A precision machine tool electric spindle health monitoring and early warning method

Abstract

The application provides a precision machine tool electric spindle health monitoring and early warning method, and relates to the field of equipment state monitoring.The method comprises the following steps: correcting standard parameter early warning indexes according to expected working condition information to determine real-time parameter early warning indexes; continuously monitoring parameters of a plurality of electric spindles in the same batch to obtain a plurality of parameter sequence sets and determine a plurality of first early warning parameter sets meeting the real-time parameter early warning indexes; constructing a plurality of parameter curve sets for abnormality identification to determine a plurality of second early warning parameter sets; and analyzing and determining a plurality of early warning electric spindles according to the plurality of first early warning parameter sets and the plurality of second early warning parameter sets.The method aims to solve the technical problem that the traditional monitoring method fails to fully consider the dynamic change characteristics of working conditions in the actual machining process and simultaneously lacks a collaborative abnormality identification mechanism, resulting in low early warning accuracy and reliability.The method can realize a high-reliability and high-confidence health monitoring and early warning mechanism and significantly improve the accuracy and reliability of fault early warning.

Description

Technical Field

[0001] This invention relates to the field of equipment condition monitoring, and in particular to a method for health monitoring and early warning of electric spindles in precision machine tools. Background Technology

[0002] As a core component of precision CNC machine tools, the stability of the electric spindle's operating status is directly related to the machining quality and equipment lifespan. Therefore, higher requirements are placed on the health monitoring and fault early warning of the electric spindle.

[0003] Currently, commonly used electric spindle health monitoring methods are mainly based on real-time acquisition of key parameters such as vibration, temperature, current, and speed, and identification of abnormal states by comparing them with fixed thresholds. By setting parameter thresholds, rapid early warning of obvious electric spindle faults can be achieved, thereby reducing the risk of equipment damage and downtime losses to a certain extent.

[0004] However, in actual machining processes, the operating conditions of the electric spindle (such as load status, machining cycle time, tool type, and material properties) have significant dynamic changes. Fixed thresholds are difficult to adapt to the normal fluctuation range of parameters under different operating conditions, which can easily lead to false alarms or missed alarms. In addition, traditional methods often rely solely on the judgment of a single parameter exceeding the limit, without establishing a collaborative analysis mechanism between multiple parameters. This makes it impossible to identify potential combined anomalies or trend degradation problems, resulting in low accuracy and reliability of early warnings and a tendency to cause false alarms and missed alarms. Consequently, these methods cannot meet the high reliability requirements for monitoring the operating status of electric spindles in high-precision machining scenarios. Summary of the Invention

[0005] The purpose of this invention is to provide a health monitoring and early warning method for precision machine tool electric spindles. This method addresses the technical problems of traditional electric spindle health monitoring methods, which primarily rely on fixed parameter thresholds for fault warning, failing to fully consider the dynamic changes in operating conditions during actual machining, and lacking a collaborative anomaly identification mechanism, resulting in low accuracy and reliability of early warnings. The method includes: acquiring expected operating condition information of the electric spindle within a preset time zone; correcting standard parameter early warning indicators based on the expected operating condition information to determine real-time parameter early warning indicators; continuously monitoring the parameters of several electric spindles processed in the same batch in the machining workshop within the preset time zone, acquiring several parameter sequence sets, and determining multiple first early warning parameter sets that satisfy the real-time parameter early warning indicators; constructing several parameter curve sets based on the several parameter sequence sets, performing anomaly identification on the several parameter curve sets, and determining multiple second early warning parameter sets; analyzing and determining multiple early warning electric spindles based on the multiple first early warning parameter sets and the multiple second early warning parameter sets, wherein any parameter of the early warning electric spindle simultaneously satisfies both the first and second early warning indicators.

[0006] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: the working condition information includes spindle operating parameters, tool characteristic parameters, feed motion parameters, workpiece attribute information, continuous machining duration, and machining environment parameters.

[0007] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: using the expected operating condition information as a condition constraint and the electric spindle as a device constraint, performing fault ratio statistics based on big data to determine the fault ratio of sample equipment; setting the ratio of the fault ratio of the sample equipment to the average historical equipment fault ratio under standard operating conditions as a fault scale coefficient; determining the parameter adjustment direction and parameter adjustment ratio according to the fault scale coefficient, expanding the standard parameter early warning index to determine the parameter early warning threshold; optimizing the parameter early warning index within the parameter early warning threshold, and outputting real-time parameter early warning index.

[0008] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: the standard parameter early warning indicators include at least standard early warning values ​​for vibration parameters, electrical parameters, temperature parameters and noise parameters.

[0009] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: randomly selecting a first parameter early warning index within the parameter early warning threshold; using a prediction error analysis model, performing prediction error analysis based on the expected working condition information and the first parameter early warning index, and outputting a first prediction error; continuing to randomly select parameter early warning indexes and perform prediction error analysis within the parameter early warning threshold until a preset number of times is reached, and setting the parameter early warning index with the minimum prediction error as the real-time parameter early warning index, wherein the preset number of times is positively correlated with the fault scale coefficient.

[0010] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: collecting a sample working condition information set and a sample parameter early warning index set based on historical monitoring records of similar electric spindles, and statistically analyzing the average proportion of fault prediction errors of similar electric spindles under different sample working condition information and sample parameter early warning indices, setting this as the sample prediction error, and obtaining a sample prediction error set; using the sample working condition information set, sample parameter early warning index set, and sample prediction error set as training data, supervising the training of the BP neural network until the network converges, thereby obtaining the prediction error analysis model.

[0011] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: judging the plurality of parameter sequence sets according to the real-time parameter early warning index; if the parameter value does not meet the corresponding parameter early warning index, then marking an early warning trigger; counting the number of early warning triggers within a preset time window; if the number of early warning triggers is greater than a preset threshold value, then setting the monitoring parameter as a first early warning parameter, thereby obtaining a plurality of first early warning parameter sets.

[0012] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: distributing the plurality of parameter sequence sets in a two-dimensional coordinate system according to the monitoring time from early to late to generate a plurality of parameter curve sets; clustering the plurality of parameter curve sets with the same parameters to obtain multiple sets of parameter curves of the same type; and identifying anomalies in the multiple sets of parameter curves of the same type to determine multiple sets of second early warning parameters.

[0013] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: randomly selecting any set of similar parameter curves, and randomly selecting a first similar parameter curve from the set of similar parameter curves; performing a similarity comparison between the first similar parameter curve and other similar parameter curves in the set of similar parameter curves to obtain multiple similarities; counting the number of curves with similarities less than a preset similarity threshold; if the number of curves is greater than a preset number threshold, then marking the first similar parameter curve as a second early warning parameter, and sequentially analyzing to obtain multiple sets of second early warning parameters.

[0014] Preferably, the method for health monitoring and early warning of a precision machine tool electric spindle further includes: randomly selecting a first electric spindle; if the first early warning parameter set and the second early warning parameter set corresponding to the first electric spindle are not empty, and any parameter simultaneously meets the first early warning parameter and the second early warning parameter, then the first electric spindle is marked as an early warning electric spindle.

[0015] The embodiments of the present invention have the following advantages: By acquiring the expected operating condition information of the electric spindle within a preset time zone, the standard parameter warning indicators are corrected based on the expected operating condition information to determine the real-time parameter warning indicators. Then, within the preset time zone, parameters of several electric spindles processed in the same batch in the machining workshop are continuously monitored to acquire several parameter sequence sets, and multiple first warning parameter sets that meet the real-time parameter warning indicators are determined. Then, several parameter curve sets are constructed based on the several parameter sequence sets, and anomaly identification is performed on the several parameter curve sets to determine multiple second warning parameter sets. Finally, multiple warning electric spindles are determined by analyzing the multiple first warning parameter sets and multiple second warning parameter sets, wherein any parameter of the warning electric spindle simultaneously meets the first warning and the second warning. In other words, by introducing an adaptive threshold adjustment mechanism based on operating parameters, the warning range of each monitoring parameter can be dynamically corrected, making the health assessment more consistent with the actual processing state, thereby effectively reducing the risk of false alarms and missed alarms caused by fixed threshold settings. By constructing parameter time series curves and using trend analysis and multi-parameter collaborative anomaly identification methods, chronic degradation and complex abnormal behaviors that are difficult to detect by traditional static methods can be identified, thereby improving the sensitivity and accuracy of the warning. Finally, by integrating the results of the first and second warnings, the final warning is triggered only when the parameter meets both warning conditions simultaneously, thereby achieving a highly reliable and high-confidence health monitoring and warning mechanism, significantly improving the accuracy and reliability of fault identification of precision machine tool electric spindles. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of a precision machine tool electric spindle health monitoring and early warning method according to the present invention. Figure 2 This is a flowchart illustrating the steps for determining real-time parameter warning indicators in a health monitoring and early warning method for a precision machine tool electric spindle according to the present invention. Detailed Implementation

[0017] This invention provides a health monitoring and early warning method for precision machine tool electric spindles. It addresses the shortcomings of traditional electric spindle health monitoring methods, which rely primarily on fixed parameter thresholds for fault warnings, failing to fully consider the dynamic changes in operating conditions during actual machining. Furthermore, the lack of a collaborative anomaly identification mechanism results in low accuracy and reliability of early warnings. By introducing an adaptive threshold adjustment mechanism based on operating parameters, the warning range of each monitoring parameter can be dynamically corrected, making the health assessment more closely aligned with the actual machining state. This effectively reduces the risk of false alarms and missed alarms caused by fixed threshold settings. By constructing parameter time series curves and employing trend analysis and multi-parameter collaborative anomaly identification methods, chronic degradation and complex abnormal behaviors that are difficult to detect using traditional static methods can be identified, thereby improving the sensitivity and accuracy of early warnings. Finally, by fusing the results of the first and second early warnings, the final early warning is triggered only when the parameters simultaneously meet both warning conditions. This achieves a highly reliable and high-confidence health monitoring and early warning mechanism, significantly improving the accuracy and reliability of fault identification for precision machine tool electric spindles.

[0018] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0019] For examples, please refer to the appendix. Figure 1 This invention provides a method for health monitoring and early warning of a precision machine tool electric spindle, specifically including the following steps: S10: Obtain the expected operating condition information of the electric spindle within the preset time zone, correct the standard parameter warning indicators based on the expected operating condition information, and determine the real-time parameter warning indicators.

[0020] Furthermore, step S10 of the present invention further includes: S11: The working condition information includes spindle operating parameters, tool characteristic parameters, feed motion parameters, workpiece attribute information, continuous machining duration, and machining environment parameters.

[0021] Specifically, firstly, the expected operating condition information of the electric spindle within a preset time zone (e.g., within the next 3 minutes) is obtained. This operating condition information includes spindle operating parameters, tool characteristic parameters, feed motion parameters, workpiece attribute information, continuous machining duration, and machining environment parameters. Among these, the spindle operating parameters include key monitoring indicators such as spindle speed, spindle load, current, and vibration, reflecting the operating status of the electric spindle in the current working state; tool characteristic parameters include tool type, wear state, tool material, and size, affecting cutting force and machining accuracy; feed motion parameters include feed speed, feed amount, and feed direction, affecting the cutting force distribution and spindle load during machining; workpiece attribute information includes workpiece material, size, and surface hardness, directly affecting the cutting state and spindle load during machining; continuous machining duration reflects the cumulative operating time of the spindle in the current machining task and is closely related to spindle wear and performance degradation; machining environment parameters include external environmental conditions such as temperature, humidity, and noise, which affect the working performance and probability of failure of the electric spindle. By acquiring this operating condition information, we can gain a comprehensive understanding of the current and future working status of the electric spindle, thus providing crucial information for subsequent health monitoring, early warning, and fault diagnosis.

[0022] Furthermore, such as Figure 2 As shown, step S10 of the present invention further includes: S12: Using the expected working condition information as a condition constraint and the electric spindle as a device constraint, perform fault ratio statistics based on big data to determine the fault ratio of the sample equipment; S13: Set the ratio of the fault ratio of the sample equipment to the average historical equipment fault ratio under standard working conditions as the fault scale coefficient.

[0023] Specifically, firstly, using the expected operating condition information as a constraint and the electric spindle as a device constraint, big data analytics is used to statistically analyze fault data under similar operating conditions. By analyzing a large amount of historical data, the failure rate of the electric spindle under different operating conditions can be determined. This data reflects the frequency and pattern of equipment failures under specific operating conditions, thereby calculating the sample equipment failure rate, i.e., the ratio of electric spindle failures under certain time or conditions. Next, the average historical equipment failure rate under standard operating conditions is obtained. The average historical equipment failure rate refers to the performance of the equipment under normal or typical working conditions, usually calculated from a large amount of historical data, and represents the normal failure frequency of the equipment under ideal operating conditions. The ratio of the sample equipment failure rate to the historical average failure rate under standard operating conditions is then set as the failure scale coefficient. A failure scale coefficient greater than 1 indicates that the equipment failure rate under the current expected operating conditions is higher than the historical level under standard operating conditions, potentially meaning the operating conditions are unsuitable or the equipment has potential problems. A failure scale coefficient equal to 1 indicates that the failure rate under the current operating conditions is comparable to that under standard operating conditions. A failure scale coefficient less than 1 indicates that the equipment failure rate under the current operating conditions is lower than that under standard operating conditions, and the equipment is in relatively good condition. By calculating and analyzing the failure scale coefficient, the health status of the equipment can be quantitatively assessed, providing a basis for subsequent early warning and decision-making.

[0024] S14: Determine the parameter adjustment direction and parameter adjustment ratio based on the fault scale coefficient, expand the standard parameter early warning index, and determine the parameter early warning threshold.

[0025] Furthermore, step S14 of the present invention also includes: S141: The standard parameter warning indicators shall include at least the standard warning values ​​for vibration parameters, electrical parameters, temperature parameters and noise parameters.

[0026] Specifically, standard parameter early warning indicators are obtained. These standard parameter early warning indicators refer to early warning reference benchmarks for various monitoring parameters formed through long-term operational data statistics and analysis under standard operating conditions, used to initially identify abnormal equipment states. These indicators include at least standard early warning values ​​for vibration, temperature, and noise parameters. The standard early warning values ​​for vibration parameters include radial vibration, axial vibration, acceleration amplitude, and spectral characteristics of the spindle, reflecting the spindle's dynamic balance and bearing wear. The standard early warning values ​​for electrical parameters include the spindle drive current. Voltage, power factor, etc., are used to assess the operating load of the electric spindle motor and the status of the electronic control system; standard warning values ​​for temperature parameters include spindle body temperature, bearing temperature, motor winding temperature, etc., reflecting the thermal load level and cooling effect of the electric spindle during operation; standard warning values ​​for noise parameters include overall sound pressure level or noise intensity in a specific frequency band, used to detect problems such as internal mechanical wear, loosening, or abnormal friction; the above standard warning values ​​are usually obtained by extracting statistical features (such as mean, standard deviation, distribution range, etc.) from a large amount of healthy operating data or by empirical judgment, serving as the basic threshold input for subsequent dynamic adjustment and warning triggering.

[0027] Next, the parameter adjustment direction and adjustment ratio are determined based on the fault scale coefficient. When the fault scale coefficient is greater than 1 (indicating a higher probability of failure under current operating conditions than under standard operating conditions), it indicates an increased risk, requiring increased vigilance and tightening of the warning threshold (i.e., shrinking towards the normal value to make the warning more sensitive). When the fault scale coefficient is less than 1 (indicating a lower probability of failure under current operating conditions than under standard operating conditions), the threshold can be appropriately relaxed to expand the warning range and avoid false alarms. When the fault scale coefficient is close to 1, the original threshold is maintained or finely adjusted. Simultaneously, the fault scale coefficient can be mapped to a specific adjustment ratio coefficient α by setting an adjustment function (such as a linear function, exponential function, or piecewise function), and then each standard warning threshold is adjusted. Furthermore, the standard parameter warning indicators are expanded according to the parameter adjustment direction and adjustment ratio, that is, the standard warning indicators for each type of health monitoring parameter (such as vibration, electrical, temperature, noise, etc.) are proportionally expanded to form parameter warning thresholds under current operating conditions, adapting to the current processing conditions and fault risk level. The implementation of this step enables dynamic adaptive updates of parameter thresholds under different operating conditions, giving the early warning mechanism the ability to perceive processing risks. By associating it with the scale and degree of fault risk, the early warning system can not only improve the sensitivity of anomaly detection, but also reduce false alarms caused by differences in operating conditions, thereby improving the robustness and practicality of the monitoring system.

[0028] S15: Optimize the parameter warning index within the parameter warning threshold and output the real-time parameter warning index.

[0029] Furthermore, step S15 of the present invention also includes: S151: Randomly select the first parameter warning indicator within the parameter warning threshold.

[0030] S152: Using the prediction error analysis model, perform prediction error analysis based on the expected working condition information and the first parameter warning index, and output the first prediction error.

[0031] Furthermore, step S152 of the present invention further includes: S1521: Based on historical monitoring records of similar electric spindles, collect sample operating condition information sets and sample parameter early warning index sets, and statistically analyze the average proportion of fault prediction errors of similar electric spindles under different sample operating condition information and sample parameter early warning indexes, set as sample prediction error, and obtain sample prediction error set; S1522: Using the sample operating condition information set, sample parameter early warning index set, and sample prediction error set as training data, supervise the training of the BP neural network until the network converges, and obtain the prediction error analysis model.

[0032] Specifically, firstly, any parameter within the parameter warning threshold is randomly selected as the first parameter warning index, including random warning values ​​for vibration parameters, electrical parameters, temperature parameters, and noise parameters. That is, for any of the above parameters, within its parameter warning threshold range after dynamic correction by the fault scale coefficient, a specific value is randomly set as the first warning reference value for that parameter, denoted as the first parameter warning index. Next, a prediction error analysis model is constructed based on a BP neural network. Using this model, prediction error analysis is performed based on the expected operating condition information and the first parameter warning index, and the first prediction error is output.

[0033] The method for constructing the prediction error analysis model is as follows: First, based on historical operating data in the workshop, monitoring records of similar electric spindles during actual processing are systematically collected and organized to obtain historical monitoring records of similar electric spindles. Further, based on these historical monitoring records, a sample operating condition information set and a sample parameter early warning index set are collected. The sample operating condition information set includes operating condition characteristic information corresponding to each historical record, such as spindle speed, load, tool material, cutting speed, feed rate, workpiece material, processing time, and ambient temperature, reflecting the processing status under different operating conditions. The sample parameter early warning index set consists of key monitoring parameter values ​​corresponding to each sample record, covering multiple dimensions such as vibration, current, temperature, and noise, representing the monitoring performance under those conditions. Then, the average proportion of fault prediction errors for similar electric spindles under different sample operating condition information and sample parameter early warning indices is statistically analyzed and set as the sample prediction error, thus obtaining the sample prediction error set.

[0034] Next, a network structure for the prediction error analysis model is constructed based on a backpropagation (BP) neural network. The BP neural network includes an input layer, multiple hidden layers, and an output layer. The input data for the input layer consists of operating condition information and parameter warning indicators, while the output data for the output layer is the prediction error value. Furthermore, the sample operating condition information set, sample parameter warning indicator set, and sample prediction error set are used as training data to supervise the training of the BP neural network. First, the input data is propagated layer by layer through the network to calculate the output prediction error value. Then, the mean squared error (MSE) is used as the loss function to calculate the difference between the predicted value and the target value. Next, based on the error value, gradient calculations are performed on the weights of each layer in the network, and backpropagation is performed using the chain rule. The gradient descent method or its improved algorithms (such as Adam, RMSprop, etc.) are then used to update the weights and biases. Iterative training continues until the loss function converges to a set threshold or the maximum number of training epochs is reached, at which point the network is considered to have converged, and the trained prediction error analysis model is obtained.

[0035] S153: Continue to randomly select parameter warning indicators and analyze prediction errors within the parameter warning threshold until a preset number of times is reached. The parameter warning indicator with the smallest prediction error is then set as the real-time parameter warning indicator, wherein the preset number of times is positively correlated with the fault scale coefficient.

[0036] Specifically, the random selection and prediction error analysis of parameter warning indicators are then performed within the parameter warning threshold. This "random selection-error analysis" process is repeated until a preset number of iterations is reached. The preset number of iterations is positively correlated with the fault scale coefficient; a larger fault scale coefficient indicates a higher potential risk under the current operating condition, requiring more refined parameter selection and thus more iterations. Finally, among all iteration results, the set of parameter warning indicators with the smallest corresponding prediction error is selected as the final real-time parameter warning indicator. This real-time warning indicator possesses the health judgment capability optimally matched to the current processing condition, significantly improving the accuracy and stability of the warning judgment. By introducing a "random sampling and prediction error evaluation" mechanism based on dynamic fault risk control, personalized adaptive adjustment of parameter thresholds can be achieved. Compared to a single fixed threshold or experience-based setting method, this significantly improves the rationality, sensitivity, and robustness of parameter warnings.

[0037] S20: Within the preset time zone, continuously monitor the parameters of several electric spindles being processed in the same batch in the processing workshop, obtain several parameter sequence sets, and determine multiple first warning parameter sets that meet the real-time parameter warning indicators.

[0038] Furthermore, step S20 of the present invention also includes: S21: Judge the several parameter sequence sets according to the real-time parameter warning indicators. If the parameter value does not meet the corresponding parameter warning indicators, mark a warning trigger. S22: Count the number of warning triggers within a preset time window. If the number of warning triggers is greater than the preset number threshold, set the monitoring parameter as the first warning parameter to obtain multiple first warning parameter sets.

[0039] Specifically, firstly, the preset time zone is divided into several sampling periods, for example, every 20 seconds is a sampling interval; within each sampling period, a comprehensive parameter acquisition is performed on all target electric spindles, including vibration parameters, electrical parameters, temperature parameters, and noise parameters; then, the above-mentioned multiple monitoring parameters are organized in chronological order to form a multi-dimensional parameter time series set for each electric spindle. For example, with a total monitoring duration of 3 minutes, data is collected once every 20 seconds, which can form 9 sets of time-series sampling data, forming a continuous parameter sequence, and obtaining several parameter sequence sets.

[0040] Then, based on the real-time parameter warning index, the several parameter sequence sets are judged, and the time series value of each monitoring parameter in the parameter sequence set is traversed. The parameter value at each monitoring time point is judged. If the value exceeds the upper and lower limits defined by the current real-time parameter warning index, it is considered as a warning trigger event. Whenever a warning is triggered, the corresponding spindle number, parameter type, timestamp, and over-limit direction (upper or lower limit) are recorded. For example, if the vibration value of a spindle exceeds the threshold consecutively at the 2nd, 4th, and 5th monitoring points, then the vibration parameter has triggered a warning 3 times in total. Next, the number of triggers for each type of parameter (such as vibration, electrical, temperature, and noise) is counted within a set time window (e.g., 3 minutes). If the cumulative number of triggers for a certain parameter type within the time window exceeds a preset threshold (e.g., greater than or equal to 3 times), then the parameter is considered to have a continuous deviation behavior, and it is marked as the first warning parameter. All parameters that meet this condition are combined into a warning feature set, which is recorded as the first warning parameter set. This method, by performing multiple sampling and statistical judgments within a short time window, can not only promptly capture sudden anomalies but also effectively identify continuous deviation trends of parameters, thereby reducing the risk of false alarms caused by single fluctuations and improving the stability and robustness of the first-level early warning.

[0041] S30: Construct several parameter curve sets based on the several parameter sequence sets, perform anomaly identification on the several parameter curve sets, and determine multiple second warning parameter sets.

[0042] Furthermore, step S30 of the present invention also includes: S31: In a two-dimensional coordinate system, distribute the parameter sequence sets from early to late according to the monitoring time to generate several parameter curve sets; S32: Cluster the several parameter curve sets with the same parameters to obtain multiple parameter curve sets of the same type.

[0043] Specifically, firstly, within a two-dimensional coordinate system, the parameter sequence sets are distributed according to the monitoring time from earliest to latest. That is, the parameter sequence sets are mapped onto the two-dimensional coordinate system in chronological order of monitoring time, with time as the horizontal axis and parameter values ​​as the vertical axis. Curves are then plotted for each type of parameter (such as vibration, electrical, temperature, and noise), resulting in a set of parameter curves for each electric spindle. Next, all electric spindles corresponding to the same type of parameter curves (such as the vibration amplitude curves of all electric spindles) are grouped together to form a set of similar parameter curves, resulting in multiple sets of similar parameter curves.

[0044] S33: Perform anomaly identification on the multiple sets of similar parameter curves respectively to determine multiple sets of second warning parameters.

[0045] Furthermore, step S33 of the present invention also includes: S331: Randomly select any set of similar parameter curves, and randomly select the first similar parameter curve from the set of similar parameter curves; S332: Perform similarity comparison between the first similar parameter curve and other similar parameter curves in the set of similar parameter curves to obtain multiple similarities; S333: Count the number of curves with similarity less than a preset similarity threshold. If the number of curves is greater than a preset number threshold, mark the first similar parameter curve as a second warning parameter, and analyze in sequence to obtain multiple sets of second warning parameters.

[0046] Specifically, firstly, a set of similar parameter curves for any parameter (such as vibration, noise, etc.) is randomly selected from the multiple sets of similar parameter curves, and a first similar parameter curve (such as amplitude, frequency, etc. in vibration parameters) is randomly selected from the set of similar parameter curves. Next, the first similar parameter curve and other similar parameter curves in the set of similar parameter curves are compared for similarity. Common similarity assessment methods include cosine similarity and Pearson correlation coefficient. The similarity value between each pair of curves is calculated, and multiple similarity results are obtained, representing the degree of similarity between the first similar parameter curve and other curves. Then, all the calculated similarities are statistically analyzed, and the number of curves with similarity less than a preset similarity threshold (such as 60%) is recorded. If this number exceeds the preset threshold, it indicates that the shape of the first similar parameter curve is significantly different from other curves, and this difference has certain commonalities in multiple principal axes. Therefore, it can be judged as a potential trend anomaly, and the first similar parameter curve is marked as a second warning parameter. The above steps are repeated to analyze and generate multiple sets of second warning parameters in sequence. By analyzing the similarity of curves of similar parameters, we can effectively capture abnormal trends in parameters and identify potential faults that are slowly degrading or not significantly exceeding the limits.

[0047] S40: Based on the analysis of the multiple first warning parameter sets and multiple second warning parameter sets, multiple warning electric spindles are determined, wherein any parameter of the warning electric spindle simultaneously satisfies the first warning and the second warning.

[0048] Furthermore, step S40 of the present invention further includes: S41: Randomly select a first electric spindle. If the first warning parameter set and the second warning parameter set corresponding to the first electric spindle are not empty, and any parameter simultaneously meets the first warning parameter and the second warning parameter, then mark the first electric spindle as a warning electric spindle.

[0049] Specifically, firstly, from all monitored electric spindles, a first electric spindle is randomly selected for further analysis to obtain the first and second warning parameter sets corresponding to this spindle. Next, it is checked whether both the first and second warning parameter sets corresponding to the selected first electric spindle are not empty. If either set is empty, it indicates that the electric spindle has not triggered an anomaly at a certain warning level, and no further judgment is needed. If both the first and second warning parameter sets are not empty, the next step of analysis is performed. Then, for the first electric spindle, all its monitoring parameters are checked sequentially. If any parameter simultaneously meets the first and second warning conditions (i.e., the value of the parameter exceeds the standard warning threshold and its curve trend is significantly different from other similar curves), the electric spindle may have a high risk of failure, and the first electric spindle is marked as a warning electric spindle. Repeated analysis is performed to identify multiple warning electric spindles. By integrating the first and second warning criteria, the system can monitor both the immediate fault manifestations (such as sudden faults) and trend faults (such as slow degradation and trend changes) of the electric spindle, effectively reducing the probability of false alarms and missed alarms. Ultimately, only those electric spindles that exhibit abnormalities in multiple dimensions and levels will be marked as warning electric spindles, significantly improving the accuracy and reliability of fault identification.

[0050] In summary, the health monitoring and early warning method for precision machine tool electric spindles provided by this invention has the following technical effects: By acquiring the expected operating condition information of the electric spindle within a preset time zone, the standard parameter warning indicators are corrected based on the expected operating condition information to determine the real-time parameter warning indicators. Then, within the preset time zone, parameters of several electric spindles processed in the same batch in the machining workshop are continuously monitored to acquire several parameter sequence sets, and multiple first warning parameter sets that meet the real-time parameter warning indicators are determined. Then, several parameter curve sets are constructed based on the several parameter sequence sets, and anomaly identification is performed on the several parameter curve sets to determine multiple second warning parameter sets. Finally, multiple warning electric spindles are determined by analyzing the multiple first warning parameter sets and multiple second warning parameter sets, wherein any parameter of the warning electric spindle simultaneously meets the first warning and the second warning. In other words, by introducing an adaptive threshold adjustment mechanism based on operating parameters, the warning range of each monitoring parameter can be dynamically corrected, making the health assessment more consistent with the actual processing state, thereby effectively reducing the risk of false alarms and missed alarms caused by fixed threshold settings. By constructing parameter time series curves and using trend analysis and multi-parameter collaborative anomaly identification methods, chronic degradation and complex abnormal behaviors that are difficult to detect by traditional static methods can be identified, thereby improving the sensitivity and accuracy of the warning. Finally, by integrating the results of the first and second warnings, the final warning is triggered only when the parameter meets both warning conditions simultaneously, thereby achieving a highly reliable and high-confidence health monitoring and warning mechanism, significantly improving the accuracy and reliability of fault identification of precision machine tool electric spindles.

[0051] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0052] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for health monitoring and early warning of a precision machine tool electric spindle, characterized in that the method... include: Obtain the expected operating condition information of the electric spindle within a preset time zone, correct the standard parameter warning indicators based on the expected operating condition information, and determine the real-time parameter warning indicators. The operating condition information includes spindle operating parameters, tool characteristic parameters, feed motion parameters, workpiece attribute information, continuous machining duration, and machining environment parameters, including: Using the expected working condition information as a condition constraint and the electric spindle as a device constraint, the failure rate of the sample equipment is determined by statistical analysis of the failure rate based on big data. The ratio of the failure rate of the sample equipment to the average failure rate of historical equipment under standard operating conditions is set as the failure scale coefficient. The parameter adjustment direction and parameter adjustment ratio are determined based on the fault scale coefficient. The standard parameter early warning index is expanded to determine the parameter early warning threshold. The standard parameter early warning index includes at least the standard early warning values ​​of vibration parameters, electrical parameters, temperature parameters and noise parameters. Within the parameter warning threshold, the parameter warning index is optimized, and real-time parameter warning indexes are output, including: Randomly select a first parameter warning indicator within the parameter warning threshold; Using a prediction error analysis model, prediction error analysis is performed based on the expected working condition information and the first parameter warning index, and the first prediction error is output. Continue to randomly select parameter warning indicators and analyze prediction errors within the parameter warning threshold until a preset number of times is reached. The parameter warning indicator with the smallest prediction error is then set as the real-time parameter warning indicator. The preset number of times is positively correlated with the fault scale coefficient. Within the preset time zone, parameters of several electric spindles processed in the same batch in the processing workshop are continuously monitored to obtain several parameter sequence sets, and multiple first warning parameter sets that meet the real-time parameter warning indicators are determined. Several parameter curve sets are constructed based on the aforementioned parameter sequence sets, and anomaly identification is performed on the aforementioned parameter curve sets to determine multiple second warning parameter sets; Multiple warning electric spindles are determined based on the analysis of the multiple first warning parameter sets and multiple second warning parameter sets, wherein any parameter of the warning electric spindle simultaneously satisfies the first warning and the second warning.

2. The method for health monitoring and early warning of a precision machine tool electric spindle according to claim 1, characterized in that, The method for constructing the prediction error analysis model includes: Based on historical monitoring records of similar electric spindles, a sample working condition information set and a sample parameter early warning index set are collected. The average proportion of fault prediction error of similar electric spindles under different sample working condition information and sample parameter early warning indexes is calculated and set as the sample prediction error. The sample prediction error set is then obtained. Using the sample operating condition information set, sample parameter early warning index set, and sample prediction error set as training data, the BP neural network is trained under supervision until the network converges, thus obtaining the prediction error analysis model.

3. The method for health monitoring and early warning of a precision machine tool electric spindle according to claim 1, characterized in that, Determine multiple sets of first warning parameters that satisfy the real-time parameter warning indicators, including: The real-time parameter warning indicators are used to judge the set of several parameter sequences. If the parameter value does not meet the corresponding parameter warning indicators, a warning is triggered. The number of warnings triggered within a preset time window is counted. If the number of warnings triggered is greater than a preset threshold, the monitoring parameter is set as the first warning parameter, resulting in multiple sets of first warning parameters.

4. The method for health monitoring and early warning of a precision machine tool electric spindle according to claim 1, characterized in that, Several parameter curve sets are constructed based on the aforementioned parameter sequence sets. Anomaly identification is performed on these parameter curve sets to determine multiple second warning parameter sets, including: Within a two-dimensional coordinate system, the parameter sequence sets are distributed according to the monitoring time from early to late to generate several parameter curve sets; Clustering the aforementioned sets of parameter curves with the same parameters yields multiple sets of parameter curves of the same type. Anomaly identification is performed on the multiple sets of similar parameter curves to determine multiple sets of second warning parameters.

5. The method for health monitoring and early warning of a precision machine tool electric spindle according to claim 4, characterized in that, Anomaly identification is performed on the multiple sets of similar parameter curves, including: Randomly select any set of similar parameter curves, and then randomly select the first similar parameter curve from the set of similar parameter curves; A similarity comparison is performed between the first similar parameter curve and other similar parameter curves in the set of similar parameter curves to obtain multiple similarity scores; The number of curves with a similarity less than a preset similarity threshold is counted. If the number of curves is greater than a preset number threshold, the first similar parameter curve is marked as a second warning parameter. Multiple sets of second warning parameters are obtained by sequential analysis.

6. The method for health monitoring and early warning of a precision machine tool electric spindle according to claim 1, characterized in that, A first electric spindle is randomly selected. If the first warning parameter set and the second warning parameter set corresponding to the first electric spindle are not empty, and any parameter simultaneously meets the first warning parameter and the second warning parameter, then the first electric spindle is marked as a warning electric spindle.

Images