Built-in drilling machining axial force monitoring method and device

Abstract

The invention relates to the technical field of intelligent monitoring, and particularly discloses a built-in drilling machining axial force monitoring method and system.The method comprises the steps that multi-channel original force signals are synchronously collected through a plurality of annular piezoelectric type force sensors connected with a spindle mounting bolt in series, and filtering, normalization and abnormity elimination preprocessing are conducted on the multi-channel original force signals; the method comprises the following steps: eliminating abnormal data by adopting consistency verification, dynamically distributing weights in combination with historical reliability of a sensor, performing fusion calculation on a total axial force value, dynamically setting a self-adaptive early warning threshold value based on learning samples of first 10 holes, performing real-time optimization, analyzing reading distribution uniformity and pre-tightening force deviation of the sensor, and identifying an eccentric load and bolt abnormality. Through closed-loop control of a standard interface and a machine tool numerical control system, machining parameters are adjusted in a graded mode or sudden stop is triggered. According to the method, the axial force monitoring precision and the anti-interference capability are improved, the early warning false alarm and missing alarm rate is reduced, the equipment health diagnosis function is expanded, the machining quality and the equipment safety are guaranteed, and the method is suitable for high-end equipment drilling machining scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent monitoring, in particular to a built-in drilling axial force monitoring method and system. BACKGROUND

[0002] Drilling is a basic process in the field of mechanical manufacturing. Axial force, as the core physical quantity in the drilling process, directly affects the dimensional accuracy, surface quality and tool life of the processed hole, and is closely related to the load state of the spindle system. If the axial force abnormally increases, it may cause tool breakage, workpiece scrap, and even spindle damage. Therefore, real-time monitoring and control of the axial force is the key to ensuring the stability of the process.

[0003] The existing axial force monitoring technology has some problems. For example, the sensor installation method is mostly external, which is easily disturbed by cutting vibration and chip impact, and additionally increases the load of the spindle system, affecting its rigidity and processing accuracy. Some built-in solutions use a single sensor, which can only collect single-point force signals and cannot reflect the distribution characteristics of the force. The signal preprocessing method is simple, and a single filtering algorithm is mostly used, which is difficult to suppress both high-frequency noise and low-frequency drift at the same time, resulting in low signal-to-noise ratio and insufficient reliability of the original signal. The data fusion method is rough, and the multi-channel signals are directly summed or simply averaged without considering the individual differences and historical reliability of the sensors. Abnormal data is directly mixed into the calculation, resulting in large deviation of the total axial force value. In addition, there is a problem of disconnection between monitoring and control. Most solutions only output an alarm signal, which requires manual adjustment of processing parameters or shutdown, with a lagging response and easy expansion of the impact of the fault.

[0004] Therefore, there is an urgent need for a built-in drilling axial force monitoring method and system to solve the above problems. SUMMARY

[0005] The purpose of the present application is to provide a built-in drilling axial force monitoring method, comprising the following steps: Synchronously collecting multi-channel original force signals in the drilling process by a plurality of ring-shaped piezoelectric force sensors, and preprocessing the original force signals; Data fusion is performed on the preprocessed multi-channel force signals to calculate the total axial force value, wherein a weighted average algorithm is used to dynamically adjust the weight of each sensor, and abnormal data is excluded according to consistency verification; Based on historical processing data and learning algorithm, an adaptive warning threshold of the axial force is dynamically set, and the total axial force value is compared with the adaptive warning threshold in real time to generate an intelligent warning signal; The distribution uniformity of the readings of each sensor is analyzed to identify eccentric load or abnormal spindle pre-tightening force, and a device health status diagnosis result is outputted; According to the intelligent early warning signal and the equipment health state diagnosis result, adaptive closed-loop control is realized through the machine tool numerical control system, and machining parameters are dynamically adjusted or emergency stop protection is triggered.

[0006] Further, the step of synchronously collecting multi-channel original force signals in the drilling process by the plurality of circular piezoelectric force sensors and pre-processing the original force signals comprises: Synchronously collecting original force signals of all force sensor channels and storing to a temporary buffer; Carrying out digital filtering processing on the original force signals, and calculating optimal estimation values of the signals through a recursive estimation method; Carrying out normalization processing on the filtered force signals, and scaling the signals of each channel to a unified dimension; Detecting and eliminating transient interference and abnormal pulses in the signals, verifying data validity by using a sliding window variance detection method, and carrying out interpolation compensation on invalid data segments.

[0007] Further, the step of data fusion on the pre-processed multi-channel force signals and calculating a total axial force value, wherein the weight of each sensor is dynamically adjusted by using a weighted average algorithm, and abnormal data is excluded according to consistency verification, comprises: Carrying out consistency verification on the pre-processed force signals of each channel, calculating the standard deviation and correlation coefficient of the readings of each sensor, and excluding abnormal data with a deviation of more than 3 times the standard deviation from the mean value; Dynamically assigning weights according to the historical reliability of each sensor, wherein the historical reliability is obtained by statistically measuring the deviation degree of the measurement data of each sensor from the reference axial force value within a preset historical period; Data fusion is carried out by using a weighted average algorithm to calculate the total axial force value, wherein the weight is dynamically updated in each sampling period; Verifying the confidence of the total axial force value, calculating the coefficient of variation of the fusion result, and triggering the sensor calibration process if the coefficient of variation is greater than the design threshold.

[0008] Further, the step of dynamically setting an adaptive early warning threshold of the axial force based on historical machining data and a learning algorithm, and comparing the total axial force value with the adaptive early warning threshold in real time to generate an intelligent early warning signal, comprises: In the initial stage of each tool and machining parameter, the axial force data of the first 10 holes is collected as a learning sample, and the mean value and standard deviation are calculated; Dynamically setting the adaptive early warning threshold based on statistical characteristics; Comparing the total axial force value with the adaptive early warning threshold in real time to generate an intelligent early warning signal, and classifying the warning level according to the deviation size; Updating the learning sample library and recalculating the statistical characteristics.

[0009] Further, the step of analyzing the distribution uniformity of each sensor reading, identifying eccentric load or spindle preload abnormality, and outputting the equipment health status diagnosis result comprises: calculating the distribution uniformity index of each sensor reading, quantifying data distribution consistency with the coefficient of variation, and determining the existence of eccentric load if the coefficient of variation is continuously greater than the coefficient threshold value; analyzing the spindle preload state, calculating the preload deviation rate by comparing each sensor reading with the preload reference value, and marking the installation bolt abnormality if the deviation rate is greater than the deviation threshold value; integrating the eccentric load and preload abnormality data to generate the equipment health status diagnosis result, including the abnormality type, severity level and processing suggestion.

[0010] Further, the step of realizing adaptive closed-loop control through the machine tool numerical control system, dynamically adjusting the machining parameters or triggering the emergency stop protection according to the intelligent early warning signal and the equipment health status diagnosis result comprises: establishing a communication connection with the machine tool numerical control system through a standard interface, and transmitting the intelligent early warning signal and the equipment health status diagnosis result in real time; if the early warning signal is mild or moderate, automatically adjusting the machining parameters, including the feed rate and the spindle speed; if the early warning signal is severe or a serious abnormality of the equipment health status is detected, sending an emergency stop signal to the machine tool numerical control system to trigger the emergency stop protection; recording all control operations and machining parameter adjustment history, generating a machining process log, and verifying the control effect through a feedback loop.

[0011] Further, the present application also discloses an embedded drilling machining axial force monitoring system, comprising: a preprocessing module for synchronously collecting multi-channel original force signals in the drilling machining process through a plurality of circular ring-shaped piezoelectric force sensors, and preprocessing the original force signals; an adjustment module for data fusion of the preprocessed multi-channel force signals, calculating the total axial force value, wherein a weighted average algorithm is used to dynamically adjust the weight of each sensor, and abnormal data is excluded according to consistency verification; an early warning module for dynamically setting an adaptive early warning threshold value of the axial force based on historical machining data and learning algorithm, and comparing the total axial force value with the adaptive early warning threshold value in real time to generate an intelligent early warning signal; a diagnosis module for analyzing the distribution uniformity of each sensor reading, identifying eccentric load or spindle preload abnormality, and outputting the equipment health status diagnosis result; A control module is configured to realize adaptive closed-loop control through a machine tool numerical control system according to the intelligent early warning signal and the equipment health state diagnosis result, and dynamically adjust a machining parameter or trigger an emergency stop protection.

[0012] Further, the adjusting module comprises: A checking unit is configured to perform consistency checking on the pretreated channel force signals, calculate a standard deviation and a correlation coefficient of the sensor readings, and exclude abnormal data deviating from the mean value by more than 3 times the standard deviation; A distribution unit is configured to dynamically distribute weights according to historical reliabilities of the sensors, wherein the historical reliability is obtained by statistically acquiring a deviation degree of the measurement data of each sensor from a reference axial force value within a preset historical period; An updating unit is configured to perform data fusion using a weighted average algorithm to calculate a total axial force value, wherein the weights are dynamically updated in each sampling period; A calculation unit is configured to verify the confidence of the total axial force value, and trigger a sensor calibration process if a variation coefficient of the fusion result is greater than a design threshold.

[0013] The application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps of the built-in drilling axial force monitoring method when executing the computer program.

[0014] The application further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program realizes the steps of the built-in drilling axial force monitoring method when executed by a processor.

[0015] The application has the following beneficial effects: The application realizes high-precision intelligent monitoring of the drilling axial force and adaptive regulation of the machining process through the coherent steps of multi-channel force signal acquisition and pretreatment, data fusion to calculate a total axial force value, dynamic setting of an adaptive early warning threshold, identification of equipment health abnormalities and realization of adaptive closed-loop control. The axial force in drilling is directly related to the tool wear degree, the machining surface quality and the equipment running state. Precise monitoring of the axial force and regulation of the machining process accordingly are the core requirements for ensuring machining precision, prolonging tool service life and improving equipment comprehensive efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The method flowchart is provided for an embodiment of the application.

[0017] Figure 2 The system structure diagram is provided for an embodiment of the application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides a method for monitoring axial force in built-in drilling processes, comprising the following steps: S1, multiple ring-shaped piezoelectric force sensors are used to synchronously acquire multi-channel original force signals during the drilling process, and the original force signals are preprocessed to suppress environmental noise and improve the signal-to-noise ratio; S2, perform data fusion on the preprocessed multi-channel force signals, calculate the total axial force value, and use a weighted average algorithm to dynamically adjust the weight of each sensor, and eliminate abnormal data based on consistency verification; S3, dynamically set the adaptive warning threshold of axial force based on historical processing data and learning algorithm, and compare the total axial force value with the adaptive warning threshold in real time to generate an intelligent warning signal; S4, analyze the uniformity of the distribution of readings from each sensor, identify abnormal eccentric loads or spindle preload, and output the equipment health status diagnosis results; S5, based on intelligent early warning signals and equipment health status diagnosis results, achieves adaptive closed-loop control through the machine tool CNC system, dynamically adjusting machining parameters or triggering emergency stop protection.

[0021] As described in steps S1-S5 above, the existing external force measurement scheme requires changing the original clamping method, which will interfere with the processing site and the measurement accuracy is affected by the distance between the sensor and the processing area. The built-in scheme has the defects of rudimentary data processing algorithm, rigid early warning mechanism, inability to diagnose potential equipment abnormalities and lack of closed-loop control capability. Therefore, it is necessary to design a complete technical solution that integrates signal optimization, intelligent analysis and real-time control.

[0022] Traditional methods often use fixed thresholds for early warning, simply summing the signals from multiple sensors. This approach cannot adapt to the dynamic changes in force signals under different cutting tools and machining parameters, nor can it identify equipment anomalies such as eccentric loads. It can only provide post-event alarms without real-time intervention capabilities.

[0023] The present application realizes high-precision intelligent monitoring of axial force in drilling and adaptive regulation of the machining process through the coherent steps of multi-channel force signal acquisition and preprocessing, data fusion to calculate the total axial force value, dynamically setting an adaptive early warning threshold, identifying device health abnormalities, and realizing adaptive closed-loop control. Axial force in drilling is directly related to tool wear, surface quality, and equipment operating status. Precise monitoring of axial force and subsequent regulation of the machining process are the core requirements for ensuring machining precision, extending tool life, and improving equipment efficiency.

[0024] The core implementation process and principle of the monitoring method are as follows: first, through the multiple circular ring-shaped piezoelectric force sensors built between the spindle and the machine tool body, which are evenly distributed along the circumference and equal in number to the spindle mounting bolts, the multi-channel raw force signals in the drilling process are synchronously acquired and preprocessed to suppress environmental noise and improve the signal-to-noise ratio. The preprocessed multi-channel force signals are then subjected to data fusion. First, abnormal data is excluded through consistency verification. The weights of each sensor are dynamically adjusted through a weighted average algorithm, with the weights being calculated based on the historical reliability of the sensors. Then, the total axial force value is calculated to improve the accuracy of force value calculation. Subsequently, the axial force adaptive early warning threshold is dynamically set based on historical machining data and learning algorithms. The axial force data of the first 10 holes corresponding to the tool and machining parameters in the initial stage are used as learning samples to calculate the mean μ and standard deviation σ to determine the threshold. Then, the total axial force value is compared with the threshold in real time to generate an intelligent early warning signal. The uniformity of the readings of each sensor is analyzed to identify eccentric load or abnormal spindle pretightening force through quantitative indicators, and the device health status diagnosis result is output. Finally, based on the intelligent early warning signal and the diagnosis result, adaptive closed-loop control is realized through a standard interface connected to the machine tool numerical control system. When the early warning is mild or moderate, the machining parameters are dynamically adjusted. When the early warning is severe or there is a serious abnormality, an emergency stop protection is triggered to ensure machining quality and equipment safety.

[0025] In one embodiment, the step of synchronously acquiring multi-channel raw force signals during the drilling process through multiple circular ring-shaped piezoelectric force sensors and preprocessing the raw force signals comprises: S11, synchronously acquiring raw force signals of all force sensor channels at a sampling frequency not lower than 10 kHz and storing them in a temporary buffer; S12, performing digital filtering on the raw force signals using Kalman filtering algorithm or wavelet threshold denoising algorithm to suppress high-frequency noise and low-frequency drift, wherein Kalman filtering calculates the optimal estimation value of the signal through recursive estimation method; S13, performing normalization processing on the filtered force signals to scale the signals of each channel to a unified dimension, eliminating individual differences between sensors; S14, detecting and removing transient disturbances and abnormal pulses in the signals, using sliding window variance detection method for data validity verification and interpolating invalid data segments.

[0026] As described in steps S11-S14, the present application realizes the suppression of environmental noise and optimization of signal quality in the original force signal by performing preprocessing operations such as synchronous acquisition, digital filtering, normalization processing, and abnormality rejection on the multi-channel original force signals collected by multiple circular ring-shaped piezoelectric force sensors.

[0027] There are environmental disturbances such as vibration generated by high-speed rotation of the machine tool spindle and impact of cooling liquid flow in the drilling processing site. These disturbances can mix high-frequency noise and low-frequency drift into the original force signal collected by the piezoelectric force sensor. At the same time, due to manufacturing process differences, different circular ring-shaped piezoelectric force sensors have individual output deviations. If the original signal is directly used for subsequent force value calculation without preprocessing, it will lead to increased calculation result deviation and cannot accurately reflect the actual axial force state of the spindle. Therefore, it is necessary to eliminate disturbances and differences through preprocessing to ensure signal reliability. However, in the prior art, some built-in force measurement schemes only perform simple filtering on the original signal or do not perform processing, which cannot effectively remove transient disturbances and sensor individual differences, resulting in low accuracy of subsequent data fusion results.

[0028] During signal preprocessing, the original force signals of all force sensors are first synchronously acquired. The used sensor is a circular ring-shaped piezoelectric force sensor built-in installed between the spindle and the machine tool body. The number of sensors is equal to the number of spindle mounting bolts and is uniformly distributed along the circumference. The sampling frequency is set to not less than 10 kHz during the acquisition process, and the acquired original signal is stored in a temporary buffer. This sampling frequency can completely capture the dynamic changes of the axial force in drilling processing, avoiding signal detail loss or distortion due to insufficient sampling frequency. Synchronous acquisition ensures that the signals of each channel are completely aligned in the time dimension, providing a time reference for the accuracy of subsequent multi-channel data fusion. After acquisition, the original force signal is subjected to digital filtering processing. Kalman filtering algorithm or wavelet threshold denoising algorithm is used. The Kalman filtering algorithm first establishes a priori estimation model of the signal based on historical signal data, and then continuously corrects the estimation result by combining the current acquired measurement value, finally obtains the optimal estimation value of the signal. This process can effectively suppress the high-frequency noise generated by machine tool vibration and the low-frequency drift caused by sensor self-temperature drift. Wavelet threshold denoising decomposes the original signal into different frequency scales, sets a threshold for the high-frequency coefficients containing noise to suppress it, and then reconstructs the signal to remove noise. Both algorithms can make the processed signal closer to the actual axial force change trend.

[0029] The filtered force signal needs to be normalized to scale the signals of each channel to a uniform dimension. Because piezoelectric force sensors of different ring shapes have slight differences in piezoelectric coefficients during manufacturing, even when subjected to the same axial force, the output signal amplitude will differ. Normalization eliminates these individual differences, ensuring that the signals of each channel are on the same order of magnitude, thus avoiding deviations in subsequent fusion calculations due to inconsistent sensor outputs. Finally, transient interference and abnormal pulses in the signal are detected and removed. A sliding window variance detection method is used to verify data validity. A fixed window length is set, and the variance of the signal within each window is calculated sequentially. If the variance exceeds the preset normal range, transient interference or abnormal pulses are identified within that window and marked as invalid data segments. Linear interpolation or nearest-neighbor interpolation methods are then used to compensate for the invalid data segments. For example, instantaneous high-amplitude pulse signals generated by chips accidentally impacting the sensor during machining can be accurately identified and removed using sliding window variance detection. Valid data is then supplemented through interpolation, ensuring the continuity and integrity of the signal and preventing abnormal data from entering subsequent processing stages and affecting the accuracy of the total axial force calculation.

[0030] In one embodiment, the step of fusing the preprocessed multi-channel force signals to calculate the total axial force value, wherein the weights of each sensor are dynamically adjusted using a weighted average algorithm, and abnormal data is eliminated based on consistency verification, includes: S21, perform consistency verification on the preprocessed force signals of each channel, calculate the standard deviation and correlation coefficient of each sensor reading, and exclude abnormal data that deviate from the mean by more than 3 times the standard deviation. S22, dynamically allocate weights based on the historical reliability of each sensor, wherein historical reliability is obtained by statistically analyzing the deviation between the measurement data of each sensor and the reference axial force value within a preset historical period; S23 uses a weighted average algorithm to fuse data and calculate the total axial force value, where the weights are dynamically updated in each sampling period; S24, verify the confidence level of the total axial force value by calculating the coefficient of variation of the fusion result. If the coefficient of variation is greater than the design threshold, the sensor calibration process is triggered.

[0031] As described in steps S21-S24 above, the present invention achieves high-precision calculation of the total axial force value through a series of steps, including consistency verification to eliminate abnormal data of the preprocessed multi-channel force signal, dynamic weight allocation based on the historical reliability of the sensor, fusion calculation of the total axial force value using a weighted average algorithm, and verification of the confidence level of the fusion result.

[0032] Although the preprocessed multi-channel force signals have eliminated environmental noise and individual sensor differences, abnormal data may still exist due to factors such as momentary sensor failures and local vibration impacts. At the same time, the deviations between historical data and benchmark values ​​of the various ring-shaped piezoelectric force sensors and the reference values ​​may vary due to differences in aging and installation conditions over long-term use, meaning that historical reliability varies. If the total axial force is calculated by simply summing the multi-channel signals, abnormal data will directly affect the accuracy of the result. Furthermore, failing to consider the differences in sensor reliability will result in the effective data of reliable sensors not being fully utilized, ultimately increasing the deviation of the total axial force calculation result and failing to accurately reflect the actual axial force state of the spindle. Therefore, data fusion processing is needed to solve the problems of abnormal data interference and sensor reliability differences.

[0033] During data fusion calculation, the consistency of the preprocessed force signals of each channel is first checked. The standard deviation and correlation coefficient of each sensor reading are calculated. Data that deviates from the mean by more than 3 times the standard deviation is identified as abnormal data and excluded. For example, a piezoelectric force sensor with a certain ring shape may have a much higher output force signal value than other sensors due to instantaneous chip impact. Calculation shows that this value deviates from the mean by more than 3 times the standard deviation and will be identified as abnormal data and excluded to avoid abnormal data entering the subsequent fusion process and causing deviation in the calculation of the total axial force value.

[0034] Weights are dynamically assigned to each sensor based on its historical reliability. This weighting is calculated using the reciprocal of the average deviation from a reference value during past data acquisitions. The reference value is the average force value acquired by the sensor under standard operating conditions. Sensors with smaller historical average deviations from the reference value have larger reciprocals of their average deviations and are assigned higher weights. For example, if sensor A's historical average deviation from the reference value is 5N and sensor B's is 10N, sensor A's reciprocal of its average deviation is greater than sensor B's, therefore sensor A has a higher weight than sensor B. This ensures that sensors with higher historical reliability contribute more to the total axial force calculation. A weighted average algorithm is then used for data fusion to calculate the total axial force value. The formula for calculating the total axial force value is as follows: ; in, This represents the total axial force value, reflecting the actual axial force experienced by the spindle after fusion, used for subsequent early warning and control. 'n' represents the number of force sensors, equal to the number of spindle mounting bolts; in this embodiment, it is set to 6. This represents the normalized weight of the i-th sensor, derived from the dynamic allocation and normalization of the historical reliability of each sensor, ensuring that the sum of the weights is 1. This represents the preprocessed force signal from the i-th circular piezoelectric force sensor. This data originates from the multi-channel force signal data after the aforementioned "synchronous acquisition, digital filtering, normalization, and anomaly removal." The weights of each sensor are dynamically updated within each sampling period. After each sampling, the historical acquisition data of each sensor is updated, and the average deviation from the benchmark value and the corresponding weights are recalculated. This ensures that the weights always match the current historical reliability of the sensors, allowing the total axial force calculation result to adjust in real time according to changes in sensor reliability, thus better reflecting actual conditions. Finally, the confidence level of the total axial force value is verified by calculating the coefficient of variation of the fused result. The coefficient of variation is the ratio of the standard deviation to the mean of the fused result. If the coefficient of variation exceeds the design threshold of 5%, the sensor calibration process is triggered. The formula for calculating the coefficient of variation of the total axial force value is: ; in, The coefficient of variation represents the total axial force value and is used to verify the confidence level of the fusion results. Represents the total axial force value within multiple sampling periods. The standard deviation is calculated from continuously sampled total axial force data. Represents the total axial force value within multiple sampling periods. The mean value is also derived from the calculation of the total axial force data obtained through continuous sampling. For example, if the coefficient of variation of the total axial force value obtained from a certain fusion calculation is 6%, exceeding the design threshold of 5%, it will trigger the sensor calibration process to calibrate each ring-shaped piezoelectric force sensor to ensure the accuracy of subsequent total axial force calculations and avoid increased deviations in calculation results due to sensor drift and other issues.

[0035] In one embodiment, the step of dynamically setting an adaptive warning threshold for axial force based on historical processing data and a learning algorithm, and comparing the total axial force value with the adaptive warning threshold in real time to generate an intelligent warning signal includes: S31, at the initial stage of each tool and machining parameter, collect the axial force data of the first 10 holes as a learning sample, and calculate the mean μ and standard deviation σ; S32, dynamically set adaptive warning threshold based on statistical characteristics. The upper limit of the threshold is the product of the mean plus the adaptive coefficient and the standard deviation, and the lower limit of the threshold is the product of the mean minus the adaptive coefficient and the standard deviation. The adaptive coefficient is slowly adjusted according to the processing time. S33 compares the total axial force value with the adaptive warning threshold in real time, generates an intelligent warning signal, and classifies the warning level according to the magnitude of the deviation; S34, update the learning sample library, recalculate statistical features, and achieve online adaptive optimization of the threshold.

[0036] As described in steps S31-S34 above, the present invention achieves dynamic adaptation of the axial force warning threshold by collecting learning samples in the initial stage of specific tools and machining parameters, dynamically setting adaptive warning thresholds based on statistical features, comparing the total axial force value with the threshold in real time to generate warning signals, and updating the sample library to optimize the thresholds. This ensures that the warning signals can accurately reflect the axial force anomalies under different machining conditions and tool wear states.

[0037] During drilling, different tools and machining parameters correspond to different axial force reference ranges. As the tool wears down normally over time, the axial force will slowly increase. If a fixed warning threshold is used, there will be problems such as false alarms due to the threshold being lower than the actual force reference in the early stage of machining with a new tool, or false alarms due to the threshold being higher than the upper limit of the actual force value in the later stage of tool wear. It cannot adapt to the dynamic changes in the machining process. Therefore, it is necessary to adjust the threshold dynamically to match the actual working conditions.

[0038] During execution, at the initial stage of each tool and machining parameter, the axial force data of the first 10 holes are collected as learning samples. These data are the total axial force values ​​calculated after multi-channel signal preprocessing and data fusion. After the collection is completed, the mean μ and standard deviation σ of the sample set are calculated. For example, when machining aluminum alloy with a carbide drill bit at a feed rate of 0.1 mm / r and a spindle speed of 3000 r / min, the total axial force values ​​of the first 10 holes are collected, and the mean μ is calculated to be 500 N and the standard deviation σ is 20 N, which provides an initial benchmark for subsequent threshold setting. Next, an adaptive warning threshold is dynamically set based on statistical characteristics. The upper limit of the threshold is the product of the mean plus the adaptive coefficient and the standard deviation, and the lower limit of the threshold is the product of the mean minus the adaptive coefficient and the standard deviation. The adaptive coefficient is slowly adjusted according to the processing time. In the first hour of processing, the adaptive coefficient is set to 3. At this time, the upper limit of the threshold is 500N + 3 × 20N = 560N and the lower limit of the threshold is 500N - 3 × 20N = 440N. After 10 hours of processing, the tool wears normally and the axial force increases overall. The adaptive coefficient is adjusted to 3.2, and the upper limit of the threshold is adjusted accordingly to 500N + 3.2 × 20N = 564N to ensure that the threshold can adapt to the force changes caused by tool wear.

[0039] The system compares the total axial force value with an adaptive warning threshold in real time, generates intelligent warning signals, and classifies the warning level according to the magnitude of the deviation. If the total axial force value is 520N, deviating from the upper limit of the threshold by 20N, it is judged as a mild warning; if the force value is 580N, deviating from the upper limit of the threshold by 20N, it is judged as a moderate warning; if the force value is 600N, exceeding the upper limit of the threshold by 36N, it is judged as a severe warning. Different warning levels correspond to different subsequent control strategies. Finally, the learning sample library is updated by adding the total axial force value of each newly processed hole to the sample library. Every 5 new samples, the mean μ and standard deviation σ are recalculated. For example, after adding the force values ​​of 5 new holes, the recalculated mean μ is 510N and standard deviation σ is 22N. The threshold is adjusted based on the new statistical characteristics to achieve online adaptive optimization of the threshold, ensuring that subsequent warnings are always accurate.

[0040] In one embodiment, the step of analyzing the uniformity of sensor readings, identifying abnormal eccentric loads or spindle preload, and outputting equipment health status diagnostic results includes: S41, calculate the distribution uniformity index of each sensor reading, use the coefficient of variation to quantify the data distribution consistency, if the coefficient of variation is continuously greater than the coefficient threshold (10%), it is determined that there is an eccentric load. S42, analyze the spindle preload status, calculate the preload deviation rate by comparing the readings of each sensor with the preload reference value, and mark the mounting bolt as abnormal if the deviation rate is greater than the deviation threshold (15%). S43 integrates abnormal data on eccentric load and preload to generate equipment health status diagnostic results, including abnormality type, severity level, and handling recommendations.

[0041] As described in steps S41-S43 above, the present invention achieves accurate identification of eccentric load and abnormal spindle preload in drilling by calculating the distribution uniformity index of each sensor reading, analyzing the spindle preload status and integrating abnormal data, and outputs equipment health status diagnosis results including abnormality type, severity level and handling suggestions.

[0042] Multiple ring-shaped piezoelectric force sensors are evenly distributed along the circumference of the spindle and connected in series with the spindle mounting bolts. Under normal machining conditions, the axial force on the spindle bearing is evenly transmitted to each sensor, and the readings of each sensor are consistent. However, if there is an eccentric load, such as workpiece clamping misalignment causing uneven force on the spindle, or abnormal spindle preload, such as loose mounting bolts changing the sensor preload state, the readings of each sensor will show significant differences. If these abnormalities are not identified in time, it will lead to increased deviation in the machining hole position accuracy, or even cause increased spindle vibration and damage to the equipment. Therefore, it is necessary to analyze the sensor reading distribution and preload state to detect potential equipment abnormalities in a timely manner.

[0043] The uniformity index of the distribution of sensor readings was calculated. The sensor readings used were pre-processed multi-channel force signals. The coefficient of variation was used to quantify the data distribution consistency. The coefficient of variation is the ratio of the standard deviation to the mean of each sensor reading. If the coefficient of variation is consistently greater than the coefficient threshold of 10%, it is determined that there is an eccentric load. For example, for 6 sensors distributed at 60° intervals along the main shaft circumference, the pre-processed readings are 500N, 502N, 580N, 505N, 498N, and 501N, respectively. The calculated mean is 514.3N, the standard deviation is 32.1N, and the coefficient of variation is 6.2%, which is within the normal range.

[0044] When workpiece clamping offset resulted in eccentric load, the readings became 500N, 505N, 620N, 510N, 495N, and 502N, with a mean of 522N, a standard deviation of 48.5N, and a coefficient of variation of 9.3%, still within the normal range. As the degree of eccentricity increased, the readings became 490N, 495N, 650N, 515N, 485N, and 498N, with a mean of 522.2N, a standard deviation of 63.8N, and a coefficient of variation reaching 12.2%, consistently exceeding the 10% threshold, indicating the presence of eccentric load. The spindle preload status was analyzed. The preload reference value was the sensor readings corresponding to a bolt preload of 600kN during sensor installation. The preload deviation rate was calculated by comparing the current sensor readings with the preload reference value. The formula for calculating the preload deviation rate is: ; Among them, the Indicates the first Preload deviation rate corresponding to each sensor Indicates the first preprocessed step The real-time force signals from the sensors come from the preprocessing steps described above. This represents the preload reference value, which is the sensor force signal reading corresponding to the bolt preload during sensor installation. If the deviation rate exceeds the deviation threshold of 15%, the installation bolt is marked as abnormal. For example, if the sensor preload reference value is 550N and the current reading is 467.5N, the calculated preload deviation rate is 15%, which is at the threshold boundary. When the installation bolt loosens, the current reading becomes 462.5N, and the preload deviation rate reaches 16%, exceeding the 15% deviation threshold, thus marking the installation bolt as abnormal. Finally, the eccentric load and preload abnormality data are integrated. If eccentric load is determined and one installation bolt is marked as abnormal, combined with processing time and vibration data, the abnormality type is determined to be workpiece clamping misalignment and single bolt loosening, with a severity level of moderate. The recommended handling is to stop the machine, re-clamp the workpiece, and tighten the corresponding bolt, generating a complete equipment health status diagnosis result.

[0045] In one embodiment, the step of dynamically adjusting machining parameters or triggering emergency stop protection by implementing adaptive closed-loop control through the machine tool CNC system based on intelligent early warning signals and equipment health status diagnosis results includes: S51 establishes a communication connection with the machine tool CNC system through a standard interface, and transmits intelligent early warning signals and equipment health status diagnosis results in real time. S52, if the warning signal is mild or moderate, automatically adjust the machining parameters, including feed rate and spindle speed, and the adjustment strategy is based on proportional, integral and derivative control algorithms; S53, if the warning signal is severe or a serious abnormality in the health status of the equipment is detected, an emergency stop signal is sent to the CNC system of the machine tool to trigger the emergency stop protection; S54 records all control operations and machining parameter adjustment history, generates a machining process log, and verifies the control effect through feedback loops.

[0046] As described in steps S51-S54 above, the present invention establishes a communication connection with the machine tool CNC system through a standard interface, performs machining parameter adjustments or emergency stop protection according to the warning signal level and equipment health status, records control operations, and verifies the effect through feedback loops. This achieves adaptive closed-loop control of the drilling process, ensuring stable machining quality and protecting the machine tool and workpiece safety.

[0047] In drilling, intelligent early warning signals reflect abnormal axial force conditions, such as mild warnings caused by normal tool wear or severe warnings caused by the risk of tool breakage. Equipment health status diagnosis results reflect structural problems such as abnormal spindle preload. If these abnormalities are not intervened in time, mild warnings will lead to a decrease in the precision of the machined surface as tool wear intensifies, while severe warnings or serious equipment abnormalities will directly cause accidents such as tool breakage and spindle vibration damage. Therefore, it is necessary to link the monitoring results with the machine tool CNC system and prevent the problem from escalating through real-time control intervention.

[0048] A communication connection is established with the machine tool CNC system through a standard interface to transmit intelligent early warning signals and equipment health status diagnosis results in real time. The intelligent early warning signal is generated by comparing the total axial force value with the adaptive early warning threshold. The equipment health status diagnosis results are obtained by analyzing the distribution of sensor readings and the spindle preload status. The standard interface adopts the industrially common OPCUA or MTConnect protocol to ensure the real-time performance and stability of data transmission and avoid untimely control due to communication delays. The system implements graded control based on the warning signal level and equipment health status. If the warning signal is mild or moderate, for example, if the total axial force exceeds the upper limit of the adaptive warning threshold by 10%, it is determined to be a mild warning; if it exceeds by 20%, it is determined to be a moderate warning. At this time, the machining parameters, including the feed rate and spindle speed, are automatically adjusted. The adjustment strategy is based on the proportional-integral-derivative (PID) control algorithm. This algorithm calculates the deviation, integral, and rate of change of the total axial force from the target value, and outputs the corresponding feed rate and spindle speed adjustment. For example, if the initial feed rate is 0.1 mm / r, and the axial force exceeds the mild warning threshold due to slight tool wear, the PID control algorithm calculates that the feed rate needs to be reduced to 0.08 mm / r, and the spindle speed needs to be finely adjusted from 3000 r / min to 2900 r / min to bring the axial force back to the target range, extend tool life, and ensure machining quality. If the warning signal is severe, such as the total axial force value exceeding the warning threshold by 30%, or a serious abnormality in the health status of the equipment is detected, such as the deviation rate of multiple mounting bolts being greater than 15%, an emergency stop signal is sent to the machine tool CNC system to trigger emergency stop protection, so as to avoid tool breakage and damage to the workpiece or spindle vibration aggravation and damage to machine tool components.

[0049] Record all control operations and machining parameter adjustment history to generate a machining process log. The log includes warning signal levels, adjusted machining parameter values, emergency stop trigger times, etc. At the same time, verify the control effect through feedback loop, that is, collect the total axial force value and equipment status data after adjustment or before shutdown, and determine whether the adjusted axial force returns to the target range and whether the equipment abnormality is controlled. For example, after adjusting the feed rate, if the total axial force drops from 580N to 520N and stabilizes within this range, it means that the control effect meets the standard. If it does not meet the standard, the adjustment strategy is re-optimized to ensure the effectiveness of adaptive closed-loop control.

[0050] like Figure 2 As shown, the present invention also discloses a built-in axial force monitoring system for drilling, comprising: The preprocessing module is used to synchronously acquire multi-channel raw force signals during the drilling process through multiple ring-shaped piezoelectric force sensors, and to preprocess the raw force signals. The adjustment module is used to perform data fusion on the preprocessed multi-channel force signals and calculate the total axial force value. The weighted average algorithm is used to dynamically adjust the weight of each sensor and to eliminate abnormal data based on consistency verification. The early warning module is used to dynamically set an adaptive early warning threshold for axial force based on historical processing data and learning algorithms, and to compare the total axial force value with the adaptive early warning threshold in real time to generate an intelligent early warning signal. The diagnostic module is used to analyze the uniformity of the distribution of readings from various sensors, identify abnormal eccentric loads or spindle preload, and output the diagnostic results of the equipment's health status. The control module is used to achieve adaptive closed-loop control through the machine tool CNC system based on intelligent early warning signals and equipment health status diagnosis results, dynamically adjusting machining parameters or triggering emergency stop protection.

[0051] In one embodiment, the adjustment module includes: The verification unit is used to verify the consistency of the preprocessed force signals of each channel, calculate the standard deviation and correlation coefficient of each sensor reading, and exclude abnormal data that deviate from the mean by more than 3 times the standard deviation. The allocation unit is used to dynamically allocate weights based on the historical reliability of each sensor. The historical reliability is obtained by statistically analyzing the deviation between the measurement data of each sensor and the reference axial force value within a preset historical period. The update unit is used to perform data fusion using a weighted average algorithm and calculate the total axial force value, wherein the weights are dynamically updated in each sampling period. The calculation unit is used to verify the confidence level of the total axial force value. It calculates the coefficient of variation of the fusion result, and if the coefficient of variation is greater than the design threshold, it triggers the sensor calibration process.

[0052] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described built-in drilling axial force monitoring method.

[0053] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for monitoring axial force in built-in drilling.

[0054] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0055] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0056] The above description is merely a preferred embodiment of the present invention and does not limit the scope of this application. Any equivalent results or equivalent process transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of protection of this application.

Claims

1. A method for monitoring axial force in built-in drilling processes, characterized in that, Includes the following steps: Multiple ring-shaped piezoelectric force sensors are used to simultaneously acquire multi-channel raw force signals during the drilling process, and the raw force signals are preprocessed. The preprocessed multi-channel force signals are fused to calculate the total axial force value. A weighted average algorithm is used to dynamically adjust the weights of each sensor, and abnormal data is eliminated based on consistency verification. Based on historical processing data and learning algorithms, an adaptive warning threshold for axial force is dynamically set, and the total axial force value is compared with the adaptive warning threshold in real time to generate an intelligent warning signal. Analyze the uniformity of the distribution of readings from each sensor to identify abnormal eccentric loads or spindle preload, and output the equipment health status diagnosis results; Based on intelligent early warning signals and equipment health status diagnosis results, adaptive closed-loop control is achieved through the machine tool CNC system to dynamically adjust machining parameters or trigger emergency stop protection.

2. The method for monitoring axial force in built-in drilling operations according to claim 1, characterized in that, The step of synchronously acquiring multi-channel raw force signals during the drilling process using multiple ring-shaped piezoelectric force sensors and preprocessing the raw force signals includes: Simultaneously acquire the raw force signals from all force sensor channels and store them in a temporary buffer; The original force signal is digitally filtered, and the optimal estimate of the signal is calculated using a recursive estimation method. The filtered force signal is normalized to scale the signals of each channel to a uniform dimension. Transient interference and abnormal pulses in the signal are detected and removed. The sliding window variance detection method is used to verify the validity of the data, and interpolation compensation is performed on invalid data segments.

3. The method for monitoring axial force in built-in drilling operations according to claim 1, characterized in that, The steps of fusing the preprocessed multi-channel force signals to calculate the total axial force value, including dynamically adjusting the weights of each sensor using a weighted average algorithm and eliminating abnormal data based on consistency checks, include: Consistency verification is performed on the preprocessed force signals of each channel, the standard deviation and correlation coefficient of each sensor reading are calculated, and abnormal data with deviations from the mean exceeding 3 times the standard deviation are excluded. Weights are dynamically assigned based on the historical reliability of each sensor. Historical reliability is obtained by statistically analyzing the deviation between the measurement data of each sensor and the reference axial force value within a preset historical period. A weighted average algorithm is used to fuse the data and calculate the total axial force value, where the weights are dynamically updated in each sampling period. Verify the confidence level of the total axial force value by calculating the coefficient of variation of the fusion result. If the coefficient of variation is greater than the design threshold, trigger the sensor calibration process.

4. The method for monitoring axial force in built-in drilling operations according to claim 1, characterized in that, The step of dynamically setting an adaptive warning threshold for axial force based on historical processing data and a learning algorithm, and comparing the total axial force value with the adaptive warning threshold in real time to generate an intelligent warning signal includes: In the initial stage of each tool and machining parameter, the axial force data of the first 10 holes are collected as a learning sample, and the mean and standard deviation are calculated. Dynamically set adaptive early warning thresholds based on statistical characteristics; The system compares the total axial force value with the adaptive warning threshold in real time, generates an intelligent warning signal, and classifies the warning level according to the magnitude of the deviation. Update the learning sample library and recalculate the statistical features.

5. The method for monitoring axial force in built-in drilling operations according to claim 1, characterized in that, The steps of analyzing the uniformity of sensor readings, identifying abnormal eccentric loads or spindle preload, and outputting equipment health status diagnostic results include: Calculate the distribution uniformity index of each sensor reading, use the coefficient of variation to quantify the data distribution consistency, and determine that there is an eccentric load if the coefficient of variation is continuously greater than the coefficient threshold. Analyze the spindle preload condition, calculate the preload deviation rate by comparing the readings of each sensor with the preload reference value, and mark the mounting bolt as abnormal if the deviation rate is greater than the deviation threshold. Integrate eccentric load and preload anomaly data to generate equipment health status diagnostic results, including anomaly type, severity level, and handling recommendations.

6. The method for monitoring axial force in built-in drilling operations according to claim 1, characterized in that, The steps of dynamically adjusting machining parameters or triggering emergency stop protection by implementing adaptive closed-loop control through the machine tool CNC system based on intelligent early warning signals and equipment health status diagnosis results include: Establish a communication connection with the machine tool CNC system through a standard interface to transmit intelligent early warning signals and equipment health status diagnosis results in real time; If the warning signal is mild or moderate, the machining parameters, including feed rate and spindle speed, will be automatically adjusted. If the warning signal is severe or a serious abnormality in the health status of the equipment is detected, an emergency stop signal is sent to the CNC system of the machine tool to trigger emergency stop protection. Record all control operations and machining parameter adjustment history, generate machining process logs, and verify the control effect through feedback loops.

7. A built-in axial force monitoring system for drilling, characterized in that, include: The preprocessing module is used to synchronously acquire multi-channel raw force signals during the drilling process through multiple ring-shaped piezoelectric force sensors, and to preprocess the raw force signals. The adjustment module is used to perform data fusion on the preprocessed multi-channel force signals and calculate the total axial force value. The weighted average algorithm is used to dynamically adjust the weight of each sensor and to eliminate abnormal data based on consistency verification. The early warning module is used to dynamically set an adaptive early warning threshold for axial force based on historical processing data and learning algorithms, and to compare the total axial force value with the adaptive early warning threshold in real time to generate an intelligent early warning signal. The diagnostic module is used to analyze the uniformity of the distribution of readings from various sensors, identify abnormal eccentric loads or spindle preload, and output the diagnostic results of the equipment's health status. The control module is used to achieve adaptive closed-loop control through the machine tool CNC system based on intelligent early warning signals and equipment health status diagnosis results, dynamically adjusting machining parameters or triggering emergency stop protection.

8. The built-in axial force monitoring system for drilling according to claim 7, characterized in that, The adjustment module includes: The verification unit is used to verify the consistency of the preprocessed force signals of each channel, calculate the standard deviation and correlation coefficient of each sensor reading, and exclude abnormal data that deviate from the mean by more than 3 times the standard deviation. The allocation unit is used to dynamically allocate weights based on the historical reliability of each sensor. The historical reliability is obtained by statistically analyzing the deviation between the measurement data of each sensor and the reference axial force value within a preset historical period. The update unit is used to perform data fusion using a weighted average algorithm and calculate the total axial force value, wherein the weights are dynamically updated in each sampling period. The calculation unit is used to verify the confidence level of the total axial force value. It calculates the coefficient of variation of the fusion result, and if the coefficient of variation is greater than the design threshold, it triggers the sensor calibration process.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images