Machine tool functional part performance analysis method and system based on big data

Abstract

The invention relates to the technical field of machine tool performance analysis, and provides a machine tool functional part performance analysis method and system based on big data. Multi-source operation information, including a vibration signal, a temperature signal, a motor current signal, an acoustic emission signal and working condition information, of functional parts of a machine tool is collected, and machine tool operation performance indexes are determined based on the information; and when any index exceeds a normal interval under the current working condition, the system can judge that the functional part of the machine tool is preliminarily abnormal, multi-dimensional abnormal information is constructed according to the working condition information, the vibration information and the temperature information, the fault risk level is determined according to the multi-dimensional abnormal information, and finally early warning information is sent and user feedback is acquired. Therefore, the problem of false alarm caused by non-fault factors such as tool wear in the prior art is effectively solved, and the difficulty that the credibility of an operator to early warning information is reduced is avoided.

Description

Technical Field

[0001] This application relates to the field of machine tool performance analysis technology, and more specifically, to a method and system for performance analysis of machine tool functional components based on big data. Background Technology

[0002] In modern manufacturing, the health of precision equipment such as high-precision five-axis machining centers directly affects product quality and production efficiency. To ensure the stable operation of critical components such as spindle bearings, advanced performance analysis systems are typically deployed to continuously collect multi-source data, including vibration, temperature, and current, during machine tool operation. These systems then utilize preset digital filters and health assessment criteria to evaluate component conditions. These filter parameters and criteria are usually determined based on machine tool factory calibration data, known background noise characteristics, and expected signal characteristics under standard machining conditions. However, after prolonged continuous operation, tool wear is inevitable. As the cutting edge of the tool dulls, its shape changes, affecting the interaction force between the tool and the workpiece material. This introduces new dynamic responses into the vibration signals generated during spindle system cutting, such as high-frequency but high-energy harmonic vibrations. This new vibration component does not originate from a mechanical failure of the spindle itself but is an accompanying phenomenon of normal tool wear, and its signal characteristics may differ from the "normal" vibration patterns expected during the initial system calibration.

[0003] Because the initial digital filter parameters were designed based on the initial tool condition and normal noise background, they may not be able to effectively identify and separate this new type of harmonic vibration caused by tool wear. If this new harmonic component happens to fall at the edge of the filter's passband, or if its frequency characteristics prevent it from being effectively suppressed, the filter will not be able to effectively identify and separate it from the raw data stream, resulting in data stream contamination. This unfiltered harmonic vibration component continues to mix into the raw data stream, directly affecting subsequent performance index calculations. When the performance analysis system incorrectly includes harmonic vibration components caused by normal tool wear, which are mistakenly considered valid information, in its calculations, it will cause a continuous, slow but stable "abnormal" increase in the performance indices derived from these formulas.

[0004] As these calculated performance metrics remain consistently high, they will gradually reach or exceed preset thresholds for determining component health. In this situation, because the abnormal rise in metrics does not stem from actual mechanical failure, the performance analysis system will frequently issue false alarms indicating "deteriorating health" of the spindle, even without any actual malfunction occurring. Over time, operators gradually lose trust in the system's warnings and tend to ignore them as "normal phenomena." This loss of trust renders the system's alerts worthless in terms of providing decision-making guidance.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses a method and system for performance analysis of machine tool functional components based on big data, aiming to solve the problem that existing machine tool performance analysis systems are prone to false alarms under the influence of non-fault factors such as tool wear, which leads to a decrease in the operator's trust in the warning information.

[0007] The technical solution of this application is as follows: In a first aspect, this application discloses a method for performance analysis of machine tool functional components based on big data, the method comprising: Collect multi-source operating information of machine tool functional components; this multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; Based on this multi-source operating information, the machine tool operating performance indicators are determined. If any machine tool operating performance indicator exceeds the normal range of the indicator under the current working condition, it is judged that there is a preliminary abnormality in the machine tool functional components. In response to the initial abnormality of the machine tool's functional components, a multi-dimensional abnormality information is constructed based on the operating condition information, vibration information, and temperature information. Based on the multidimensional anomaly information, the fault risk level of the machine tool's functional components is determined, and a warning message including the fault risk level is sent to the user, and user feedback is obtained.

[0008] Furthermore, in this big data-based machine tool functional component performance analysis method, the machine tool's operating performance indicators include vibration indicators, temperature indicators, current indicators, and acoustic emission indicators. Among them, the vibration indicators include the root mean square value, peak factor, kurtosis, or energy percentage within a specific frequency range of the vibration signal; the temperature indicators include the real-time temperature value or rate of change of the temperature signal; the current indicators include the harmonic content, fluctuation amplitude, or specific frequency component energy of the motor current signal; and the acoustic emission indicators include the root mean square value, count rate, or energy of the acoustic emission signal.

[0009] Based on this, in response to any machine tool operating performance index exceeding the normal range under the current working conditions, it is determined that there is a preliminary abnormality in the machine tool functional components, including: determining the abnormality of the machine tool functional components and recording the time of occurrence of the abnormality, the type of abnormal index, and the degree of deviation of the abnormal index; the degree of deviation of the abnormal index includes at least one of the abnormality degree of vibration, abnormality degree of temperature, and abnormality degree of acoustic emission.

[0010] Furthermore, the multidimensional anomaly information includes operating condition correlation information and multi-sensor cross-validation information. Based on the operating condition information, vibration information, and temperature information, the multidimensional anomaly information is constructed, including: comparing the current operating condition information with a preset operating condition feature library, calculating the degree of matching between the current anomaly data and the preset feature patterns in the preset operating condition feature library, and determining this as operating condition correlation information; wherein, the preset operating condition feature library stores preset feature patterns caused by known reasons under different operating conditions; checking whether the data from multiple types of sensors show a correlated anomaly trend within the time window of the initial anomaly occurrence, and assigning weights to the degree of anomaly and the consistency between different sensor signals, calculating a multi-sensor consistency score.

[0011] Based on the above, the fault risk level of the machine tool functional components is determined according to the multidimensional anomaly information, including: determining the fault risk level by weighted summation of the working condition correlation information, the consistency score of the multi-sensor and the degree of deviation of the anomaly indicators.

[0012] As a technological improvement, in this big data-based machine tool functional component performance analysis method, the vibration signal is a high-frequency vibration signal; the multi-source operating information also includes the three-dimensional position of the machine tool and the feed rate vector; the method further includes: identifying transient impact pulses from the high-frequency vibration signal, and synchronously analyzing the occurrence time of each impact pulse with the cutting event determined based on the three-dimensional position of the tool and the feed rate vector to determine the transient impact pulse source information; the transient impact pulse source information includes tool wear or spindle bearing degradation; detecting the transient fluctuations of the acoustic emission signal and / or the temperature signal, evaluating the temporal alignment between the transient fluctuations and the transient impact pulses, and calculating the resonance confidence level characterizing the consistency of the multi-source signals.

[0013] To improve the solution, the method sends a warning message including the fault risk level to the user and obtains user feedback. Then, the method further includes: assessing the warning confidence level based on the transient impact pulse source information and the resonance confidence level; if the warning confidence level is lower than a preset threshold, the warning message is marked as a false alarm, and the upper limit of the normal range corresponding to the machine tool's operating performance indicators is increased; if the warning confidence level is higher than the preset threshold, the warning message is marked as a confirmed fault, and the upper limit of the normal range corresponding to the machine tool's operating performance indicators is decreased.

[0014] As a further improvement, while identifying transient impact pulses from the high-frequency vibration signal, the occurrence time and energy characteristics of each transient impact pulse are recorded. The occurrence time of each impact pulse is then synchronously analyzed with the cutting event determined based on the tool's three-dimensional position and feed rate vector to determine the source information of the transient impact pulse. This includes: synchronously comparing the occurrence time of the transient impact pulse with the cutting occurrence time calculated based on the tool's three-dimensional position and feed rate vector to obtain a synchronous comparison result; simultaneously, analyzing the instantaneous current response of the motor current signal before and after the occurrence time of the impact pulse; based on the synchronous comparison result and the instantaneous current response, calculating the correlation score between the transient impact pulse originating from tool wear or spindle bearing degradation, and determining the source of the anomaly.

[0015] In some preferred embodiments, the correlation score includes a first correlation score characterizing the correlation between the transient impact pulse and the cutting process; and a second correlation score characterizing the correlation between the transient impact pulse and the bearing failure frequency; if the first correlation score is higher than a first threshold, the transient impact pulse is determined to originate from tool wear; if the first correlation score is lower than the first threshold and the second correlation score is higher than the second threshold, the transient impact pulse is determined to originate from spindle bearing degradation.

[0016] Secondly, this application also discloses a big data-based machine tool functional component performance analysis system. The system includes: a data acquisition module for acquiring multi-source operating information of the machine tool functional components; the multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; an analysis and judgment module for determining machine tool operating performance indicators based on the multi-source operating information; and, in response to any machine tool operating performance indicator exceeding the normal range under the current operating condition, determining that the machine tool functional component has a preliminary abnormality; and a construction module for, in response to the preliminary abnormality of the machine tool functional component, constructing multi-dimensional abnormality information based on the operating condition information, vibration information, and temperature information. The feedback module determines the fault risk level of the machine tool's functional components based on the multidimensional anomaly information, sends a warning message including the fault risk level to the user, and obtains user feedback.

[0017] Beneficial effects This application discloses a big data-based method for analyzing the performance of machine tool functional components. It collects multi-source operational information from machine tool functional components, including vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information, and determines machine tool performance indicators based on this information. When any indicator exceeds the normal range under current operating conditions, the system determines that the machine tool functional component has a preliminary abnormality. Based on this, multi-dimensional anomaly information is constructed according to the operating condition information, vibration information, and temperature information. The system then determines the fault risk level based on this multi-dimensional anomaly information, and finally sends a warning message and obtains user feedback. This method effectively solves the problem of false alarms caused by non-fault factors such as tool wear in existing technologies, avoiding the dilemma of reduced operator trust in warning information. By comprehensively analyzing multi-source data and combining it with operating condition information to construct multi-dimensional anomalies, this application can more accurately identify the true abnormal state of machine tool functional components, distinguish between normal wear and actual faults, thereby significantly improving the accuracy and reliability of warnings and providing a more scientific and effective decision-making basis for machine tool maintenance and management. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of the performance analysis method for machine tool functional components based on big data disclosed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a machine tool functional component performance analysis system based on big data, as disclosed in an embodiment of the present invention. Detailed Implementation

[0019] The implementation details of the technical solution in this embodiment are described in detail below: In modern manufacturing, the health of precision equipment such as high-precision five-axis machining centers directly affects product quality and production efficiency. To ensure the stable operation of critical components such as spindle bearings, advanced performance analysis systems are typically deployed to continuously collect multi-source data, including vibration, temperature, and current, during machine tool operation. These systems utilize preset digital filters and health assessment criteria to evaluate component conditions. However, after prolonged continuous operation, tool wear is inevitable, and the resulting harmonic vibrations may not be effectively identified and separated by existing filters. This leads to a continuous, slow, but stable "abnormal" increase in performance indicators, resulting in frequent false alarms and causing operators to lose trust in the system's warnings.

[0020] In response, this application proposes a method for performance analysis of machine tool functional components based on big data, such as... Figure 1 As shown, the method includes: S101 collects multi-source operating information of machine tool functional components; this multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; S102, Based on the multi-source operating information, determine the machine tool operating performance indicators. If any machine tool operating performance indicator exceeds the normal range of the indicator under the current working condition, it is determined that there is a preliminary abnormality in the machine tool functional components. S103, in response to the initial abnormality of the machine tool's functional components, construct multi-dimensional abnormality information based on the operating condition information, vibration information, and temperature information; S104. Based on the multidimensional anomaly information, determine the fault risk level of the machine tool's functional components, send a warning message including the fault risk level to the user, and obtain user feedback.

[0021] This application aims to improve the accuracy of abnormal judgment of machine tool functional components and reduce false alarms by fusing multi-source data and constructing multi-dimensional abnormal information, thereby enhancing the reliability of the early warning system.

[0022] The term "machine tool functional components" as used in this application refers to independent units that constitute a machine tool and perform specific functions, such as the spindle, feed system, tool magazine, and cooling system. The normal operation of these components is essential for ensuring the overall performance of the machine tool. "Multi-source operating information" refers to various data obtained from different sensors and data sources that reflect the operating status of machine tool functional components, including but not limited to vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information. This information collectively depicts a comprehensive picture of the machine tool's functional components' operation. "Machine tool operating performance indicators" are parameters calculated or evaluated based on multi-source operating information to quantify the health status of machine tool functional components, such as the root mean square value of vibration, real-time temperature, and current harmonic content. Fluctuations in these indicators or deviations from normal ranges usually indicate potential anomalies. "Operating condition information" refers to the machine tool's operating status parameters at a specific point in time, such as the machining program, cutting parameters (feed rate, spindle speed, depth of cut, etc.), type of material being machined, and type of tool. Operating condition information is crucial for understanding the background of performance indicators and determining the nature of anomalies.

[0023] The big data-based machine tool functional component performance analysis method of this application first requires the collection of multi-source operating information from machine tool functional components. This information forms the basis for assessing the machine tool's health status. For example, vibration signals can be collected by installing accelerometers at key locations such as the machine tool spindle, bearing housing, and motor; temperature signals can be obtained through thermocouples or infrared sensors; motor current signals can be monitored through current transformers; and acoustic emission signals can be captured through acoustic emission sensors. Furthermore, the machine tool control system can provide real-time operating condition information, such as spindle speed, feed rate, depth of cut, tool type, and machining material. This multi-source information is transmitted to the data processing unit in real-time or near real-time. After acquiring the multi-source operating information, machine tool operating performance indicators need to be determined based on this information. For example, for vibration signals, statistical indicators such as root mean square (RMS), peak factor, and kurtosis can be calculated; for temperature signals, real-time temperature values ​​can be used directly; for motor current signals, harmonic content or fluctuation amplitude can be analyzed; and for acoustic emission signals, RMS or count rate can be calculated. These indicators can be calculated using standard signal processing algorithms, such as Fourier transform and wavelet analysis. Subsequently, the determined machine tool operating performance indicators are compared with the normal ranges for the current operating conditions. These normal ranges are preset based on historical data, expert experience, or machine tool factory standards and can be dynamically adjusted according to different operating conditions. For example, under high-speed cutting conditions, the normal range for spindle vibration indicators may be higher than under low-speed idling conditions. If any machine tool operating performance indicator exceeds its normal range for the current operating conditions, a preliminary judgment is made that there is an abnormality in the machine tool's functional components. For example, if the root mean square value of spindle bearing vibration consistently exceeds a preset threshold under specific operating conditions, a preliminary abnormality is considered to exist.

[0024] When a machine tool's functional components exhibit initial abnormalities, multi-dimensional anomaly information needs to be constructed based on operating condition, vibration, and temperature data. For example, specific operating parameters at the time of the anomaly, such as spindle speed and feed rate, can be analyzed to determine if the anomaly is related to a specific operating condition. Simultaneously, vibration and temperature signals before and after the anomaly are analyzed in greater depth; for instance, checking for new frequency components in the vibration spectrum or abnormal temperature trends. This information is integrated to form a multi-dimensional anomaly description. Finally, the fault risk level of the machine tool's functional components is determined based on the constructed multi-dimensional anomaly information. For example, the fault risk level can be classified as "low risk," "medium risk," or "high risk" based on the severity of the anomaly, its duration, and its match with historical fault patterns. Once the fault risk level is determined, the system generates a warning message containing that risk level and sends it to the user via SMS, email, or the machine tool control interface. Simultaneously, the system obtains user feedback, such as whether the user confirmed the fault, ignored the warning, or took any maintenance measures. This feedback information will be used for subsequent system optimization and model iteration.

[0025] The analytical method of this application collects multi-source operational information from machine tool functional components, including vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information, providing a rich data foundation for comprehensively assessing the health status of the machine tool. Based on this multi-source information, the system can determine a series of machine tool operating performance indicators. When any indicator exceeds the normal range under the current operating condition, the system can promptly determine that there is an initial anomaly in the machine tool functional component, thereby avoiding false alarms or missed alarms that may be caused by judging a single indicator in traditional methods. After the initial anomaly occurs, this application further constructs multi-dimensional anomaly information based on operating condition information, vibration information, and temperature information. This step is one of the key innovations of this application. By integrating different types of information, the system can cross-validate and deeply analyze anomalies from multiple dimensions, such as determining whether the anomaly is related to a specific operating condition, or whether there is a consistent anomaly trend among different sensor data. This multi-dimensional analysis helps to more accurately identify the nature and potential causes of anomalies, avoiding the limitations of issuing warnings based solely on a single indicator anomaly. Finally, the system determines the fault risk level of the machine tool functional components based on the constructed multi-dimensional anomaly information, sends warning information to the user, and simultaneously obtains user feedback. This tiered early warning mechanism allows users to take appropriate maintenance measures based on the risk level, avoiding unnecessary downtime or excessive maintenance. The acquisition of user feedback creates a closed loop, enabling the system to continuously learn and optimize its judgment model, further improving the accuracy and reliability of its early warnings.

[0026] Compared to existing technologies, the big data-based machine tool functional component performance analysis method of this application has significant advantages. Traditional methods often rely on preset digital filters and fixed health judgment criteria. When factors such as tool wear introduce new harmonic vibrations, these filters may fail to effectively identify and separate them, leading to abnormal increases in performance indicators and frequent false alarms. Such false alarms not only reduce operators' trust in the early warning system but may also cause actual faults to be overlooked.

[0027] This application significantly improves the accuracy of anomaly detection by introducing multi-source operational information acquisition and multi-dimensional anomaly information construction. For example, when tool wear causes new harmonic components to appear in the vibration signal, a traditional system might directly identify it as an anomaly. However, this application, by combining operating condition information, temperature information, and other sensor data, can analyze this anomaly more comprehensively. If this vibration anomaly is highly correlated with a specific cutting condition, and other indicators such as temperature do not show signs of spindle bearing degradation, the system can more accurately determine that this may be tool wear rather than spindle failure, thereby avoiding false alarms.

[0028] Furthermore, this application establishes an intelligent early warning and optimization closed loop by determining the fault risk level and obtaining user feedback. User feedback helps the system continuously adjust and optimize its judgment model, enabling it to better adapt to changes in the machine tool operating environment and new fault modes. This adaptive capability is not possessed by traditional fixed threshold judgment methods. Therefore, this application not only effectively solves the problems of high false alarm rate and low reliability in existing technologies, but also provides more accurate and reliable performance analysis and early warning services for machine tool functional components, thereby improving production efficiency and equipment utilization. Specifically, the following methods can be used to determine machine tool operating performance indicators.

[0029] The aforementioned machine tool operating performance indicators include vibration indicators, temperature indicators, current indicators, and acoustic emission indicators. Among them, vibration indicators include the root mean square value, peak factor, kurtosis, or energy percentage within a specific frequency range of the vibration signal; temperature indicators include the real-time temperature value or rate of change of the temperature signal; current indicators include the harmonic content, fluctuation amplitude, or energy of specific frequency components of the motor current signal; and acoustic emission indicators include the root mean square value, count rate, or energy of the acoustic emission signal.

[0030] Specifically, vibration indices are quantitative descriptions of the vibration state of machine tool functional components. The root mean square (RMS) value of the vibration signal reflects the average level of vibration energy; the peak factor characterizes the intensity of the impact component in the vibration signal, which is important for early fault detection; kurtosis sensitively reflects the impact of the signal and is often used for fault diagnosis of components such as rolling bearings; the energy proportion within a specific frequency range can focus on frequency components related to specific fault modes, such as gear meshing frequency or bearing fault characteristic frequency. Temperature indices are used to monitor the thermal state of machine tool functional components. The real-time temperature value of the temperature signal directly reflects the current operating temperature of the component; excessively high temperatures usually indicate abnormal wear or poor lubrication; the rate of change of the temperature signal reflects the speed of temperature rise, and a rapid temperature rise is often an early warning signal of an impending fault.

[0031] In practical applications, current indicators are used to analyze the operating status of motors. The harmonic content of the motor current signal can reflect abnormalities in the motor windings or power supply system; the fluctuation amplitude can indicate the stability of the motor load or abnormalities in the mechanical transmission system; the energy of specific frequency components can be correlated with mechanical or electrical fault modes within the motor, such as rotor eccentricity or bearing failure. Furthermore, acoustic emission indicators are used to capture transient elastic waves generated by microscopic damage or friction within materials. The root mean square value of the acoustic emission signal reflects the average intensity of acoustic emission activity; the count rate represents the number of acoustic emission events occurring per unit time; a high count rate is usually associated with crack propagation or frictional wear; and the energy quantifies the intensity of each acoustic emission event, providing a valuable reference for assessing the severity of damage.

[0032] This application's solution refines machine tool performance indicators into vibration, temperature, current, and acoustic emission indicators, and further clarifies the specific calculation methods for these indicators. This enables multi-dimensional and refined quantitative evaluation of the operating status of machine tool functional components. The introduction of these specific indicators allows the system to capture potential anomalies in machine tool functional components from different physical phenomena levels. For example, vibration indicators focus on the smoothness of mechanical motion, temperature indicators focus on thermal load, current indicators reflect electrical and load characteristics, and acoustic emission indicators can detect microscopic damage to materials. By monitoring and analyzing these specific indicators, it is possible to more comprehensively and accurately identify whether the operating performance of machine tool functional components deviates from the normal state, thus providing a solid data foundation for subsequent preliminary anomaly judgment.

[0033] Through the above technical solutions, the detailed definition and quantification of machine tool operating performance indicators make the assessment of the health status of machine tool functional components more accurate and sensitive. Specifically, by monitoring the root mean square value, peak factor, and kurtosis of vibration signals, mechanical wear, imbalance, or impact faults can be detected earlier; real-time temperature values ​​and rates of change can provide timely warnings of overheating risks; harmonic content and fluctuation amplitude of current signals can effectively identify electrical or load anomalies in motors; and the root mean square value, count rate, or energy of acoustic emission signals can capture microscopic damage within materials. This enables early, multi-source, and high-precision detection of potential faults in machine tool functional components, significantly improving the accuracy and reliability of anomaly judgment.

[0034] Specifically, the above-mentioned steps for determining that there is a preliminary abnormality in the functional components of the machine tool when any machine tool operating performance index exceeds the normal range under the current working conditions can be further refined as follows.

[0035] If any machine tool operating performance index exceeds the normal range under the current operating conditions, it is determined that there is a preliminary abnormality in the machine tool functional component, including: determining that the machine tool functional component is abnormal, and recording the time of occurrence of the abnormality, the type of abnormal index, and the degree of deviation of the abnormal index; the degree of deviation of the abnormal index includes at least one of vibration abnormality, temperature abnormality, and acoustic emission abnormality.

[0036] Specifically, when any of the machine tool's operating performance indicators (such as vibration, temperature, current, or acoustic emission indicators) determined based on multi-source operating information exceeds the normal range under the current operating conditions, the system determines that there is an abnormality in the machine tool's functional components. Based on this determination, detailed records of this abnormality are necessary for subsequent fault diagnosis and risk assessment. The abnormality occurrence time refers to the point in time when the machine tool's operating performance indicator first exceeds the normal range; accurately recording this time helps track the evolution of the abnormality. The abnormal indicator type refers to the specific machine tool operating performance indicator that exceeds the normal range, such as vibration, temperature, current, or acoustic emission indicators. The degree of deviation of the abnormal indicator quantifies the extent to which the indicator exceeds the normal range, and can include at least one of the following: vibration abnormality degree, temperature abnormality degree, and acoustic emission abnormality degree. For example, the vibration abnormality degree can be expressed as the percentage deviation of the root mean square value, peak factor, or kurtosis of the vibration signal relative to the normal threshold; the temperature abnormality degree can be expressed as the absolute or relative value of the real-time temperature value or rate of change exceeding the normal range; and the acoustic emission abnormality degree can be expressed as the extent to which the root mean square value, count rate, or energy of the acoustic emission signal exceeds the normal threshold. By recording these detailed information, a richer data foundation can be provided for subsequent fault analysis.

[0037] This application's solution, upon determining an initial anomaly in a machine tool's functional components, goes beyond simply providing an anomaly assessment result. Instead, it records the anomaly's occurrence time, type, and degree of deviation, providing a more detailed description of the component's abnormal state. This detailed recording mechanism allows subsequent fault diagnosis and risk assessment to be based on more comprehensive information, avoiding misjudgments or insufficient information that might arise from relying on a single anomaly signal. Quantifying the degree of anomaly deviation provides a clear understanding of the anomaly's severity, offering a basis for subsequent decision-making.

[0038] In some embodiments described above in this application, a method is proposed to determine machine tool operating performance indicators based on multi-source operating information, and to determine that a preliminary abnormality exists in a machine tool functional component when any machine tool operating performance indicator exceeds the normal range under the current operating conditions. Based on this, multi-dimensional abnormality information needs to be constructed based on operating condition information, vibration information, and temperature information. Specifically, the construction of multi-dimensional abnormality information based on operating condition information, vibration information, and temperature information may further include the following steps.

[0039] The multidimensional anomaly information includes operating condition correlation information and cross-validation information from multiple sensors; the construction of multidimensional anomaly information based on the operating condition information, vibration information, and temperature information includes: The current operating condition information is compared with the preset operating condition feature library, and the degree of matching between the current abnormal data and the preset feature patterns in the preset operating condition feature library is calculated to determine the operating condition correlation information; wherein, the preset operating condition feature library stores preset feature patterns caused by known causes under different operating conditions. Examine whether the data from multiple types of sensors show a correlated abnormal trend within the time window of the initial anomaly occurrence, assign weights to the degree of anomaly of different sensor signals and their consistency, and calculate a multi-sensor consistency score.

[0040] Specifically, multidimensional anomaly information aims to provide a more comprehensive and in-depth view of anomalies, rather than simply a single indicator exceeding its threshold. Among these, operating condition correlation information is used to assess the correlation between the currently detected anomaly and typical failure modes under known operating conditions. The pre-set operating condition feature library can be understood as a knowledge base containing typical feature patterns of sensor data (such as vibration signals and temperature signals) caused by specific known causes (such as tool wear, early bearing failure, and insufficient coolant) under different machine tool operating conditions (e.g., different cutting speeds, feed rates, and machining materials). By comparing the current anomaly data with these pre-set feature patterns, the degree of matching can be calculated, thereby determining whether the current anomaly is highly correlated with a known failure mode under a specific operating condition.

[0041] Furthermore, cross-validation of multi-sensor information is used to enhance the reliability of anomaly detection. Within the time window of the initial anomaly occurrence, the system checks whether data from different types of sensors (e.g., vibration sensors, temperature sensors, acoustic emission sensors, etc.) simultaneously exhibit correlated anomaly trends. For example, if the vibration signal shows an anomaly while the temperature signal also shows a synchronous increase, this multi-sensor consistency greatly increases the confidence level of the anomaly detection. To quantify this consistency, weights can be assigned to the degree of anomaly in different sensor signals and their mutual consistency, and a multi-sensor consistency score can be calculated based on these weights. The higher the score, the higher the authenticity of the anomaly and the lower the probability of false alarms.

[0042] This application's solution constructs multi-dimensional anomaly information by introducing operating condition correlation information and multi-sensor cross-validation information. Its working principle improves the accuracy and reliability of anomaly detection from two main dimensions. First, operating condition correlation information allows the system to analyze current anomalies within the context of specific operating conditions. Traditional methods may judge anomalies solely based on whether indicators exceed fixed thresholds, but some indicators may naturally be elevated under specific operating conditions, not necessarily indicating a true fault precursor. By comparing with preset feature patterns in a pre-set operating condition feature library, the system can identify anomalies inconsistent with the current operating conditions or confirm anomalies highly consistent with known fault patterns under specific operating conditions, thereby avoiding false alarms or missed alarms. Second, multi-sensor cross-validation information utilizes the inherent correlation between data from different sensors. A single sensor may be affected by noise interference or localized sporadic events, leading to misjudgments. When multiple different types of sensors simultaneously show correlated anomaly trends within the same time window, this mutual verification of multi-source information greatly enhances the confidence of anomaly judgment. By assigning weights to the degree of anomaly and consistency of different sensor signals and calculating consistency scores, the system can more objectively and comprehensively assess the authenticity of anomalies, thereby providing a more solid data foundation for subsequent determination of fault risk levels.

[0043] Through the above technical solutions, this application can significantly improve the accuracy and reliability of anomaly detection for machine tool functional components. Specifically, by constructing operating condition correlation information, it can effectively distinguish between normal operating condition fluctuations and actual fault precursors, reducing false alarms caused by changes in operating conditions, and making anomaly judgment more targeted and effective. Simultaneously, through cross-validation of multi-sensor information, utilizing the complementarity and redundancy between different sensor data, it can effectively suppress interference from single sensor noise or sporadic events, significantly reducing the false alarm rate and improving the ability to capture real anomalies. Therefore, the constructed multi-dimensional anomaly information can more comprehensively and accurately reflect the actual operating status of machine tool functional components, providing a more reliable basis for subsequent fault risk level assessment, thereby enhancing the practical value and decision support capability of the entire performance analysis method.

[0044] In some of the embodiments described above in this application, although it is possible to collect multi-source operational information, determine machine tool performance indicators based on this information, and then judge whether there are preliminary anomalies in machine tool functional components, and record the degree of deviation of the anomaly indicators, while constructing multi-dimensional anomaly information including working condition correlation information and multi-sensor consistency scores, there is still room for further optimization in effectively integrating these anomaly information of different dimensions and natures to accurately and objectively determine the fault risk level of machine tool functional components. If a systematic quantitative method is lacking to comprehensively consider these factors, the accuracy and reliability of fault risk assessment may be insufficient, affecting the effectiveness of early warning.

[0045] In this regard, this application further proposes a step for determining the fault risk level of the machine tool functional components based on the above-mentioned multi-dimensional anomaly information, including: determining the fault risk level by weighted summation of the working condition correlation information, the multi-sensor consistency score, and the degree of deviation of the anomaly index.

[0046] Specifically, determining the fault risk level is achieved by weighted summation of operating condition correlation information, multi-sensor consistency scores, and the degree of deviation of abnormal indicators. Operating condition correlation information reflects the degree of matching between the current abnormal data and known fault modes in a pre-set operating condition feature library; a higher value indicates a stronger correlation between the current anomaly and a specific fault mode. The multi-sensor consistency score characterizes the degree to which data from different types of sensors exhibit correlated abnormal trends within the time window of the initial anomaly occurrence, as well as the reliability of these anomaly degrees and their consistency; a higher score indicates a higher confidence level in the anomaly judgment. The degree of deviation of abnormal indicators, such as the degree of vibration anomaly, temperature anomaly, and acoustic emission anomaly, quantifies the severity of machine tool operating performance indicators exceeding the normal range. By assigning different weights to these three types of information and performing a weighted summation, the fault risk of machine tool functional components can be comprehensively assessed. For example, these weights can be determined based on historical data, expert experience, or machine learning models to ensure the accuracy and reliability of the risk assessment.

[0047] This application's solution effectively addresses the challenge of comprehensively considering multi-dimensional anomaly information when determining the fault risk level of machine tool functional components by introducing a weighted summation method. Specifically, operating condition correlation information can associate current anomalies with known operating condition characteristic patterns, thus providing directional guidance for preliminary fault diagnosis. Multi-sensor consistency scores significantly improve the confidence of anomaly judgments by cross-validating the anomaly trends of different sensor data, avoiding erroneous judgments caused by false alarms from a single sensor or noise interference. Simultaneously, the degree of deviation of anomaly indicators directly quantifies the degree of performance degradation, providing a severity basis for risk assessment. By weighting and summing these three types of information, this application can assign different weights based on their respective contributions to fault risk, thereby achieving a comprehensive, objective, and quantitative assessment of the fault risk level of machine tool functional components, avoiding inaccuracies caused by subjective judgments or one-sided information.

[0048] Through the above technical solution, this application enables a more accurate and objective assessment of the fault risk level of machine tool functional components. The weighted summation method allows for the systematic integration of abnormal information from different sources and of different natures, fully considering multiple key factors such as operating conditions, multi-sensor verification, and the severity of the abnormality, thereby improving the comprehensiveness and reliability of fault risk assessment. This effectively reduces false alarm and false negative rates, making early warning information more instructive and helping users take timely maintenance measures to avoid potential equipment damage and production interruptions, ultimately improving the operating efficiency and safety of machine tools.

[0049] In some preferred embodiments, it is assumed that under a certain cutting condition, the vibration signal, temperature signal, and acoustic emission signal of the machine tool spindle bearing all show abnormalities. Through analysis, the calculated condition correlation information is 0.8 (highly matching the preset characteristic pattern of bearing wear), the multi-sensor consistency score is 0.9 (the abnormal trends of vibration, temperature, and acoustic emission signals are highly consistent), and the deviation degree of the abnormal index is 0.7 (the root mean square value of vibration exceeds the normal range by 70%). If the preset weights are: condition correlation information weight W1 = 0.4, multi-sensor consistency score weight W2 = 0.3, and abnormal index deviation degree weight W3 = 0.3, then the fault risk level can be calculated as: Risk level = 0.8 * 0.4 + 0.9 * 0.3 + 0.7 * 0.3 = 0.32 + 0.27 + 0.21 = 0.80. According to the preset risk level classification standard (e.g., above 0.7 is high risk), the system will determine that the current machine tool functional component has a high fault risk and send corresponding warning information.

[0050] In some embodiments described above in this application, traditional existing machine tool functional component performance analysis methods based on big data can determine the presence of preliminary abnormalities in machine tool functional components by collecting multi-source operating information and determining machine tool operating performance indicators, and further construct multi-dimensional abnormality information to determine the fault risk level. However, in practical applications, for some transient, high-frequency fault phenomena, such as impacts caused by tool wear or spindle bearing degradation, relying solely on macroscopic index analysis of conventional vibration, temperature, current, and acoustic emission signals may be insufficient to accurately identify their specific sources, thus affecting the accuracy of fault diagnosis and the timeliness of early warning. If these problems are not addressed, misjudgments or omissions may occur, thereby affecting the normal operation and maintenance efficiency of the machine tool. To address this, this application proposes a more refined analysis method that introduces high-frequency vibration signals and combines them with tool motion information to determine the source of transient impact pulses, and verifies this by utilizing the consistency of multi-source signals, thereby improving the accuracy and reliability of fault diagnosis.

[0051] In some embodiments of this application described above, the vibration signal is specifically set to a high-frequency vibration signal; the multi-source operating information further includes the three-dimensional position of the machine tool and the feed speed vector.

[0052] The method further includes: identifying transient impact pulses from the high-frequency vibration signal, and synchronously analyzing the occurrence time of each impact pulse with the cutting event determined based on the tool's three-dimensional position and feed rate vector to determine the transient impact pulse source information; the transient impact pulse source information includes tool wear or spindle bearing degradation; detecting transient fluctuations in the acoustic emission signal and / or the temperature signal, evaluating the temporal alignment between the transient fluctuations and the transient impact pulses, and calculating the resonance confidence level characterizing the consistency of multi-source signals.

[0053] Specifically, limiting the vibration signal to high-frequency vibration signals aims to more effectively capture minute, rapid impacts or vibrations caused by tool wear, bearing damage, etc. These events are typically insignificant at lower frequencies but exhibit significant energy characteristics at higher frequencies. High-frequency vibration signals can be understood as vibration data typically in the kilohertz or even higher frequency range, which can be acquired using piezoelectric accelerometers, for example. The tool's three-dimensional position and feed rate vectors added to the multi-source operational information refer to the real-time coordinates of the tool in space during machining, as well as its speed and direction of movement. This information can be directly acquired by the machine tool's CNC system, with the aim of accurately predicting or determining the timing and location of cutting events.

[0054] In practical applications, identifying transient impact pulses from high-frequency vibration signals specifically refers to extracting impact signals with short-duration, high-energy characteristics from continuous high-frequency vibration signals using signal processing techniques such as envelope demodulation, wavelet analysis, and peak detection. The occurrence time of each impact pulse refers to the start or peak time of the impact signal on the time axis. Furthermore, the occurrence time of each impact pulse is synchronously analyzed with the cutting event determined based on the tool's three-dimensional position and feed rate vector. The purpose is to distinguish whether the impact pulse originates from a normal cutting process or an abnormal component failure. The cutting event refers to the instant when the tool contacts the workpiece and removes material. Using the tool's three-dimensional position and feed rate vector, the exact time and location of the cutting action can be calculated, thus determining the occurrence time of the cutting event. Synchronous analysis involves comparing the occurrence time of the impact pulse with the occurrence time of the cutting event to determine if there is a temporal correlation between the two. This allows for the identification of transient impact pulse source information, which includes tool wear or spindle bearing degradation. For example, if the impact pulse is highly synchronized with the cutting event, it may indicate tool wear; if the impact pulse is less correlated with the cutting event but correlated with the spindle's rotational frequency or its harmonics, it may indicate spindle bearing degradation.

[0055] Furthermore, detecting transient fluctuations in acoustic emission signals and / or temperature signals refers to observing whether short-term, drastic changes also occur in these signals within the time window of the impact pulse. For example, tool wear or bearing failure may generate vibration and impact along with a local temperature increase or the generation of high-frequency sound waves. Assessing the temporal alignment between these transient fluctuations and the transient impact pulse aims to improve the reliability of fault diagnosis through cross-validation of multi-sensor information. Based on this, calculating the resonance confidence level, which characterizes the consistency of multi-source signals, involves comprehensively considering factors such as the impact characteristics of the high-frequency vibration signal, the temporal alignment of transient fluctuations in acoustic emission signals and / or temperature signals with the impact pulse, and calculating a quantitative index to evaluate the consistency or synergy of these different types of signals in indicating the same fault event. A higher resonance confidence level indicates a more consistent indication of the same fault event by the multi-source signals, and thus a higher reliability of fault diagnosis.

[0056] This application's solution, by introducing high-frequency vibration signals and combining them with the tool's three-dimensional position and feed rate vector, enables a more in-depth analysis of the operating status of machine tool functional components. Because high-frequency vibration signals are more sensitive to minute impacts and early faults, the system can identify transient impact pulses that are difficult to capture with conventional vibration signals. Simultaneously, by accurately acquiring the tool's three-dimensional position and feed rate vector, the timing of cutting events can be accurately predicted. Synchronizing the analysis of the transient impact pulse's occurrence with the cutting event allows the system to effectively distinguish between impacts caused by normal cutting processes and abnormal impacts caused by component failures (such as tool wear or spindle bearing degradation). This synchronous analysis mechanism, combined with the detection of transient fluctuations in acoustic emission signals and / or temperature signals and the assessment of their time alignment with the impact pulse, achieves cross-validation of multi-source information. Therefore, by calculating the resonance confidence level characterizing the consistency of multi-source signals, this application's solution can confirm the authenticity and source of abnormal events from multiple dimensions, thereby overcoming the potential for misjudgment or omission in single-signal analysis and significantly improving the accuracy and reliability of fault diagnosis.

[0057] In some preferred embodiments, a specific example is given below. Assume a CNC milling machine is machining, and its vibration sensors (especially high-frequency vibration sensors) collect a series of high-frequency vibration signals. Simultaneously, the machine tool's CNC system provides real-time information on the tool's three-dimensional position and feed rate vector. When the system detects one or more transient impact pulses in the high-frequency vibration signals, for example, by performing envelope demodulation on the vibration signals to identify obvious impact characteristics, it records the precise occurrence time of these impact pulses. At the same time, the system uses the tool's three-dimensional position and feed rate vector, combined with the machine tool's geometric model and machining program, to accurately calculate the expected moment when the tool and workpiece will cut, i.e., the cutting event.

[0058] Subsequently, the system synchronously compares the occurrence time of the identified transient impact pulses with the calculated cutting event time. If the occurrence time of an impact pulse closely matches the timing of a cutting event, for example, within a very short time window (e.g., milliseconds), it is preliminarily determined that the impact may be related to an anomaly in the cutting process, such as localized tool chipping or severe wear. If the synchronization between the impact pulse and the cutting event is poor, but its occurrence frequency is related to the spindle's rotational frequency or its harmonic frequencies, the system tends to determine that the impact originates from early degradation of the spindle bearings.

[0059] To further verify this determination, the system also detects whether the acoustic emission sensor also detects high-energy transient acoustic emission signals within the time window of the impact pulse, and whether the temperature sensor detects a momentary increase in local temperature. For example, if tool wear leads to impact, it is usually accompanied by an increase in temperature in the cutting area and high-frequency acoustic emission. The system evaluates the temporal alignment of the transient fluctuations of these acoustic emission signals and / or temperature signals with the vibration impact pulse.

[0060] Ultimately, the system comprehensively considers the synchronicity of vibration and cutting events, their correlation with bearing characteristic frequencies, and the synergy of acoustic emission and temperature signals to calculate a resonance confidence level. For example, if the vibration and cutting events are highly synchronized, and the acoustic emission and temperature signals also exhibit transient fluctuations that align with the vibration and shock, the resonance confidence level will be very high. This confirms that the anomaly is caused by tool wear, and accordingly sends a warning message containing a high failure risk level to the user.

[0061] In some of the embodiments described above in this application, a scheme is proposed to send early warning information, including the fault risk level, to the user and obtain user feedback. However, in its implementation, if user feedback is not effectively utilized to continuously optimize the system's early warning accuracy, false alarms or missed alarms may occur, thereby reducing user trust in the system and affecting the actual application effect. To address this, this application further proposes that after obtaining user feedback, the early warning confidence level of the user feedback is evaluated based on transient impact pulse source information and resonance confidence level, and the upper limit of the normal range of the machine tool's operating performance indicators is dynamically adjusted accordingly to achieve adaptive optimization of the early warning mechanism.

[0062] In this embodiment, the method further includes: evaluating the confidence level of the user-feedback warning based on the transient impact pulse source information and the resonance confidence level; if the warning confidence level is lower than a preset threshold, marking the warning information as a false alarm and increasing the upper limit of the normal range of the machine tool operating performance index; if the warning confidence level is higher than the preset threshold, marking the warning information as a confirmed fault and decreasing the upper limit of the normal range of the machine tool operating performance index.

[0063] Specifically, the warning confidence level refers to the system's assessment of the accuracy of user-reported warning information, aiming to quantify the reliability of the warning. This assessment is based on transient impact pulse source information and resonance confidence level. Transient impact pulse source information provides a preliminary judgment on the source of the anomaly, such as tool wear or spindle bearing degradation, while resonance confidence level characterizes the temporal alignment and consistency of multi-source signals (such as acoustic emission signals, temperature signals, and transient impact pulses), reflecting the authenticity of the abnormal event. By integrating this information, the validity of user feedback can be judged more accurately. The preset threshold is a configurable parameter used to distinguish between high-confidence and low-confidence warnings. When the assessed warning confidence level is lower than the preset threshold, it indicates that the system considers the warning potentially a false alarm, meaning the user's feedback of "no fault" or "false alarm" is credible. In this case, to reduce future false alarms, the system automatically raises the upper limit of the normal range for the machine tool's operating performance indicators, so that under similar operating conditions, the indicators need to deviate more significantly to be considered abnormal. Conversely, when the confidence level of the warning is higher than the preset threshold, it indicates that the system considers the warning to be accurate, meaning that the user's feedback of "fault" or "confirmed fault" is credible. At this time, in order to improve the sensitivity to real faults, the system will automatically lower the upper limit of the normal range of the machine tool's operating performance indicators, making it easier to judge as abnormal under similar working conditions, even if the deviation of the indicators is small.

[0064] This application's solution effectively addresses the problem of insufficient utilization of user feedback in the basic solution by introducing a warning confidence assessment mechanism. Specifically, after the system sends a warning and receives user feedback, it doesn't simply accept the feedback result. Instead, it combines the transient impact pulse source information identified through high-frequency vibration signals and the resonance confidence calculated based on the consistency of multi-source signals to perform a secondary assessment of the authenticity of the user feedback, thus obtaining the warning confidence. For example, if the user reports "no fault," but the transient impact pulse source information detected by the system clearly points to spindle bearing degradation, and the resonance confidence is high, this may mean that the user failed to detect the problem in time or made a mistake in judgment; in this case, the warning confidence will be relatively low. Conversely, if the user reports "no fault," and the transient impact pulse source information is unclear, and the resonance confidence is also low, then the warning confidence will be high. It is precisely because of this mechanism of intelligently assessing user feedback based on multi-source information that the system can distinguish between real faults and false alarms, and adaptively adjust the upper limit of the normal range of indicators accordingly. When the confidence level of the warning is low, the system will raise the upper limit of the normal range of the indicator to avoid repeated false alarms; when the confidence level of the warning is high, the system will lower the upper limit of the normal range of the indicator to improve the sensitivity to real faults.

[0065] Through the above technical solution, this application enables adaptive optimization of the performance analysis method for machine tool functional components. By introducing a warning confidence assessment mechanism, the system no longer passively accepts user feedback but can intelligently judge the reliability of user feedback and dynamically adjust the warning threshold accordingly. This significantly improves the accuracy and reliability of warnings, effectively reduces false alarms and missed alarms, thereby increasing user trust in the system. Furthermore, this self-learning and adaptive capability allows the system to better adapt to the operating characteristics and wear patterns of machine tools under different working conditions, extending the service life of machine tool functional components and reducing maintenance costs.

[0066] In some preferred embodiments, a specific example is given below. Suppose that during operation, the vibration index of a machine tool suddenly exceeds the normal range for the current operating conditions. Based on this, the system determines that a preliminary anomaly exists, further constructs multi-dimensional anomaly information, and ultimately determines the fault risk level to be "medium risk," sending a warning message to the user. After receiving the warning, the user inspects the machine tool but finds no obvious fault, thus reporting "no fault." At this point, the system initiates the warning confidence assessment process. First, the system reviews the transient impact pulse source information identified through high-frequency vibration signals at the time of the anomaly. For example, a weak impact pulse may be identified, but its source information is determined to be "uncertain." Simultaneously, the system checks the time alignment between the transient fluctuations of the acoustic emission signal and temperature signal and the impact pulse, calculating the resonance confidence. Suppose the resonance confidence is low, indicating weak consistency among the multi-source signals. Based on the factors of "uncertain transient impact pulse source information" and "low resonance confidence," the system assesses the user's warning confidence as 0.3. Because 0.3 is below a preset threshold (e.g., the preset threshold is 0.5), the system marks this warning as a "false alarm." To prevent false alarms from recurring in similar situations, the system automatically increases the upper limit of the normal range for this vibration index, for example, from 1.5g to 1.8g. This means that in subsequent operations, the vibration index needs to reach a higher value before the system will judge it as abnormal, thus reducing the possibility of false alarms. Conversely, if the user reports a "fault," and the system's assessed warning confidence level is 0.8 (higher than the preset threshold of 0.5), the system will mark this warning as a "confirmed fault." To improve sensitivity to actual faults, the system automatically decreases the upper limit of the normal range for this vibration index, for example, from 1.5g to 1.3g. In this way, even if the deviation of the vibration index is small in the future, the system can issue a warning earlier, thereby achieving more timely fault detection and handling.

[0067] This application further proposes to identify transient impact pulses from the high-frequency vibration signal while recording the occurrence time and energy characteristics of each transient impact pulse; the above-mentioned synchronous analysis of the occurrence time of each impact pulse with the cutting event determined based on the tool's three-dimensional position and feed rate vector to determine the source information of the transient impact pulse includes: synchronously comparing the occurrence time of the transient impact pulse with the cutting occurrence time calculated based on the tool's three-dimensional position and feed rate vector to obtain a synchronous comparison result; simultaneously, analyzing the instantaneous current response of the motor current signal before and after the occurrence time of the impact pulse; based on the synchronous comparison result and the instantaneous current response, calculating the correlation score of the transient impact pulse originating from tool wear or spindle bearing degradation, and determining the source of the anomaly.

[0068] Specifically, when identifying transient impact pulses from high-frequency vibration signals, in addition to recording their occurrence time, the energy characteristics of each transient impact pulse are also recorded. These energy characteristics can be understood as the intensity or amplitude of the impact pulse, such as the peak value, root mean square value, or energy integral within a specific time window. The purpose is to quantify the severity of the impact and provide richer information for subsequent fault diagnosis. Specifically, the occurrence time of the transient impact pulse is synchronously compared with the cutting occurrence time calculated based on the tool's three-dimensional position and feed rate vector to determine whether the impact pulse is directly related to the cutting process. The synchronous comparison result can be a time difference or a correlation index, used to characterize the degree of temporal agreement between the impact pulse and the cutting event. In practical applications, analyzing the instantaneous current response of the motor current signal before and after the occurrence of the impact pulse refers to monitoring the instantaneous changes in the motor current before and after the impact pulse. For example, when the tool is worn or the spindle bearing is faulty, it may cause instantaneous fluctuations in the motor load, thus exhibiting specific instantaneous response characteristics in the current signal. The purpose is to provide auxiliary diagnostic information independent of the vibration signal and enhance the reliability of the diagnosis. Furthermore, based on the synchronous comparison results and the instantaneous current response, a correlation score is calculated indicating whether the transient impact pulse originates from tool wear or spindle bearing degradation. This correlation score can be a quantitative indicator used to assess the correlation between the impact pulse and a specific fault type (such as tool wear or spindle bearing degradation). By comparing the correlation scores for different fault types, the source of the anomaly can be determined more accurately.

[0069] This application's solution effectively addresses the potential accuracy and robustness deficiencies of the aforementioned basic schemes in determining transient impact pulse source information by introducing energy characteristic recording of transient impact pulses, instantaneous response analysis of motor current signals, and correlation score calculation. Specifically, recording the energy characteristics of the impact pulse allows for a more comprehensive assessment of the impact severity, going beyond just the timing of its occurrence. Simultaneously, by synchronously comparing the timing of the impact pulse with the cutting event, a preliminary determination can be made as to whether the impact is directly related to the cutting process. More importantly, the introduction of instantaneous response analysis of the motor current signal before and after the impact pulse occurrence provides multi-dimensional data support for fault diagnosis. Tool wear and spindle bearing degradation often produce different instantaneous load changes in the motor current; analyzing these current responses provides crucial auxiliary evidence for distinguishing between these two types of faults. Finally, based on the synchronous comparison results and the instantaneous current response, the calculated correlation score quantifies the correlation between the impact pulse and different fault sources, thereby achieving accurate determination of the anomaly source. This multi-source information fusion and quantitative evaluation method eliminates reliance on single indicators in fault diagnosis, significantly improving the accuracy and reliability of the diagnosis.

[0070] In some preferred embodiments, a specific example is illustrated below. Suppose that during machine tool operation, a high-frequency vibration sensor detects a transient impact pulse. First, the system records the occurrence time of the impact pulse and its energy characteristics, such as a peak amplitude of XG and a duration of Y ms. Simultaneously, the system calculates whether a cutting event exists before and after the occurrence time of the impact pulse based on the machine tool's three-dimensional tool position and feed rate vector. If the occurrence time of the impact pulse is found to be highly synchronized with a cutting event, it is preliminarily determined that it may be related to the cutting process. Further, the system analyzes the instantaneous response of the motor current signal before and after the occurrence time of the impact pulse. For example, if a brief, sharp rise or fall in the current signal is observed during the occurrence of the impact pulse, this may indicate an instantaneous change in the motor load. Subsequently, based on the synchronization comparison results of the impact pulse and the cutting event (e.g., time alignment score) and the characteristics of the instantaneous response of the motor current (e.g., current fluctuation amplitude or duration), the system calculates two correlation scores: one characterizing that the impact pulse originates from tool wear, and the other characterizing that the impact pulse originates from spindle bearing degradation. For example, if the impact pulse is highly synchronized with the cutting event and the current response characteristics match a known tool wear mode, the correlation score for tool wear will be high. Conversely, if the impact pulse is less synchronized with the cutting event, but its frequency components match the characteristic frequency of the spindle bearing and the current response characteristics match a known bearing failure mode, the correlation score for spindle bearing degradation will be high. By comparing these two correlation scores, the system ultimately determines whether the abnormal source of the transient impact pulse is tool wear or spindle bearing degradation. For example, if the tool wear correlation score is much higher than the spindle bearing degradation correlation score, it is determined to be tool wear.

[0071] This application further proposes that the aforementioned correlation score includes a first correlation score for characterizing the correlation between the transient impact pulse and the cutting process; and a second correlation score for characterizing the correlation between the transient impact pulse and the bearing failure frequency; if the first correlation score is higher than a first threshold, it is determined that the transient impact pulse originates from tool wear; if the first correlation score is lower than the first threshold and the second correlation score is higher than the second threshold, it is determined that the transient impact pulse originates from spindle bearing degradation.

[0072] Specifically, the correlation score is refined into two independent indicators: a first correlation score and a second correlation score. The first correlation score aims to quantify the correlation between transient impact pulses and the cutting process. For example, it can be calculated by analyzing the synchronicity between the occurrence time of the impact pulse and the tool cutting event, and the relationship between the energy characteristics of the impact pulse and changes in cutting force. Its purpose is to identify impact events directly related to the cutting operation, which are usually closely related to changes in tool condition such as wear and chipping. Furthermore, the second correlation score is used to characterize the correlation between transient impact pulses and spindle bearing failure frequencies. For example, it can be used to perform spectral analysis on high-frequency vibration signals to identify whether there are impact pulse frequency components corresponding to the inherent failure frequencies of the bearing (such as outer ring failure frequency, inner ring failure frequency, rolling element failure frequency, or cage failure frequency), and to calculate their energy proportion or amplitude. Its purpose is to identify periodic impacts caused by internal defects in the spindle bearing (such as raceway spalling, rolling element damage, etc.).

[0073] In practical applications, to achieve accurate fault source identification, a first threshold and a second threshold are set. The first threshold is a preset value used to determine whether the first correlation score is high enough to indicate that the impact pulse mainly originates from tool wear. The second threshold is another preset value used to determine whether, when the first correlation score is insufficient to indicate tool wear, the second correlation score is high enough to indicate that the impact pulse mainly originates from spindle bearing degradation. These thresholds can be calibrated and optimized based on historical data, expert experience, or experimental testing.

[0074] This application's solution effectively addresses the problem of accurately distinguishing different fault sources by decomposing a single correlation score into two correlation scores with clear physical meanings and introducing threshold-based decision logic. Specifically, the first correlation score focuses on the correlation between transient impact pulses and the cutting process. When its value is higher than a first threshold, it indicates that the impact pulse is highly synchronized with the tool's cutting activity and their characteristics match, thus directly pointing to tool wear as the primary fault source. This utilizes the principle that tool wear generates specific impact characteristics during the cutting process. On the other hand, when the first correlation score is insufficient to indicate tool wear, the system further evaluates the second correlation score. The second correlation score focuses on the correlation between transient impact pulses and the spindle bearing failure frequency. When its value is higher than a second threshold, it indicates that the impact pulse has periodic or spectral characteristics consistent with the bearing failure frequency, thus locking the fault source to spindle bearing degradation. This hierarchical and targeted judgment mechanism allows the system to more precisely capture the unique fingerprints of different fault modes, avoiding the confusion that may be caused by a single indicator, and significantly improving the accuracy and reliability of fault source determination.

[0075] In some preferred embodiments, it is assumed that a transient impact pulse is detected in a high-frequency vibration signal during machine tool operation. The system first calculates the correlation between the impact pulse and the cutting process, obtaining a first correlation score. For example, if the calculated first correlation score is 0.85, and the preset first threshold is 0.70, since 0.85 is higher than 0.70, the system will determine that the transient impact pulse mainly originates from tool wear. This means that the occurrence of the impact pulse is highly synchronized with the cutting event, and its energy characteristics are consistent with the impact characteristics generated by tool wear during the cutting process. As another specific implementation, it is assumed that the calculated first correlation score is 0.60, which is lower than the preset first threshold of 0.70. In this case, the system will further calculate the correlation between the impact pulse and the spindle bearing failure frequency, obtaining a second correlation score. If the calculated second correlation score is 0.92, and the preset second threshold is 0.80, since 0.60 is lower than 0.70 and 0.92 is higher than 0.80, the system will determine that the transient impact pulse mainly originates from spindle bearing degradation. This indicates that although the impact pulse is not highly correlated with the cutting process, its spectral analysis results show a high degree of consistency with the specific failure frequency of the spindle bearing, thus indicating that the bearing is degraded.

[0076] Furthermore, this application also proposes a machine tool functional component performance analysis system based on big data, such as... Figure 2 As shown, the system includes: The acquisition module 201 is used to acquire multi-source operating information of the machine tool's functional components; the multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; The analysis and judgment module 202 is used to determine the machine tool operating performance indicators based on the multi-source operating information. If any machine tool operating performance indicator exceeds the normal range of the indicator under the current working condition, it is judged that there is a preliminary abnormality in the machine tool functional components. Construction module 203 is used to construct multi-dimensional abnormality information based on the working condition information, vibration information, and temperature information in response to the initial abnormality of the machine tool functional components; The feedback module 204 determines the fault risk level of the machine tool functional components based on the multi-dimensional anomaly information, sends early warning information including the fault risk level to the user, and obtains user feedback.

[0077] This system, through the collaborative operation of its internal acquisition, analysis and judgment, construction, and feedback modules, achieves comprehensive monitoring and intelligent early warning of the performance of machine tool functional components. The acquisition module is responsible for acquiring multi-source operational information, providing a data foundation for subsequent analysis; the analysis and judgment module evaluates performance indicators and identifies initial anomalies based on this information; the construction module integrates multi-dimensional information to deeply analyze the nature of the anomaly when it occurs; finally, the feedback module determines the fault risk level based on the analysis results and interacts with the user. This application aims to improve the accuracy of anomaly judgment for machine tool functional components and reduce false alarms through systematic multi-source data fusion and multi-dimensional anomaly information construction, thereby enhancing the reliability of the early warning system and user trust.

[0078] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for performance analysis of machine tool functional components based on big data, characterized in that, The method includes: Collect multi-source operating information of machine tool functional components; the multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; Based on the multi-source operating information, the machine tool operating performance indicators are determined. If any machine tool operating performance indicator exceeds the normal range of the indicator under the current working condition, it is determined that there is a preliminary abnormality in the machine tool functional components. In response to the presence of a preliminary abnormality in the machine tool's functional components, multi-dimensional abnormality information is constructed based on the operating condition information, vibration information, and temperature information. Based on the multidimensional anomaly information, the fault risk level of the machine tool functional components is determined, and a warning message including the fault risk level is sent to the user, and user feedback is obtained.

2. The method for performance analysis of machine tool functional components based on big data according to claim 1, characterized in that, The machine tool operating performance indicators include vibration indicators, temperature indicators, current indicators, and acoustic emission indicators; wherein, the vibration indicators include the root mean square value, peak factor, kurtosis, or energy percentage within a specific frequency range of the vibration signal; the temperature indicators include the real-time temperature value or rate of change of the temperature signal; the current indicators include the harmonic content, fluctuation amplitude, or specific frequency component energy of the motor current signal; and the acoustic emission indicators include the root mean square value, count rate, or energy of the acoustic emission signal.

3. The method for performance analysis of machine tool functional components based on big data according to claim 2, characterized in that, If any machine tool operating performance index exceeds the normal range under current operating conditions, it is determined that there is a preliminary abnormality in the machine tool's functional components, including: Determine if a machine tool functional component is abnormal, and record the time of occurrence, type of abnormal indicator, and degree of deviation of the abnormal indicator; the degree of deviation of the abnormal indicator includes at least one of vibration abnormality, temperature abnormality, and acoustic emission abnormality.

4. The method for performance analysis of machine tool functional components based on big data according to claim 3, characterized in that, The multidimensional anomaly information includes operating condition correlation information and cross-validation information from multiple sensors; the construction of multidimensional anomaly information based on the operating condition information, vibration information, and temperature information includes: The current operating condition information is compared with the preset operating condition feature library, and the degree of matching between the current abnormal data and the preset feature patterns in the preset operating condition feature library is calculated to determine the operating condition correlation information; wherein, the preset operating condition feature library stores preset feature patterns caused by known causes under different operating conditions. Examine whether the data from multiple types of sensors show a correlated abnormal trend within the time window of the initial anomaly occurrence, assign weights to the degree of anomaly of different sensor signals and their consistency, and calculate a multi-sensor consistency score.

5. The method for performance analysis of machine tool functional components based on big data according to claim 4, characterized in that, Determining the fault risk level of the machine tool functional components based on the multidimensional anomaly information includes: The fault risk level is determined by weighted summation of the working condition correlation information, the multi-sensor consistency score, and the degree of deviation of abnormal indicators.

6. The method for performance analysis of machine tool functional components based on big data according to claim 1, characterized in that, The vibration signal is a high-frequency vibration signal; the multi-source operating information also includes the three-dimensional position of the machine tool and the feed speed vector; The method further includes: identifying transient impact pulses from the high-frequency vibration signal, and synchronously analyzing the occurrence time of each impact pulse with the cutting event determined based on the tool's three-dimensional position and feed rate vector to determine the transient impact pulse source information; the transient impact pulse source information includes tool wear or spindle bearing degradation; The transient fluctuations of the acoustic emission signal and / or the temperature signal are detected, the temporal alignment of the transient fluctuations with the transient impact pulse is evaluated, and the resonance confidence level characterizing the consistency of the multi-source signals is calculated.

7. The method for performance analysis of machine tool functional components based on big data according to claim 6, characterized in that, Sending warning information including the fault risk level to the user and obtaining user feedback, the method further includes: The confidence level of the warning feedback from the user is evaluated based on the transient impact pulse source information and the resonance confidence level. If the confidence level of the warning is lower than the preset threshold, the warning information is marked as a false alarm, and the upper limit of the normal range of the corresponding machine tool operating performance index is increased; if the confidence level of the warning is higher than the preset threshold, the warning information is marked as a confirmed fault, and the upper limit of the normal range of the corresponding machine tool operating performance index is decreased.

8. The method for performance analysis of machine tool functional components based on big data according to claim 6, characterized in that, While identifying transient impact pulses from the high-frequency vibration signal, the occurrence time and energy characteristics of each transient impact pulse are recorded; The step of synchronously analyzing the occurrence time of each impact pulse with the cutting event determined based on the tool's three-dimensional position and feed rate vector to determine the transient impact pulse source information includes: The occurrence time of the transient impact pulse is synchronously compared with the cutting occurrence time calculated based on the tool's three-dimensional position and feed rate vector to obtain the synchronous comparison result; at the same time, the instantaneous current response of the motor current signal before and after the occurrence time of the impact pulse is analyzed. Based on the synchronous comparison results and the instantaneous current response, the correlation score of the transient impact pulse originating from tool wear or spindle bearing degradation is calculated, and the source of the anomaly is determined.

9. The method for performance analysis of machine tool functional components based on big data according to claim 8, characterized in that, The correlation score includes a first correlation score characterizing the correlation between the transient impact pulse and the cutting process; and a second correlation score characterizing the correlation between the transient impact pulse and the bearing failure frequency. If the first correlation score is higher than the first threshold, the transient impact pulse is determined to originate from tool wear; if the first correlation score is lower than the first threshold and the second correlation score is higher than the second threshold, the transient impact pulse is determined to originate from spindle bearing degradation.

10. A machine tool functional component performance analysis system based on big data, characterized in that, The system includes: The acquisition module is used to acquire multi-source operating information of the machine tool's functional components; the multi-source operating information includes vibration signals, temperature signals, motor current signals, acoustic emission signals, and operating condition information; The analysis and judgment module is used to determine the machine tool's operating performance indicators based on the multi-source operating information. If any machine tool operating performance indicator exceeds the normal range of the indicator under the current working condition, it is judged that there is a preliminary abnormality in the machine tool's functional components. A construction module is used to construct multi-dimensional anomaly information based on the working condition information, vibration information, and temperature information in response to the presence of a preliminary anomaly in the machine tool functional components. The feedback module determines the fault risk level of the machine tool functional components based on the multi-dimensional anomaly information, sends early warning information including the fault risk level to the user, and obtains user feedback.

Images