A method and system for detecting torsional stiffness of a direct-connected spindle

Abstract

The present invention relates to the technical field of direct-connected spindle detection, and in particular to a detection method and system for the torsional stiffness of a direct-connected spindle, the method comprising: cyclically collecting the motion data of the direct-connected spindle and performing classification analysis, fitting the single-collected direct-connected spindle motion data to generate a torsional feature vector; defining the torsional stiffness state of the direct-connected spindle and dividing it into different stiffness levels according to the torsional feature vector, and generating a torsional stiffness node value for each stiffness level; taking the torsional stiffness node value as the aggregation center, aggregating the torsional feature vector to obtain a number of vector aggregation results, and generating a torsional stiffness threshold value according to the vector aggregation result; establishing a torsional stiffness detection model according to the torsional stiffness threshold value, and the torsional stiffness detection model detects the torsional stiffness of direct-connected spindles of different specifications according to different torsional stiffness threshold values. Through the present invention, the problem of inaccurate torsional stiffness detection of the direct-connected spindle is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of direct-connected spindle detection, and in particular to a method and system for detecting torsional stiffness of a direct-connected spindle. Background Art

[0002] In machine tool manufacturing and high-precision processing equipment, the torsional stiffness of the spindle directly affects the processing accuracy and equipment stability. The torsional stiffness of the spindle is related to its stability and durability under high load and high-speed operation, especially in the production of high-precision CNC machine tools, aerospace manufacturing equipment and automotive parts, the demand for torsional stiffness is particularly important.

[0003] Traditional spindle stiffness detection methods usually rely on static detection and offline analysis to test the spindle under a single working condition, but cannot effectively capture the torsional characteristics under dynamic conditions, resulting in low detection accuracy and failure to reflect the true stiffness state of the spindle in actual applications. In addition, existing detection methods can usually only provide an overall judgment of the spindle stiffness, and it is difficult to refine it to different stiffness levels, and cannot meet the stiffness detection needs of spindles of different specifications under different load conditions.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art known to those skilled in the art. Summary of the invention

[0005] The present invention provides a method and system for detecting the torsional stiffness of a direct-connected main shaft, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for detecting torsional stiffness of a direct-connected main shaft, the method comprising:

[0008] Circularly collecting the direct-connected spindle motion data and performing classification analysis, and fitting the direct-connected spindle motion data collected once to generate a torsion feature vector;

[0009] Defining and dividing the torsional stiffness state of the direct-connected main shaft into different stiffness levels according to the torsional characteristic vector, and generating a torsional stiffness node value for each stiffness level;

[0010] Taking the torsional stiffness node value as the aggregation center, aggregating the torsional feature vectors to obtain a plurality of vector aggregation results, and generating a torsional stiffness threshold value according to the vector aggregation results;

[0011] A torsional stiffness detection model is established according to the torsional stiffness threshold value, and the torsional stiffness detection model detects the torsional stiffness of directly connected spindles of different specifications according to different torsional stiffness threshold values.

[0012] Furthermore, fitting the single direct-connected spindle motion data to generate a torsion feature vector includes:

[0013] A plurality of time windows are set according to the detection time sequence, and samples are selected from the cyclically collected direct-connected spindle motion data according to the time windows to obtain a plurality of time sample data;

[0014] Performing composite fitting on the time sample data, wherein the composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected main shaft under different load and speed conditions;

[0015] Constructing a parameter recognition layer to extract key parameters for torsion detection in the fitting motion data;

[0016] The torsion detection key parameter is used as an accompanying feature to generate a torsion feature vector containing multi-dimensional information.

[0017] Furthermore, a parameter recognition layer is constructed, including:

[0018] Establishing a time-frequency analysis domain, extracting the direct-connected spindle torsion characteristics from the fitted motion data;

[0019] Capturing and analyzing the torsion characteristics of the direct-connected main shaft to capture the corresponding torsion postures in different time windows;

[0020] Applying a screening detection algorithm to optimize the extracted characteristic parameters of the torsional behavior, screening the characteristic parameters of the torsional behavior according to the strong positive correlation of the torsional stiffness detection, and obtaining the characteristic parameters with strong positive correlation;

[0021] The operating condition of the direct-connected main shaft is detected to obtain the operating condition result, the strongly positively correlated characteristic parameters are used as a basis for parameter identification, and a parameter identification layer is constructed according to the operating condition result.

[0022] Further, generating a torsional stiffness node value for each stiffness level includes:

[0023] Generating different stiffness levels corresponding to different torsional stiffnesses, and selecting a number of torsional eigenvectors as initial aggregation centers according to the data median from the torsional eigenvectors generated by fitting each stiffness level;

[0024] Perform multiple iterations of comparison on each of the initial aggregation centers, calculate distance values ​​between other torsion feature vectors and the current initial aggregation center, and adjust the torsion feature vector of the initial aggregation center;

[0025] In each iteration, the twisted eigenvector that is closer to the initial aggregation center is replaced with the aggregation center, the new aggregation center position is recalculated, and the iteration is repeated until the aggregation center no longer changes;

[0026] Through multiple comparisons and iterations, the torsional eigenvector contained in the selected aggregation center is used as the torsional stiffness node value.

[0027] Furthermore, multiple iterative comparisons are performed on each initial aggregation center, including:

[0028] Calculating the spatial distance between each of the initial aggregation centers and the twisted eigenvectors around it;

[0029] An aggregation limit value is set according to the calculated spatial distance, and all torsion feature vectors whose spatial distance is greater than the aggregation limit value are attributed to the corresponding initial aggregation center to form an initial aggregation domain;

[0030] In each iteration, the aggregation average of all the torsion feature vectors in the initial aggregation domain is calculated, the torsion feature vector close to the aggregation average is used as a new aggregation center, and the position of the aggregation center is adjusted;

[0031] After each iteration is completed, the change in the aggregation center position is determined. If the change is less than the aggregation limit value, the aggregation center is considered to have converged. Otherwise, the iterative optimization is continued until the aggregation center no longer changes.

[0032] Further, a torsional stiffness threshold is generated, including:

[0033] Collecting a plurality of torsional stiffness node values ​​generated by each of the stiffness levels, calculating a node value range for each level, and characterizing the torsional characteristics of each stiffness level;

[0034] A transition interval is set between the node value ranges of adjacent stiffness levels, and an upper limit and a lower limit of the transition interval are determined according to the distribution characteristics of the torsional stiffness node values ​​of adjacent levels;

[0035] A statistical fitting analysis is performed on the transition intervals of the torsional stiffness node values ​​of adjacent levels to generate a torsional stiffness threshold value for each stiffness level.

[0036] Further, a statistical fitting analysis is performed on the transition intervals of the torsional stiffness node values ​​of adjacent levels, including:

[0037] Performing statistical analysis on the distribution characteristics of torsional stiffness node values ​​of adjacent stiffness levels within the transition interval to quantify the concentration and dispersion of the node values;

[0038] Based on the distribution characteristics of the transition interval, parameter estimation is performed on the node values ​​in the transition interval to calculate the key parameters of the node;

[0039] Verify the accuracy of the key parameters of the node and evaluate the fitting effect;

[0040] Based on the evaluated key parameters of the nodes, the node value that best represents the critical points of two adjacent levels in each transition interval is set as the torsional stiffness threshold.

[0041] Furthermore, a torsional stiffness detection model is established, including:

[0042] Establishing a data input process for receiving the cyclically collected direct-connected spindle motion data and converting it into the torsional feature vector;

[0043] Based on the generated torsional stiffness threshold, a stiffness level identification mechanism is constructed to determine the stiffness level of the current direct-connected spindle according to each torsional characteristic vector;

[0044] Establishing a threshold mapping relationship, establishing a corresponding relationship between the torsional stiffness threshold and directly coupled spindles of different specifications, so as to adapt to the threshold adjustment requirements under different directly coupled spindle specifications and workloads in the future;

[0045] Formulate a real-time detection and feedback process, take the torsional characteristic vector in the operation of the direct-connected spindle as input, and analyze and evaluate the current stiffness state through the torsional stiffness detection model;

[0046] Define an adaptive update strategy to regularly adjust the torsional stiffness threshold and stiffness level based on historical inspection data and the actual working conditions of the directly connected spindle.

[0047] A detection system for torsional stiffness of a direct-connected main shaft, the system comprising:

[0048] The data fitting module cyclically collects the direct-connected spindle motion data and performs classification analysis, and generates a torsion feature vector by fitting the direct-connected spindle motion data collected once;

[0049] A level classification module defines and divides the torsional stiffness state of the direct-connected spindle into different stiffness levels according to the torsional eigenvector, and generates a torsional stiffness node value for each stiffness level;

[0050] A threshold generation module, which uses the torsional stiffness node value as the aggregation center, aggregates the torsional feature vectors to obtain a number of vector aggregation results, and generates a torsional stiffness threshold according to the vector aggregation results;

[0051] The stiffness detection module establishes a torsional stiffness detection model according to the torsional stiffness threshold value. The torsional stiffness detection model detects the torsional stiffness of direct-connected spindles of different specifications according to different torsional stiffness threshold values.

[0052] Furthermore, the data fitting module includes:

[0053] The time acquisition unit sets a number of time windows according to the detection time sequence, selects samples of the cyclically acquired direct-connected spindle motion data according to the time windows, and obtains a number of time sample data;

[0054] The composite fitting unit performs composite fitting on the time sample data. The composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected spindle under different load and speed conditions;

[0055] The parameter identification unit constructs a parameter identification layer to extract key parameters for torsion detection in the fitted motion data;

[0056] The vector generating unit uses the key parameters of the torsion detection as accompanying features to generate a torsion feature vector containing multi-dimensional information.

[0057] The technical solution of the present invention can achieve the following technical effects:

[0058] Improve the detection accuracy of torsional stiffness, realize dynamic grading of stiffness, adapt to the detection requirements of spindles of different specifications, and provide adaptive update strategies to effectively improve the stability of equipment operation and processing accuracy, and meet the strict requirements of modern high-precision processing equipment for spindle stiffness.

[0059] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 It is a flow chart of a method for detecting torsional stiffness of a direct-connected spindle;

[0062] Figure 2 A schematic diagram of the process of generating a torsion eigenvector;

[0063] Figure 3 Conceptual diagram of nodal values ​​of torsional stiffness generated for aggregates;

[0064] Figure 4Schematic diagram of the structure for obtaining the torsional stiffness threshold. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0067] Embodiment 1

[0068] like Figure 1 As shown, the present application provides a method for detecting the torsional stiffness of a direct-connected main shaft, the method comprising:

[0069] S10: cyclically collecting the direct-connected spindle motion data and performing classification analysis, and fitting the direct-connected spindle motion data collected once to generate a torsion feature vector;

[0070] S20: defining and dividing the torsional stiffness state of the direct-connected main shaft into different stiffness levels according to the torsional characteristic vector, and generating a torsional stiffness node value for each stiffness level;

[0071] S30: taking the torsional stiffness node value as the aggregation center, aggregating the torsional characteristic vectors to obtain a number of vector aggregation results, and generating a torsional stiffness threshold according to the vector aggregation results;

[0072] S40: establishing a torsional stiffness detection model according to the torsional stiffness threshold, wherein the torsional stiffness detection model detects the torsional stiffness of directly connected spindles of different specifications according to different torsional stiffness thresholds.

[0073] Specifically, the motion data of the direct-connected spindle is collected cyclically, and the data is classified and analyzed. Each collected data includes the motion characteristics of the spindle under different loads, speeds and operating conditions. The collected single motion data is fitted into a torsional feature vector. During the fitting process, the time window is used to select samples of the data, and a composite fitting method is used to extract the key torsional parameters in the motion to generate a multi-dimensional feature vector. The torsional stiffness state of the direct-connected spindle is defined and classified according to the torsional feature vector. By analyzing the feature vectors of different stiffness states, the corresponding torsional stiffness node values ​​are generated to represent different stiffness levels. With the node values ​​of each stiffness level as the initial aggregation center, all torsional feature vectors are aggregated and calculated to obtain several aggregation results. Then, the characteristic distribution range of each stiffness level is calculated based on these aggregation results, thereby generating a torsional stiffness threshold. Based on the generated torsional stiffness threshold, a torsional stiffness detection model is constructed. The detection model can detect the current stiffness level of the spindle according to the stiffness threshold of spindles of different specifications by analyzing and judging the torsional feature vectors collected in real time. The model is suitable for spindles of different specifications and supports torsional stiffness detection under various working conditions.

[0074] Through the technical solution of the present invention, the torsional stiffness detection accuracy is improved, and an adaptive update strategy is provided to effectively improve the stability of equipment operation and processing accuracy, meeting the strict requirements of modern high-precision processing equipment on spindle stiffness.

[0075] Further, if Figure 2 As shown, the torsion feature vector is generated by fitting the single direct-connected spindle motion data, including:

[0076] A number of time windows are set according to the detection time sequence, and samples of the directly connected spindle motion data collected cyclically are selected according to the time windows to obtain a number of time sample data;

[0077] Performing composite fitting on the time sample data, the composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected spindle under different load and speed conditions;

[0078] Construct a parameter recognition layer to extract key parameters for torsion detection in the fitted motion data;

[0079] The key parameters of torsion detection are used as accompanying features to generate a torsion feature vector containing multi-dimensional information.

[0080] As a preferred embodiment of the above embodiment, several time windows are set according to the time sequence of detection, each time window represents a specific detection cycle, which is used to capture the spindle motion data within the time period, and sample selection is performed on the cyclically collected data within each time window, so as to obtain several time sample data containing the spindle torsional behavior, and these time sample data cover the dynamic state of the spindle at different times; the obtained time sample data are subjected to composite fitting processing, and composite fitting refers to multi-dimensional and multi-parameter fitting analysis of the spindle torsional behavior under different load and speed conditions, so as to generate representative fitting motion data, and these fitting data can reflect the actual torsional characteristics of the spindle under different working conditions; based on the motion data obtained by composite fitting, a parameter recognition layer is constructed to extract key torsional detection parameters from the fitting data, and these key parameters may include the torque, speed, torsional angle and other characteristic indicators of the spindle reflecting torsional stiffness; the extracted key torsional detection parameters are used as accompanying features to generate a torsional feature vector containing multi-dimensional information, and the feature vector integrates the torsional behavior information of the spindle under different loads, speeds and operating conditions, has high dimensionality and high representativeness, and provides accurate data information for stiffness detection.

[0081] Furthermore, the parameter recognition layer is constructed, including:

[0082] Establish a time-frequency analysis domain to extract the torsion characteristics of the direct-connected spindle from the fitted motion data;

[0083] Capture and analyze the torsion characteristics of the direct-connected spindle and capture the corresponding torsion postures in different time windows;

[0084] The extracted torsional behavior characteristic parameters are optimized by applying the screening detection algorithm, and the torsional behavior characteristic parameters are screened according to the strong positive correlation of torsional stiffness detection to obtain the strong positive correlation characteristic parameters;

[0085] The operating conditions of the direct-connected spindle are detected to obtain the operating condition results. The strongly positively correlated characteristic parameters are used as the basis for parameter identification, and a parameter identification layer is constructed based on the operating condition results.

[0086] As a preferred embodiment of the above, based on the motion data generated by the composite fitting, a time-frequency analysis domain is constructed. The time-frequency analysis domain is used to analyze the time domain and frequency domain characteristics of the direct-connected spindle motion data, so as to extract key information reflecting the spindle torsional characteristics from the fitted motion data. The analysis can reveal the change of the torsional stiffness of the spindle in different time periods; in the time-frequency analysis domain, the torsional characteristics of the direct-connected spindle are captured and analyzed, and the torsional posture of the spindle in each time window is captured through the set time window, including key behaviors such as angle change and torque characteristics. These torsional postures can reflect the dynamic torsional characteristics of the spindle under real-time operation; the screening detection algorithm is applied to optimize the captured torsional behavior characteristic parameters. In the screening process, according to the strong positive correlation principle of torsional stiffness detection, , the characteristic parameters of torsional behavior are screened and filtered, and only the characteristic parameters that are highly correlated with torsional stiffness (such as angle and torque at a specific speed) will be selected to form a set of characteristic parameters with strong positive correlation; the operating conditions of the direct-connected spindle (such as load, speed, temperature, etc.) are detected to obtain the specific conditions under the current operating conditions, and the results of these operating conditions are combined with the strongly positively correlated characteristic parameters as the basis for constructing the parameter identification layer. By considering the operating conditions, the parameter identification layer can adapt to different operating states and improve its detection accuracy under actual conditions; based on the strongly positively correlated characteristic parameters and the operating condition results, the parameter identification layer is constructed, and the parameter identification layer can dynamically adjust the torsional characteristic parameters to adapt to different spindle conditions, thereby forming a more stable and accurate torsional characteristic vector.

[0087] Further, if Figure 3 As shown, nodal values ​​for torsional stiffness are generated for each stiffness level, including:

[0088] Different torsional stiffnesses are correspondingly generated into different stiffness levels, and a number of torsional eigenvectors are selected as initial aggregation centers according to the median of the data from the torsional eigenvectors generated by fitting each stiffness level;

[0089] Perform multiple iterations of comparisons on each initial aggregation center, calculate distance values ​​between other torsion feature vectors and the current initial aggregation center, and adjust the torsion feature vector of the initial aggregation center;

[0090] In each iteration, the twisted eigenvector that is closer to the initial aggregation center is replaced with the aggregation center, the new aggregation center position is recalculated, and the iteration is repeated until the aggregation center no longer changes;

[0091] Through multiple mutual comparisons and iterations, the torsional eigenvectors contained in the selected aggregation centers are used as the torsional stiffness node values.

[0092] As a preferred embodiment of the above, firstly, according to the characteristics of different torsional stiffnesses, the stiffness state of the direct-connected main shaft is divided into multiple stiffness levels, and each stiffness level corresponds to a set of torsional eigenvectors, which are generated according to the torsional performance of the main shaft under different working conditions; in the torsional eigenvectors generated by fitting at each stiffness level, several eigenvectors are selected according to the median of the data as the initial aggregation center, and the median selection ensures that the initial aggregation center is representative and avoids the influence of extreme values ​​on the accuracy of aggregation; for each initial aggregation center, multiple iterative comparisons are performed, and firstly the distance value between the current initial aggregation center and other torsional eigenvectors is calculated to determine the proximity of these eigenvectors in space, and the eigenvectors with closer distances are selected. The eigenvectors are classified as the initial aggregation center, and the average position of these eigenvectors is reset to the new aggregation center position. In each iteration, the torsion eigenvector closer to the aggregation center is used as the new aggregation center, the distances of all torsion eigenvectors are calculated again, and the aggregation center is updated. This process is repeated until the aggregation center position no longer changes significantly, that is, it converges to a stable position. After multiple mutual comparisons and iterative adjustments, the aggregation center finally screened out represents the typical torsion eigenvector of the stiffness level. The torsion eigenvectors of these aggregation centers are defined as the torsion stiffness node values ​​of the stiffness level. The torsion stiffness node values ​​can be used as representative values ​​of the stiffness level for subsequent stiffness detection and analysis.

[0093] Furthermore, multiple iterative comparisons are performed on each initial aggregation center, including:

[0094] The spatial distance between each initial aggregation center and its surrounding twisted eigenvectors is calculated;

[0095] An aggregation limit value is set according to the calculated spatial distance, and all torsion eigenvectors whose spatial distance is greater than the aggregation limit value are attributed to the corresponding initial aggregation center to form an initial aggregation domain;

[0096] In each iteration, the aggregated average of all torsion eigenvectors in the initial aggregation domain is calculated, and the torsion eigenvector close to the aggregated average is used as the new aggregation center, and the position of the aggregation center is adjusted;

[0097] After each iteration, the change in the aggregation center position is determined. If the change is less than the aggregation limit, the aggregation center is considered to have converged. Otherwise, the iterative optimization continues until the aggregation center no longer changes.

[0098] As a preferred embodiment of the above embodiment, for each initial aggregation center and the torsion eigenvectors around it, the spatial distance between them is calculated, and the spatial distance reflects the similarity between the eigenvector and the aggregation center, which helps to determine which eigenvectors are closer to the aggregation center; according to the calculated spatial distance, an aggregation limit value is set, and all torsion eigenvectors with a spatial distance less than the limit value are attributed to the corresponding initial aggregation center, thereby forming an initial aggregation domain, which contains eigenvectors that are relatively close to the aggregation center; in each iteration, the aggregation average value of all torsion eigenvectors in the current aggregation domain is calculated, and the torsion eigenvectors close to the average value are re-used as new aggregation centers; after each iteration, the position change between the current aggregation center and the aggregation center of the previous iteration is calculated, if the change is less than the preset aggregation limit value, it means that the aggregation center has reached a convergence state, otherwise, iterative optimization is continued to adjust the position of the aggregation center until the position change is less than the limit value; after multiple iterations, when the position change of the aggregation center is less than the aggregation limit value, the aggregation center is considered to have converged, and the aggregation center position at this time is the final position after optimization, which is used to define the torsion stiffness node value of the stiffness level.

[0099] Further, if Figure 4 As shown, the torsional stiffness threshold is generated, including:

[0100] Collect multiple torsional stiffness node values ​​generated for each stiffness level, calculate the node value range for each level, and characterize the torsional characteristics of each stiffness level;

[0101] A transition interval is set between the node value ranges of adjacent stiffness levels, and the upper and lower limits of the transition interval are determined according to the distribution characteristics of the torsional stiffness node values ​​of adjacent levels;

[0102] The transition intervals of torsional stiffness node values ​​of adjacent levels are statistically fitted and analyzed to generate the torsional stiffness threshold for each stiffness level.

[0103] As a preferred embodiment of the above embodiment, the torsional stiffness node values ​​of each stiffness level that have been generated are collected to calculate the node value range of each stiffness level, and these torsional stiffness node value ranges reflect the torsional characteristics of each stiffness level; a transition interval is set between the node value ranges of adjacent stiffness levels to buffer the boundaries between different stiffness levels, and the upper and lower limits of each transition interval are determined according to the distribution characteristics of the node values ​​of adjacent levels; a statistical analysis is performed on the distribution of the torsional stiffness node values ​​of adjacent levels in the transition interval to quantify the degree of concentration and dispersion of these node values, and a transition relationship between the various stiffness levels is established through statistical fitting analysis, thereby effectively identifying the characteristic boundaries of each stiffness level; based on the statistical fitting results of the node values ​​in the transition interval, the critical points of each stiffness level are calculated as torsional stiffness thresholds, and these thresholds can clearly distinguish different stiffness levels, becoming the standard basis for the detection model, which is used to determine the stiffness state of the spindle in actual operation.

[0104] Furthermore, a statistical fitting analysis is performed on the transition intervals of adjacent level torsional stiffness node values, including:

[0105] Statistical analysis is performed on the distribution characteristics of the torsional stiffness node values ​​of adjacent stiffness levels in the transition interval to quantify the concentration and dispersion of the node values;

[0106] Based on the distribution characteristics of the transition interval, the node values ​​in the transition interval are estimated and the key parameters of the nodes are calculated;

[0107] Verify the accuracy of key node parameters and evaluate the fitting effect;

[0108] Based on the evaluated key parameters of the nodes, the node value that best represents the critical points of two adjacent levels in each transition interval is set as the torsional stiffness threshold.

[0109] As a preferred embodiment of the above embodiment, within the transition interval of adjacent stiffness levels, the distribution of torsional stiffness node values ​​is statistically analyzed, focusing on the concentration and dispersion of node values ​​(such as mean, variance, etc.), and analyzing the distribution characteristics of node values ​​between adjacent stiffness levels; based on the distribution characteristics of node values ​​in the transition interval, parameter estimation is performed on these node values, aiming to calculate key statistical parameters (such as mean, variance, standard deviation, etc.), which can represent the central tendency and discreteness of node values ​​in the transition interval, thereby characterizing the transition characteristics of node values ​​of adjacent levels; the calculated node key parameters are precision verified, the fitting effect is evaluated, the accuracy of parameter estimation is ensured, and it is tested whether the fitting model can effectively represent the distribution of node values ​​in the transition interval. If the fitting effect is not ideal, it is necessary to readjust the parameters or increase sample data to optimize the results; based on the verified node key parameters, the node value that best represents the critical point of two adjacent stiffness levels in the transition interval is set as the torsional stiffness threshold, and the torsional stiffness threshold marks the boundary between the two stiffness levels, so that the detection model can accurately identify the change of the spindle stiffness state.

[0110] Furthermore, a torsional stiffness detection model is established, including:

[0111] Establishing data input processing for receiving cyclically collected direct-connected spindle motion data and converting it into torsional feature vectors;

[0112] Based on the generated torsional stiffness threshold, a stiffness level identification mechanism is constructed to determine the stiffness level of the current direct-connected spindle according to each torsional eigenvector;

[0113] Establish a threshold mapping relationship, establish a corresponding relationship between the torsional stiffness threshold and different specifications of the direct-coupled spindle, and adapt to the threshold adjustment requirements under different direct-coupled spindle specifications and workloads in the future;

[0114] Develop a real-time detection and feedback process, take the torsional eigenvector of the direct-connected spindle as input, and analyze and evaluate the current stiffness state through the torsional stiffness detection model;

[0115] Define an adaptive update strategy to regularly adjust the torsional stiffness threshold and stiffness level based on historical inspection data and the actual working conditions of the directly connected spindle.

[0116] As a preferred embodiment of the above, a data input processing module is designed to receive motion data collected from the direct-connected spindle cycle, convert the data into torsional feature vectors, so that the detection model can use these feature vectors to reflect the torsional stiffness state of the spindle, and provide input data support for subsequent analysis and judgment; based on the generated torsional stiffness threshold, a stiffness level recognition mechanism is constructed, and different stiffness levels are distinguished by each stiffness threshold, and the stiffness level of the current direct-connected spindle is judged according to each torsional feature vector input in real time, so as to ensure that the spindle operation state can be detected and classified in time; a mapping relationship is established between the torsional stiffness threshold and the direct-connected spindles of different specifications. A system is designed to adapt to the detection requirements under different spindle specifications and workloads. This mapping relationship allows the automatic matching of the applicable torsional stiffness threshold according to the spindle specifications, enhancing the versatility and adaptability of the model. A real-time detection and feedback process is designed, and the real-time collected torsional feature vector is used as input to analyze and evaluate the current stiffness state of the spindle through the detection model. Based on historical detection data and the actual operating conditions of the spindle, an adaptive update strategy is formulated to regularly adjust the torsional stiffness threshold and stiffness level. This strategy ensures that the detection model can continue to maintain high accuracy under different environments and load conditions, and dynamically optimizes the threshold setting and stiffness level according to actual conditions.

[0117] Embodiment 2

[0118] Based on the same inventive concept as the method for detecting the torsional stiffness of a direct-connected spindle in the aforementioned embodiment, the present invention further provides a system for detecting the torsional stiffness of a direct-connected spindle, the system comprising:

[0119] The data fitting module cyclically collects the direct-connected spindle motion data and performs classification analysis, and generates a torsion feature vector by fitting the direct-connected spindle motion data collected once;

[0120] A level classification module defines and divides the torsional stiffness state of the direct-connected spindle into different stiffness levels according to the torsional eigenvector, and generates a torsional stiffness node value for each stiffness level;

[0121] A threshold generation module, which uses the torsional stiffness node value as the aggregation center, aggregates the torsional feature vectors to obtain a number of vector aggregation results, and generates a torsional stiffness threshold according to the vector aggregation results;

[0122] The stiffness detection module establishes a torsional stiffness detection model according to the torsional stiffness threshold value. The torsional stiffness detection model detects the torsional stiffness of direct-connected spindles of different specifications according to different torsional stiffness threshold values.

[0123] The above-mentioned adjustment system in the present invention can effectively realize a method for detecting the torsional stiffness of a direct-connected main shaft, and the technical effects that can be achieved are as described in the above-mentioned embodiments, which will not be repeated here.

[0124] Specifically, the data fitting module includes:

[0125] The time acquisition unit sets a number of time windows according to the detection time sequence, selects samples of the cyclically acquired direct-connected spindle motion data according to the time windows, and obtains a number of time sample data;

[0126] The composite fitting unit performs composite fitting on the time sample data. The composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected spindle under different load and speed conditions;

[0127] The parameter identification unit constructs a parameter identification layer to extract key parameters for torsion detection in the fitted motion data;

[0128] The vector generating unit uses the key parameters of the torsion detection as accompanying features to generate a torsion feature vector containing multi-dimensional information.

[0129] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the optimization effects corresponding to the method in Example 1, which will not be repeated here.

[0130] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined therein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A method for detecting torsional stiffness of a direct-connected main shaft, characterized in that: The method comprises: cyclically collecting the direct-connected spindle motion data and performing classification analysis, and fitting the direct-connected spindle motion data collected once to generate a torsion feature vector; Defining and dividing the torsional stiffness state of the direct-connected main shaft into different stiffness levels according to the torsional characteristic vector, and generating a torsional stiffness node value for each stiffness level; Taking the torsional stiffness node value as the aggregation center, aggregating the torsional feature vectors to obtain a plurality of vector aggregation results, and generating a torsional stiffness threshold value according to the vector aggregation results; A torsional stiffness detection model is established according to the torsional stiffness threshold, wherein the torsional stiffness detection model detects the torsional stiffness of directly connected spindles of different specifications according to different torsional stiffness thresholds; The torsion characteristic vector is generated by fitting the single direct-connected spindle motion data, including: A plurality of time windows are set according to the detection time sequence, and samples are selected from the cyclically collected direct-connected spindle motion data according to the time windows to obtain a plurality of time sample data; Performing composite fitting on the time sample data, wherein the composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected spindle under different load and speed conditions; Constructing a parameter recognition layer to extract key parameters for torsion detection in the fitting motion data; Using the torsion detection key parameter as an accompanying feature to generate a torsion feature vector containing multi-dimensional information; A torsional stiffness node value is generated for each of the stiffness levels, including: Generating different stiffness levels corresponding to different torsional stiffnesses, and selecting a number of torsional eigenvectors as initial aggregation centers according to the data median from the torsional eigenvectors generated by fitting each stiffness level; Perform multiple iterations of comparison on each of the initial aggregation centers, calculate distance values ​​between other torsion feature vectors and the current initial aggregation center, and adjust the torsion feature vector of the initial aggregation center; In each iteration, the twisted eigenvector that is closer to the initial aggregation center is replaced with the aggregation center, the new aggregation center position is recalculated, and the iteration is repeated until the aggregation center no longer changes; Through multiple comparisons and iterations, the torsional eigenvector contained in the selected aggregation center is used as the torsional stiffness node value.

2. A method for detecting torsional stiffness of a direct-connected main shaft according to claim 1, characterized in that: Construct the parameter recognition layer, including: Establishing a time-frequency analysis domain, extracting the direct-connected spindle torsion characteristics from the fitted motion data; Capturing and analyzing the torsion characteristics of the direct-connected main shaft to capture the corresponding torsion postures in different time windows; Applying a screening detection algorithm to optimize the extracted characteristic parameters of the torsional behavior, screening the characteristic parameters of the torsional behavior according to the strong positive correlation of the torsional stiffness detection, and obtaining the characteristic parameters with strong positive correlation; The operating condition of the direct-connected main shaft is detected to obtain the operating condition result, the strongly positively correlated characteristic parameters are used as a basis for parameter identification, and a parameter identification layer is constructed according to the operating condition result.

3. The method for detecting torsional stiffness of a direct-connected main shaft according to claim 1, characterized in that: Multiple iterative comparisons are performed for each initial aggregation center, including: Calculating the spatial distance between each of the initial aggregation centers and the twisted eigenvectors around it; An aggregation limit value is set according to the calculated spatial distance, and all torsion feature vectors whose spatial distance is greater than the aggregation limit value are attributed to the corresponding initial aggregation center to form an initial aggregation domain; In each iteration, the aggregation average of all the torsion feature vectors in the initial aggregation domain is calculated, the torsion feature vector close to the aggregation average is used as a new aggregation center, and the position of the aggregation center is adjusted; After each iteration is completed, the change in the aggregation center position is determined. If the change is less than the aggregation limit value, the aggregation center is considered to have converged. Otherwise, the iterative optimization is continued until the aggregation center no longer changes.

4. The method for detecting torsional stiffness of a direct-connected main shaft according to claim 1, characterized in that: Generates torsional stiffness thresholds, including: Collecting a plurality of torsional stiffness node values ​​generated by each of the stiffness levels, calculating a node value range for each level, and characterizing the torsional characteristics of each stiffness level; A transition interval is set between the node value ranges of adjacent stiffness levels, and an upper limit and a lower limit of the transition interval are determined according to the distribution characteristics of the torsional stiffness node values ​​of adjacent levels; A statistical fitting analysis is performed on the transition intervals of the torsional stiffness node values ​​of adjacent levels to generate a torsional stiffness threshold value for each stiffness level.

5. A method for detecting torsional stiffness of a direct-connected main shaft according to claim 4, characterized in that: Performing statistical fitting analysis on the transition intervals of the torsional stiffness node values ​​of adjacent levels, including: Performing statistical analysis on the distribution characteristics of torsional stiffness node values ​​of adjacent stiffness levels within the transition interval to quantify the concentration and dispersion of the node values; Based on the distribution characteristics of the transition interval, parameter estimation is performed on the node values ​​in the transition interval to calculate the key parameters of the node; Verify the accuracy of the key parameters of the node and evaluate the fitting effect; Based on the evaluated key parameters of the nodes, the node value that best represents the critical points of two adjacent levels in each transition interval is set as the torsional stiffness threshold.

6. A method for detecting torsional stiffness of a direct-connected main shaft according to claim 1, characterized in that: Establish a torsional stiffness detection model, including: Establishing a data input process for receiving the cyclically collected direct-connected spindle motion data and converting it into the torsional feature vector; Based on the generated torsional stiffness threshold, a stiffness level identification mechanism is constructed to determine the stiffness level of the current direct-connected spindle according to each torsional characteristic vector; Establishing a threshold mapping relationship, establishing a corresponding relationship between the torsional stiffness threshold and directly coupled spindles of different specifications, so as to adapt to the threshold adjustment requirements under different directly coupled spindle specifications and workloads in the future; Formulate a real-time detection and feedback process, take the torsional characteristic vector in the operation of the direct-connected spindle as input, and analyze and evaluate the current stiffness state through the torsional stiffness detection model; Define an adaptive update strategy to regularly adjust the torsional stiffness threshold and stiffness level based on historical inspection data and the actual working conditions of the directly connected spindle.

7. A detection system for torsional stiffness of a direct-connected spindle, characterized in that: Using the method for detecting torsional stiffness of a direct-connected main shaft as claimed in claim 1, the system comprises: The data fitting module cyclically collects the direct-connected spindle motion data and performs classification analysis, and generates a torsion feature vector by fitting the direct-connected spindle motion data collected once; A level classification module defines and divides the torsional stiffness state of the direct-connected spindle into different stiffness levels according to the torsional characteristic vector, and generates a torsional stiffness node value for each stiffness level; A threshold generation module, which uses the torsional stiffness node value as the aggregation center, aggregates the torsional feature vectors to obtain a number of vector aggregation results, and generates a torsional stiffness threshold according to the vector aggregation results; The stiffness detection module establishes a torsional stiffness detection model according to the torsional stiffness threshold. The torsional stiffness detection model detects the torsional stiffness of direct-connected spindles of different specifications according to different torsional stiffness thresholds. The data fitting module comprises: The time acquisition unit sets a number of time windows according to the detection time sequence, selects samples of the cyclically acquired direct-connected spindle motion data according to the time windows, and obtains a number of time sample data; The composite fitting unit performs composite fitting on the time sample data. The composite fitting generates fitting motion data by fitting the torsional behavior of the direct-connected spindle under different load and speed conditions; The parameter identification unit constructs a parameter identification layer to extract key parameters for torsion detection in the fitted motion data; The vector generating unit uses the key parameters of the torsion detection as accompanying features to generate a torsion feature vector containing multi-dimensional information.

Images