Real-time monitoring and inhibiting method for cutting vibration of numerical control machine tool

Abstract

The invention discloses a real-time monitoring and inhibiting method for cutting vibration of a numerical control machine tool. The method comprises the following steps: (1) collecting a vibration signal; (2) signal processing and feature extraction; (3) vibration abnormity judgment; (4) if the abnormal condition exists, dynamically adjusting the cutting parameters; (5) the monitored vibration data and cutting parameter adjustment records are stored in a cloud server; and (6) establishing a cutting parameter optimization model. According to the technical scheme, cutting vibration can be monitored in real time in multiple directions, cutting parameters are dynamically adjusted, data are stored, model optimization parameters can be constructed, and machining quality and efficiency are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of numerical control machine tools, in particular to a cutting vibration real-time monitoring and suppression method for numerical control machine tools. BACKGROUND

[0002] In the machining process of numerical control machine tools, cutting vibration can cause vibration marks on the machined surface, seriously reducing the dimensional accuracy and surface roughness of the parts, making the parts unable to meet the high-precision requirements; at the same time, vibration can exacerbate tool wear, shorten tool life, and increase production costs; in addition, strong vibration can also cause damage to the machine tool structure, affecting the stability and reliability of the machine tool.

[0003] The traditional vibration monitoring method has many drawbacks. On the one hand, it relies on a single sensor, such as using only an acceleration sensor to collect spindle vibration signals. This single-dimensional monitoring cannot fully reflect the vibration conditions of multiple parts such as tools and workpieces during cutting, resulting in inaccurate judgment of the vibration source. On the other hand, the traditional method often uses offline analysis, that is, a large amount of data is collected first and then analyzed and processed afterwards. This method has obvious response lag problems. When the cutting vibration exceeds the threshold, manual intervention to adjust the cutting parameters often leads to reduced machining efficiency, because manual adjustment takes time, and the machine tool may need to be stopped or slowed down during the adjustment process, affecting production progress. In addition, manual adjustment mainly relies on the experience of the operator, and it is difficult to adapt to complex and variable machining conditions, such as large differences in vibration characteristics under different materials, different tools, and different combinations of cutting parameters. The experience-based adjustment method lacks scientificity and precision. Therefore, there is an urgent need for a real-time and dynamic vibration monitoring and suppression method to realize intelligent optimization of the cutting process and meet the needs of modern manufacturing for high-precision and high-efficiency machining. SUMMARY

[0004] In view of the shortcomings of the prior art described above, the purpose of the present application is to provide a cutting vibration real-time monitoring and suppression method for numerical control machine tools, aiming to solve the technical problems of incomplete cutting vibration monitoring and parameter adjustment relying on manual experience in the prior art.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions: A cutting vibration real-time monitoring and suppression method for numerical control machine tools, comprising the following steps: (1) Two acceleration sensors are arranged at the spindle part of the numerical control machine tool to collect spindle radial vibration acceleration signals and axial vibration acceleration signals; an acceleration sensor and a displacement sensor are arranged at the tool clamping part to collect tool vibration acceleration signals and tool relative workpiece vibration displacement signals; an acceleration sensor and a displacement sensor are arranged at the workpiece clamping part to collect workpiece vibration acceleration signals and workpiece relative tool vibration displacement signals; (2) The collected vibration signal is transmitted to a signal processing module, which is filtered by a digital filter, and then wavelet transform is performed to complete noise reduction and feature extraction. The extracted feature parameters include the frequency, amplitude, phase and energy of the vibration signal; (3) The processed vibration signal feature parameters are analyzed using a machine learning algorithm. During the analysis process, it is determined whether there is cutting vibration abnormality by judging whether the frequency of the vibration signal is close to the resonance frequency of the machine tool structure, whether the amplitude exceeds the preset threshold, and whether the energy change rate is abnormal; (4) If it is determined that there is cutting vibration abnormality, the cutting parameters of the numerical control machine tool are dynamically adjusted according to the feature parameters of the vibration signal; (5) The monitored vibration data and cutting parameter adjustment records are stored in a cloud server for subsequent data analysis and fault diagnosis; (6) A cutting parameter optimization model is established based on the historical data stored in the cloud server, and the optimal cutting parameter combination is predicted through the model to achieve long-term stable suppression of cutting vibration.

[0006] Further, in the numerical control machine tool cutting vibration real-time monitoring and suppression method, in step (1), the sampling frequency of each acceleration sensor is not less than 10 kHz, and the measurement accuracy of each displacement sensor is not less than 0.1 μm.

[0007] Further, in the numerical control machine tool cutting vibration real-time monitoring and suppression method, in step (2), the digital filter uses an adaptive filtering algorithm, which can dynamically adjust the filter cutoff frequency according to the frequency characteristics of the vibration signal, and the filtering accuracy of the frequency band greater than 1000 Hz in the vibration signal is controlled within ± 5 Hz.

[0008] Further, in the numerical control machine tool cutting vibration real-time monitoring and suppression method, in step (3), the training data set of the machine learning algorithm contains at least 200 normal cutting samples and 150 abnormal cutting samples.

[0009] Further, in the numerical control machine tool cutting vibration real-time monitoring and suppression method, in step (4), the specific way of dynamically adjusting the cutting parameters of the numerical control machine tool is: adjusting the cutting speed according to the frequency and amplitude of the vibration signal to avoid the resonance frequency area; adjusting the feed rate according to the intensity of the vibration signal to reduce the cutting force; adjusting the cutting depth according to the change trend of the vibration signal to reduce the cutting load.

[0010] Further, in the cutting vibration real-time monitoring and suppression method of the numerical control machine tool, in step (4), the adjustment range of the cutting speed is ±8% to 15% of the current value, and the adjustment range of the feed rate is ±5% to 12% of the current value, and the adjusted cutting parameters do not exceed the rated working range of the machine tool.

[0011] Further, in the cutting vibration real-time monitoring and suppression method of the numerical control machine tool, in step (4), when the amplitude is not reduced to 70% of the threshold value after single adjustment, secondary adjustment is started, the secondary adjustment range is 1.2 to 1.5 times of the first adjustment range, and the adjustment range is 1.2 to 1.5 times of the previous adjustment range according to the progressive adjustment rule, and the adjustment direction is consistent with the first adjustment until the vibration is suppressed or the stop protection is triggered.

[0012] Further, in the cutting vibration real-time monitoring and suppression method of the numerical control machine tool, in step (5), the cloud server adopts an edge computing and cloud computing collaborative architecture, real-time data is stored through an edge node, and historical data is uploaded to the cloud server.

[0013] Further, in the cutting vibration real-time monitoring and suppression method of the numerical control machine tool, in step (6), the cutting parameter optimization model is constructed based on a genetic algorithm, the minimum amplitude is taken as an objective function, the optimization variables include the cutting speed, the feed rate and the cutting depth, the optimization iteration number is not less than 50 generations, and the iteration stops when the change amount of the objective function is less than 0.01 mm.

[0014] Advantages: The present application provides a cutting vibration real-time monitoring and suppression method of a numerical control machine tool, which has at least the following advantages compared with the prior art: (1) By arranging acceleration sensors and displacement sensors at the spindle, the tool clamping part and the workpiece clamping part, vibration acceleration and displacement signals are collected in multiple dimensions, overcoming the shortcomings of traditional single sensor monitoring, and more accurately capturing vibration sources and characteristics, laying a reliable data foundation for subsequent analysis.

[0015] (2) Adaptive filtering and wavelet transform are used for signal processing, vibration characteristic parameters are analyzed in real time combined with a machine learning algorithm, vibration abnormalities can be quickly judged, cutting parameters are dynamically adjusted, and the vibration suppression effect is ensured through a progressive adjustment mechanism, solving the problems of response lag in traditional offline analysis and low efficiency of manual adjustment, and ensuring the continuity and stability of the machining process.

[0016] (3) Data is stored by means of an edge computing and cloud computing collaborative architecture, and a cutting parameter optimization model is constructed based on historical data using a genetic algorithm, the optimal parameter combination can be predicted, and long-term stable suppression of cutting vibration is realized. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The application provides a flow chart of a cutting vibration real-time monitoring and suppression method of a numerical control machine tool. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and effects of the present application clearer and more explicit, the present application is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0019] The cutting vibration real-time monitoring and suppression method of the numerical control machine tool comprises the following steps: (1) two acceleration sensors are arranged at a spindle part of the numerical control machine tool, for collecting spindle radial vibration acceleration signals and spindle axial vibration acceleration signals; one acceleration sensor and one displacement sensor are arranged at a tool clamping part, for collecting tool vibration acceleration signals and tool vibration displacement signals relative to a workpiece; one acceleration sensor and one displacement sensor are arranged at a workpiece clamping part, for collecting workpiece vibration acceleration signals and workpiece vibration displacement signals relative to the tool; (2) the collected vibration signals are transmitted to a signal processing module, the signal processing module is filtered by a digital filter, and then noise reduction and feature extraction are completed through wavelet transform, and the extracted feature parameters include frequency, amplitude, phase and energy of the vibration signals; the vibration signals mentioned in the present application are the vibration acceleration signals and the vibration displacement signals mentioned in step (1).

[0020] (3) a machine learning algorithm is used to analyze the processed vibration signal feature parameters; in the analysis process, whether the vibration signal frequency is close to the machine tool structure resonance frequency, whether the amplitude exceeds the preset threshold value, and whether the energy change rate is abnormal are determined to determine whether there is cutting vibration abnormality; (4) if it is determined that there is cutting vibration abnormality, the cutting parameters of the numerical control machine tool are dynamically adjusted according to the feature parameters of the vibration signals; (5) the monitored vibration data and the cutting parameter adjustment records are stored to a cloud server, for subsequent data analysis and fault diagnosis; (6) a cutting parameter optimization model is established based on the stored historical data, the optimal cutting parameter combination is predicted through the model, and long-term stable suppression of the cutting vibration is realized.

[0021] As a preferred, the above-mentioned each acceleration sensor selects a piezoelectric acceleration sensor, the range is ± 50g, to meet the high-frequency vibration measurement demand; the frequency response is 0.5Hz~20kHz, which can cover the main frequency band of machine tool cutting vibration; the sampling frequency is not less than 10kHz, to ensure the capture of high-frequency resonance signal; the output signal type is voltage signal (0~5V), which is convenient for signal processing module to collect directly. The above-mentioned each displacement sensor adopts a laser displacement sensor, the measurement range is 0~50mm, the measurement accuracy is ± 0.1μm (i.e. not less than 0.1μm), to meet the high-precision processing scene; the sampling frequency is 5kHz; the non-contact measurement reduces the interference to the tool or workpiece movement. The acceleration sensor and the displacement sensor use the same time reference (such as sharing clock source) for sampling, to ensure that the signals collected by different sensors have a clear corresponding relationship on the time axis, so as to avoid data timing deviation.

[0022] Further, the mounting seat of the acceleration sensor and the displacement sensor has the following structure requirements: using butyronitrile rubber material with 60~70 shore hardness, which has shock absorption and oil resistance characteristics; and the corresponding part of the numerical control machine tool is connected by threads. The bottom of each mounting seat is provided with a 0.1mm thick polytetrafluoroethylene gasket to enhance insulation and prevent electromagnetic interference.

[0023] Further, in step (2), the digital filter uses an adaptive filtering algorithm, which can dynamically adjust the filter cutoff frequency according to the frequency characteristics of the vibration signal, and the filtering accuracy of the frequency band greater than 1000Hz (hereinafter referred to as "high frequency band") in the vibration signal is controlled within ± 5Hz.

[0024] In the cutting process, the vibration signal contains different frequency components, among which the high frequency band is usually related to machine tool structure resonance, tool high-frequency chatter, etc. Controlling the filtering accuracy of the high frequency band within ± 5Hz is to ensure accurate filtering and processing of the vibration signal of this key frequency band, which not only effectively eliminates high-frequency noise (such as electromagnetic interference generated by clutter), but also completely retains the true characteristics of high-frequency vibration (such as resonance frequency, amplitude change), providing a reliable data basis for subsequent feature parameter (frequency, amplitude, etc.) extraction through wavelet transform and machine learning algorithm to judge vibration abnormalities.

[0025] The adaptive filtering algorithm adopts a least mean square (LMS) adaptive filtering algorithm, is suitable for real-time signal processing, and can dynamically adjust the weight of the filter according to the input signal to minimize the mean square value of the error signal. The iteration step size μ is dynamically adjusted according to the signal noise intensity, and the step size is smaller when the noise is larger, that is, in the case of larger noise, the smaller step size can reduce the oscillation of the weight and improve the stability of the filter; when the noise is small, the larger step size can speed up the convergence speed. Preferably, the iteration step size μ is set to 0.01-0.05, which can balance the convergence speed and stability. Preferably, the filter order is set to 32 orders, which can ensure the filtering effect while avoiding excessive calculation.

[0026] The logic of dynamically adjusting the filter cutoff frequency is as follows: when the high-frequency component ratio in the input vibration signal exceeds 30%, the cutoff frequency is automatically raised to 1.2 kHz; when the ratio is less than 10%, the cutoff frequency is reduced to 800 Hz; and in the intermediate state, the cutoff frequency is linearly adjusted. Specifically, when the high-frequency component ratio exceeds 30%, it means that there are many high-frequency components in the signal, which may be caused by resonance or chatter. At this time, the cutoff frequency is raised to 1.2 kHz, which can better filter out these high-frequency components. When the high-frequency component ratio is less than 10%, it means that there are few high-frequency components in the signal, which may be caused by noise or other non-critical signals. At this time, the cutoff frequency is reduced to 800 Hz, which can reduce the filtering of low-frequency signals and retain more useful information. In the intermediate state, the cutoff frequency is transitioned through linear adjustment, which can avoid sudden changes in filter performance and improve system stability.

[0027] The wavelet transform is an important part of the signal processing process. Its main functions include: (1) noise reduction: removing high-frequency noise from the signal through wavelet transform and retaining the true vibration signal; (2) feature extraction: extracting feature parameters of the vibration signal from the wavelet transform results, such as frequency, amplitude, phase, and energy; (3) multi-scale analysis: decomposing the signal into different scale details and approximations to help analyze vibration characteristics in different frequency ranges.

[0028] Further, in step (3), the training data set of the machine learning algorithm contains at least 200 normal cutting samples and 150 abnormal cutting samples. The 200 normal samples cover different machining materials (including steel, aluminum, titanium alloy, etc.), tool types (including end mills and turning tools, etc.), and cutting parameter combinations; the 150 abnormal samples include typical abnormal scenarios such as resonance, tool wear, and workpiece loosening, each sample lasting 10 seconds.

[0029] Further, in step (4), the specific way of dynamically adjusting the cutting parameters of the numerical control machine tool is: adjusting the cutting speed according to the frequency and amplitude of the vibration signal to avoid the resonance frequency area; adjusting the feed rate according to the intensity of the vibration signal to reduce the cutting force; and adjusting the cutting depth according to the change trend of the vibration signal to reduce the cutting load.

[0030] Further, in step (4), the adjustment range of the cutting speed is ±8% to 15% of the current value, which can quickly and effectively avoid the resonance area and avoid excessive adjustment leading to system instability. The adjustment range of the feed rate is ±5% to 12% of the current value, which can effectively reduce the cutting force and avoid excessive adjustment leading to a significant decrease in processing efficiency. The adjusted cutting parameters do not exceed the rated working range of the machine tool.

[0031] Further, in step (4), when the amplitude is not reduced to 70% of the threshold value after single adjustment, secondary adjustment is started, and the secondary adjustment range is 1.2 to 1.5 times the first adjustment range, and the adjustment is recursively adjusted according to this rule (i.e., the adjustment range is 1.2 to 1.5 times the previous adjustment range) until the vibration is suppressed or the stop protection is triggered. The secondary adjustment range refers to the adjustment ratio of the cutting parameter being 1.2 to 1.5 times the adjustment amount of the parameter in the first adjustment, and the adjustment direction is consistent with the first adjustment (such as reducing the cutting depth, and the second adjustment is also reducing). For ease of understanding, the following example of adjusting the cutting speed when the frequency is close to resonance is described.

[0032] Suppose the current cutting speed is 1000 mm / min, the feed rate is 100 mm / min, the amplitude threshold is 1 mm, and the amplitude detected is 1.5 mm (exceeding the threshold), which needs to be adjusted according to the following logic: (1) First adjustment, the first adjustment amount of the cutting speed = 1000 mm / min x 10% = 100 mm / min.

[0033] The adjusted cutting speed = 1000-100 = 900 mm / min.

[0034] The adjusted amplitude is reduced to 1.2 mm (still higher than 70% of the threshold, i.e., 0.7 mm), which does not meet the standard and needs to start secondary adjustment.

[0035] (2) Secondary adjustment, the second adjustment amount of the cutting speed = the first adjustment amount (100 mm / min) x 1.2 = 120 mm / min.

[0036] The adjusted cutting speed = 900-120 = 780 mm / min (does not exceed the rated range of the machine tool).

[0037] The adjusted amplitude is reduced to 0.6 mm (lower than 70% of the threshold), and the vibration is suppressed, and the adjustment is completed.

[0038] If the secondary adjustment still does not suppress the vibration, continue to make tertiary adjustment, and so on, with each adjustment being 1.2-1.5 times the previous adjustment, until the vibration is suppressed or the shutdown protection is triggered.

[0039] When any of the following conditions are met, the shutdown protection is triggered: Amplitude over limit: the amplitude after adjustment exceeds 2 times the threshold value (e.g. if the threshold value is 1 mm, the amplitude after adjustment > 2 mm).

[0040] Adjustment number over limit: after the cumulative number of adjustments reaches the preset upper limit (e.g. 5 times), the amplitude is still not reduced to below 70% of the threshold value.

[0041] Parameter out of normal range: the cutting parameter after adjustment exceeds the machine tool rated range (e.g. cutting speed below the minimum operating speed of the equipment, feed rate exceeding the safety threshold).

[0042] Further, in step (5), the cloud server adopts an edge computing and cloud computing collaborative architecture, and real-time data (including vibration signals and adjustment instructions) are stored through edge nodes, and historical data (data exceeding 24 hours) are uploaded to the cloud server.

[0043] The above edge node is deployed in the local control cabinet of the machine tool and stores real-time data for 24 hours. The edge node communicates with the PLC of the machine tool through Ethernet, and the data transmission delay is ≤10 ms.

[0044] The above cloud server (which can use Aliyun IoT platform and other platforms) stores historical data in the format of "processing batch + timestamp", supports concurrent writing of 1000 data per second, and is used for subsequent model training.

[0045] The edge node synchronizes data to the cloud once an hour; when a major anomaly (such as a shutdown event) is detected, real-time alarm data is pushed to the cloud.

[0046] Further, in step (6), the cutting parameter optimization model is constructed based on a genetic algorithm, with amplitude minimization as the objective function; the optimization variables include cutting speed, feed rate, and cutting depth; the optimization iteration number is not less than 50 generations, and the iteration stops when the change of the objective function is less than 0.01 mm.

[0047] In step (6), the model is established based on the historical data stored in the cloud server, and its main significance lies in that it can predict the optimal cutting parameter combination (including cutting speed, feed rate, cutting depth, etc.). These parameter combinations are obtained through algorithm iteration optimization for different processing scenarios (such as different materials and tool types), which can guide the actual processing process, stably suppress cutting vibration from a long-term perspective, and avoid repeated vibration affecting processing quality.

[0048] The above merely provides the preferred embodiments of the present application, and is not intended to limit the present application in any form. It can be understood by those skilled in the art that the technical solutions and the inventive concept of the present application can be equivalently replaced or changed by those skilled in the art, and all these changes or replacements shall fall within the protection scope of the present application.

Claims

1. A method for real-time monitoring and suppression of cutting vibrations in a numerically controlled machine tool, characterized by, The method comprises the following steps: (1) Two acceleration sensors are arranged at the spindle part of the numerical control machine tool to collect the radial vibration acceleration signal and the axial vibration acceleration signal; an acceleration sensor and a displacement sensor are arranged at the tool clamping part to collect the vibration acceleration signal of the tool and the vibration displacement signal of the tool relative to the workpiece; an acceleration sensor and a displacement sensor are arranged at the workpiece clamping part to collect the vibration acceleration signal of the workpiece and the vibration displacement signal of the workpiece relative to the tool; (2) The collected vibration signals are transmitted to a signal processing module, the signal processing module is filtered by a digital filter, and then noise reduction and feature extraction are completed through wavelet transform, and the extracted feature parameters include the frequency, amplitude, phase and energy of the vibration signal; (3) The processed vibration signal feature parameters are analyzed by using a machine learning algorithm; in the analysis process, whether the vibration signal frequency is close to the machine tool structure resonance frequency, whether the amplitude exceeds the preset threshold, and whether the energy change rate is abnormal are judged to determine whether there is cutting vibration abnormality; (4) If it is judged that there is cutting vibration abnormality, the cutting parameters of the numerical control machine tool are dynamically adjusted according to the feature parameters of the vibration signal; (5) The monitored vibration data and the cutting parameter adjustment record are stored to a cloud server for subsequent data analysis and fault diagnosis; (6) A cutting parameter optimization model is established based on the historical data stored in the cloud server, the optimal cutting parameter combination is predicted through the model, and the long-term stable suppression of cutting vibration is realized.

2. The method of claim 1, wherein, In step (1), the sampling frequency of each acceleration sensor is not less than 10 kHz, and the measurement accuracy of each displacement sensor is not less than 0.1 μm.

3. The method of claim 1, wherein the method further comprises: In step (2), the digital filter adopts an adaptive filtering algorithm, which can dynamically adjust the filter cutoff frequency according to the frequency characteristics of the vibration signal, and the filtering accuracy of the frequency band greater than 1000 Hz in the vibration signal is controlled within ± 5 Hz.

4. The method of claim 1, wherein, In step (3), the training data set of the machine learning algorithm contains at least 200 normal cutting samples and 150 abnormal cutting samples.

5. The method of real-time monitoring and suppression of cutting vibrations in a CNC machine tool according to claim 1, characterized in that, In step (4), the specific way of dynamically adjusting the cutting parameters of the numerical control machine tool is: adjusting the cutting speed according to the frequency and amplitude of the vibration signal to avoid the resonance frequency area; adjusting the feed rate according to the intensity of the vibration signal to reduce the cutting force; adjusting the cutting depth according to the change trend of the vibration signal to reduce the cutting load.

6. The method of real-time monitoring and suppression of cutting vibrations of a numerically controlled machine tool according to claim 5, characterized in that, In step (4), the adjustment range of the cutting speed is ± 8% to 15% of the current value, the adjustment range of the feed rate is ± 5% to 12% of the current value, and the adjusted cutting parameters do not exceed the rated working range of the machine tool.

7. The method of real-time monitoring and suppression of cutting vibrations of a numerically controlled machine tool according to claim 6, characterized in that, In step (4), when the amplitude is not reduced to 70% of the threshold after single adjustment, secondary adjustment is started, the secondary adjustment range is 1.2 to 1.5 times of the first adjustment range, and the adjustment range is 1.2 to 1.5 times of the previous adjustment range according to this rule, and the adjustment direction is consistent with the first one, until the vibration is suppressed or the stop protection is triggered.

8. The method of real-time monitoring and suppression of cutting vibrations in a CNC machine tool according to claim 1, characterized in that, In step (5), the cloud server adopts an edge computing and cloud computing collaborative architecture, real-time data is stored through an edge node, and historical data is uploaded to the cloud server.

9. The method of real-time monitoring and suppression of cutting vibrations of CNC machine tools as claimed in claim 1 wherein, In step (6), the cutting parameter optimization model is constructed based on a genetic algorithm, with amplitude minimization as an objective function; the optimization variables include cutting speed, feed rate and cutting depth; the optimization iteration number is not less than 50 generations, and the iteration stops when the change of the objective function is less than 0.01 mm.

Images