulse response sequence through an encoder. The two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, where the first amplitude is smaller than the second amplitude; analyzes the pulse arrival time sequence of the two-order impulse response sequence, calculates the reverse response delay, hysteresis length, and speed transient characteristics; inputs the reverse response delay, hysteresis length, and speed transient characteristics into the axis transmission anomaly detection model for analysis, and outputs the backlash detection value of the key machining axes. In other words, by constructing an axis transmission anomaly detection model for the key machining axes of a CNC machine tool, combined with a two-order reverse test sequence, and outputting the backlash detection value of the key machining axes, the accuracy of machining anomaly detection is improved, thereby increasing the production efficiency of CNC machine tools.

[0065] Example 2: Based on the same inventive concept as the machine tool machining axis anomaly detection method based on two-stage excitation in Example 1, this application also provides a machine tool machining axis anomaly detection system based on two-stage excitation. Please refer to the appendix. Figure 3 The machine tool machining axis anomaly detection system based on dual-order excitation includes: The system includes a critical machining axis determination module 11 for determining the critical machining axes of a CNC machine tool; a model construction module 12 for constructing a shaft transmission anomaly detection model for the critical machining axis, which is obtained by training through identifying multiple historical abnormal transmission data samples of the critical machining axis; a two-order impulse response module 13 for injecting a two-order reverse test sequence into the critical machining axis through a controller and recording the corresponding two-order impulse response sequence through an encoder, wherein the two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, the first amplitude being smaller than the second amplitude; an arrival time analysis module 14 for analyzing the pulse arrival time sequence of the two-order impulse response sequence and calculating the reverse response delay, hysteresis length, and speed transient characteristics; and a detection result output module 15 for inputting the reverse response delay, hysteresis length, and speed transient characteristics into the shaft transmission anomaly detection model for analysis and outputting the backlash detection value of the critical machining axis.

[0066] Furthermore, the model building module 12 in the machine tool machining axis anomaly detection system based on dual-order excitation is also used to: construct an axis transmission anomaly detection model for the key machining axis, wherein the axis transmission anomaly detection model is obtained by training by identifying multiple historical abnormal transmission data samples of the key machining axis; wherein, the multiple historical abnormal transmission data samples include normal transmission data samples, abnormal transmission data samples corresponding to known backlash samples, and dual-order impulse response sequence samples corresponding to the same dual-order reverse test sequence; extract the reverse response delay feature vector, quiescent segment length feature vector, and velocity transient feature vector of the dual-order impulse response sequence samples as a feature vector set for regression training, and output the trained axis transmission anomaly detection model.

[0067] Furthermore, the two-order pulse response module 13 in the machine tool machining axis anomaly detection system based on two-order excitation is also used to: inject a two-order reverse test sequence into the critical machining axis through a controller, the controller being connected to a SEPIC chopper circuit; adjust the duty cycle of the SEPIC chopper circuit with the first amplitude to obtain a pulse test sequence for outputting the first amplitude, and adjust the duty cycle of the SEPIC chopper circuit with the second amplitude to obtain a pulse test sequence for outputting the second amplitude; the controller injects the pulse test sequence of the first amplitude or the pulse test sequence of the second amplitude into the critical machining axis.

[0068] Furthermore, the two-order pulse response module 13 in the machine tool machining axis anomaly detection system based on two-order excitation is also used for: recording the first pulse response sequence corresponding to the pulse test sequence of the first amplitude through an encoder, performing anomaly identification on the first pulse response sequence, and obtaining anomaly identification results; if the anomaly identification result returns empty, the controller is not activated to inject the pulse test sequence of the second amplitude into the critical machining axis; if the anomaly identification result returns not empty, the controller is activated to inject the pulse test sequence of the second amplitude into the critical machining axis, and the second pulse response sequence corresponding to the pulse test sequence of the second amplitude is recorded through an encoder.

[0069] Furthermore, the two-order pulse response module 13 in the machine tool machining axis anomaly detection system based on two-order excitation is also used for: wherein the SEPIC chopper circuit is connected to the conduction mode controller; identifying historical transmission sample data of the key machining axis, analyzing the temporal continuity of the historical transmission sample data, and the conduction mode controller determining whether the key machining axis is in continuous transmission mode based on the temporal continuity; if it is in continuous transmission mode, setting the SEPIC chopper circuit to continuous conduction mode, adjusting the duty cycle of the SEPIC chopper circuit with a first amplitude or a second amplitude, and constructing a two-order reverse test sequence.

[0070] Furthermore, the two-order pulse response module 13 in the machine tool machining axis anomaly detection system based on two-order excitation is also used to: determine a pulse test sequence template, the pulse test sequence template including a plurality of pulse units, wherein each pulse unit includes a forward injection segment, a forward injection observation segment, a reverse injection segment and a recovery stabilization segment; convert the first amplitude and the second amplitude into a first duty cycle and a second duty cycle; the controller outputs a pulse test signal based on the pulse test sequence template by controlling the duty cycle of the first amplitude or the second amplitude, thereby generating a two-order reverse test sequence.

[0071] Furthermore, the double-order impulse response module 13 in the machine tool machining axis anomaly detection system based on double-order excitation is also used to: the first amplitude is less than the second amplitude, and the ratio of the first amplitude to the second amplitude is in the range of [0.2, 0.5].

[0072] Furthermore, the machine tool machining axis anomaly detection system based on two-order excitation is also used for: calibrating the position of the key machining axis to obtain multiple feature positions; injecting two-order reverse test sequences into the key machining axis at the multiple feature positions through the controller to obtain multiple two-order pulse response sequences corresponding to the multiple feature positions; analyzing the pulse arrival time sequence of the multiple two-order pulse response sequences to update the backlash detection value of the key machining axis.

[0073] Furthermore, the machine tool machining axis anomaly detection system based on dual-order excitation is also used to: analyze the pulse arrival time sequence of the multiple dual-order pulse response sequences to obtain multiple backlash detection values; calculate the weighted minimum variance fusion of the multiple backlash detection values ​​to update the backlash detection value of the key machining axis.

[0074] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The machine tool machining axis anomaly detection method and specific examples based on two-stage excitation in Example 1 are also applicable to the machine tool machining axis anomaly detection system based on two-stage excitation in this embodiment. Through the foregoing detailed description of the machine tool machining axis anomaly detection method based on two-stage excitation, those skilled in the art can clearly understand the machine tool machining axis anomaly detection system based on two-stage excitation in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0075] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0076] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for detecting machine tool machining axis anomalies based on dual-order excitation, characterized in that, include: Identify the critical machining axes of CNC machine tools; A shaft transmission anomaly detection model for the key machining axis is constructed. The shaft transmission anomaly detection model is obtained by training and identifying multiple historical abnormal transmission data samples of the key machining axis. The controller injects a two-order reverse test sequence into the key machining axis, and the encoder records the two-order pulse response sequence corresponding to the two-order reverse test sequence. The two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, wherein the first amplitude is smaller than the second amplitude. Analyze the pulse arrival time sequence of the two-order impulse response sequence, and calculate the inverse response delay, stagnation length, and velocity transient characteristics; The reverse response delay, static lag length, and transient speed characteristics are input into the shaft transmission anomaly detection model for analysis, and the backlash detection value of the key machining shaft is output.

2. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 1, characterized in that, A two-stage reverse test sequence is injected into the key machining axis by a controller connected to a SEPIC chopper circuit. The duty cycle of the SEPIC chopper circuit is adjusted with the first amplitude to obtain a pulse test sequence for outputting the first amplitude, and the duty cycle of the SEPIC chopper circuit is adjusted with the second amplitude to obtain a pulse test sequence for outputting the second amplitude. The controller injects the critical machining axis according to the pulse test sequence of the first amplitude or the pulse test sequence of the second amplitude.

3. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 1, characterized in that, The method involves injecting a two-stage reverse test sequence into the critical machining axis via a controller. include: Specifically, the encoder records the first pulse response sequence corresponding to the pulse test sequence of the first amplitude, performs anomaly identification on the first pulse response sequence, and obtains the anomaly identification result. If the anomaly identification result returns empty, the controller is not activated to inject a pulse test sequence of the second amplitude onto the critical machining axis. If the anomaly identification result is not empty, the controller is activated to inject a pulse test sequence of the second amplitude into the key machining axis, and the encoder records the second pulse response sequence corresponding to the pulse test sequence of the second amplitude.

4. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 2, characterized in that, The method for adjusting the duty cycle of the SEPIC chopper circuit with the first amplitude includes: The SEPIC chopper circuit is connected to the conduction mode controller. The historical transmission sample data of the key machining axis is identified, and the temporal continuity of the historical transmission sample data is analyzed. The conduction mode controller determines whether the key machining axis is in continuous transmission mode based on the temporal continuity. If in continuous drive mode, the SEPIC chopper circuit is set to continuous conduction mode, and the duty cycle of the SEPIC chopper circuit is adjusted with a first amplitude or a second amplitude to construct a two-stage reverse test sequence.

5. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 4, characterized in that, The duty cycle of the SEPIC chopper circuit is adjusted using either a first or a second amplitude value to construct a two-order reverse test sequence. The method includes: A pulse test sequence template is determined, which includes several pulse units, wherein each pulse unit includes a forward injection segment, a forward injection observation segment, a reverse injection segment, and a recovery stabilization segment; Convert the first amplitude and the second amplitude into a first duty cycle and a second duty cycle; The controller controls the duty cycle of the first amplitude or the second amplitude to output a pulse test signal based on the pulse test sequence template, thereby generating a two-order reverse test sequence.

6. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 1, characterized in that, A shaft transmission anomaly detection model for the key machining axis is constructed. The shaft transmission anomaly detection model is obtained by training and identifying multiple historical abnormal transmission data samples of the key machining axis. The plurality of historical abnormal transmission data samples include normal transmission data samples, abnormal transmission data samples corresponding to known backlash samples, and double-order impulse response sequence samples corresponding to the same double-order reverse test sequence. The inverse response delay feature vector, quiescent segment length feature vector, and velocity transient feature vector of the two-order impulse response sequence sample are extracted as a feature vector set for regression training, and the trained shaft transmission anomaly detection model is output.

7. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 1, characterized in that, The method further includes injecting a two-stage reverse test sequence into the key machining axis via a controller: The key machining axis is calibrated to obtain multiple feature positions; By injecting double-order reverse test sequences into the key machining axes at the multiple feature locations using the controller, multiple double-order impulse response sequences corresponding to the multiple feature locations are obtained. Analyze the pulse arrival time series of the multiple two-order impulse response sequences to update the backlash detection value of the key machining axis.

8. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 7, characterized in that, Analyzing the pulse arrival time series of the multiple two-order impulse response sequences and updating the backlash detection value of the critical machining axis, the method includes: Analyze the pulse arrival time series of the multiple double-order impulse response sequences to obtain multiple reverse gap detection values; The weighted minimum variance fusion of the multiple backlash detection values ​​is calculated, and the backlash detection value of the key machining axis is updated.

9. The machine tool machining axis anomaly detection method based on dual-stage excitation as described in claim 1, characterized in that, The first amplitude is smaller than the second amplitude, and the ratio of the first amplitude to the second amplitude is in the range of [0.2, 0.5].

10. A machine tool machining axis anomaly detection system based on dual-stage excitation, characterized in that, The steps for implementing the machine tool machining axis anomaly detection method based on dual-order excitation according to any one of claims 1 to 9, wherein the machine tool machining axis anomaly detection system based on dual-order excitation comprises: The critical machining axis determination module is used to determine the critical machining axes of CNC machine tools; The model building module is used to build a shaft transmission anomaly detection model for the key machining axis. The shaft transmission anomaly detection model is obtained by training and identifying multiple historical abnormal transmission data samples of the key machining axis. A two-order impulse response module is used to inject a two-order reverse test sequence into the key machining axis through a controller, and to record the two-order impulse response sequence corresponding to the two-order reverse test sequence through an encoder. The two-order reverse test sequence includes a pulse test sequence with a first amplitude and a pulse test sequence with a second amplitude, wherein the first amplitude is smaller than the second amplitude. The arrival time analysis module is used to analyze the pulse arrival time sequence of the two-order impulse response sequence and calculate the reverse response delay, stagnation length, and velocity transient characteristics. The detection result output module is used to input the reverse response delay, static lag length and speed transient characteristics into the shaft transmission anomaly detection model for analysis, and output the backlash detection value of the key machining shaft.

Images