Spindle life prediction method and system based on application analysis

Abstract

The present invention relates to the field of equipment life management technology, and in particular to a method and system for predicting the life of a spindle based on application analysis. The method comprises: establishing a modal perception network, collecting multi-source operating data during the operation of the spindle, and generating a multi-source operating characteristic matrix; based on the multi-source operating characteristic matrix, establishing a multi-application feature map, and generating a multi-scenario feature vector describing the operating status of the equipment; performing multi-dimensional coupled fatigue analysis on the multi-scenario feature vector, and outputting the fatigue damage accumulation value of the spindle; based on the fatigue damage accumulation value, acquiring and combining the historical motion data of the spindle, establishing a heterogeneous parameter life prediction model, and outputting the remaining life prediction value based on the heterogeneous parameter life prediction model. Through the present invention, the problems of the traditional spindle life prediction method, such as a single data source, insufficient prediction accuracy under complex working conditions, difficulty in modeling nonlinear fatigue damage, and lack of dynamic feedback and correction capabilities, are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment life management, and in particular to a method and system for predicting the life of a spindle based on application analysis. Background Art

[0002] The spindle is a critical component in various types of mechanical equipment, and its operating status directly impacts the overall performance and reliability of the equipment. However, due to the long-term complex and changing operating environment of the spindle, fatigue damage gradually accumulates, potentially leading to failure. Therefore, accurately analyzing spindle fatigue damage and predicting its remaining life are important research topics in the field of mechanical equipment health management.

[0003] Currently, traditional life prediction methods primarily rely on empirical formulas or single-sensor monitoring. These methods suffer from problems such as a single data source, insufficient model accuracy, and a lack of dynamic feedback mechanisms. These methods typically analyze data from monitoring data of a single physical quantity (such as temperature, vibration, or stress), failing to fully reflect the multi-source characteristics of the spindle during operation. Furthermore, prediction models often assume stable spindle operation and ignore the dynamic changes under multi-application operating conditions, resulting in significant deviations in prediction results. Furthermore, traditional methods struggle to provide real-time feedback and dynamic adjustments to the spindle's operating status, making them unable to adapt to fatigue evolution under complex applications.

[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 an admission or any form of suggestion that the information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0005] The present invention provides a life prediction method and system for a main shaft based on application analysis, 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 life prediction method for a main shaft based on application analysis, the method comprising:

[0008] Establish a modal perception network to collect multi-source operation data during the spindle operation and generate a multi-source operation characteristic matrix;

[0009] Based on the multi-source operation characteristic matrix, a multi-application feature map is established, and a multi-scenario feature vector describing the operation status of the device is generated;

[0010] Performing multi-dimensional coupled fatigue analysis on the multi-scenario feature vectors, and outputting a fatigue damage accumulation value of the main shaft;

[0011] According to the fatigue damage accumulation value, the historical motion data of the main shaft is obtained and combined to establish a different parameter life prediction model, and the remaining life prediction value is output based on the different parameter life prediction model.

[0012] Furthermore, outputting a remaining life prediction value based on the heterogeneous life prediction model includes:

[0013] Extracting fatigue damage accumulation patterns from the spindle's historical operating data based on multiple application scenarios, and generating application fatigue correlation feedback as input to the heterogeneous parameter life prediction model;

[0014] According to the fatigue damage accumulation value and the application fatigue correlation feedback, combined with the time series analysis method, a fatigue damage accumulation curve in the current application scenario is established;

[0015] Based on the fatigue damage accumulation curve, the fatigue accumulation trend of the main shaft under various applications is simulated, and the fatigue critical point is predicted according to the trend direction and the trend intersection time sequence;

[0016] Integrate the fatigue critical points and calculate the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set, wherein the comprehensive remaining life is the predicted life under the current application;

[0017] According to the comprehensive remaining life, combined with the single application and multi-application superposition conditions of the main spindle, the remaining life prediction value is calculated and output.

[0018] Furthermore, the fatigue critical point is predicted based on the trend direction and the trend intersection time sequence, including:

[0019] Analyze the fatigue damage accumulation rules of various spindles under multiple application operating conditions, organize the trend according to the fatigue damage accumulation rules and set the fatigue critical point judgment conditions;

[0020] Establishing a fatigue evolution model based on the applied fatigue correlation feedback and the fatigue damage accumulation curve to simulate the nonlinear dynamic characteristics of the main shaft fatigue damage accumulation;

[0021] Taking the multi-scenario feature vector as input, analyzing the weights of fatigue influencing factors of the main shaft under various operating scenarios, and using the fatigue influencing factors as feedback to adjust the prediction accuracy of the fatigue evolution model;

[0022] According to the fatigue impact factor, for a plurality of the multi-scenario feature vectors, a timing of the main shaft reaching the fatigue critical point is predicted according to the trend and corresponding application characteristics to generate a simulation prediction result;

[0023] The simulation prediction results are compared with the fatigue critical point judgment conditions to perform dynamic correction and generate fatigue critical points for current applications and future applications.

[0024] Furthermore, the nonlinear dynamic characteristics of the spindle fatigue damage accumulation are simulated, including:

[0025] By structured analysis of the multi-source operation data, a fatigue damage accumulation simulation framework is constructed, and the multi-scenario feature vectors of the spindle operation are mapped to the fatigue damage accumulation simulation framework;

[0026] Simulate the nonlinear evolution of fatigue damage under multiple application superposition conditions in various operating scenarios, and describe the changes in coupling characteristics and fatigue accumulation rates in multiple applications;

[0027] Calculating a fatigue accumulation coefficient according to the fatigue accumulation rate to adjust the fatigue damage accumulation simulation framework;

[0028] The cumulative laws of the main shaft in various stages are integrated, and the simulation results of each stage are integrated to generate the nonlinear trend of fatigue damage accumulation throughout the life cycle;

[0029] By comparing the historical operating data of the spindle with the fatigue damage critical point judgment conditions, the accuracy of the simulation results is verified and the deviation is corrected.

[0030] Furthermore, a multi-application feature map is established, including:

[0031] Extracting key features related to the spindle operation state based on the multi-source operation data during the spindle operation process, and generating a multi-source operation characteristic matrix containing time series information;

[0032] Classifying the multi-source operating characteristic matrix and generating corresponding application category labels according to the various operating states;

[0033] Identify core characteristic parameters closely related to the spindle operating state based on a statistical correlation analysis method between the key characteristics;

[0034] A mapping is established to map the multi-source operating characteristic matrix to the application category label, a multi-application feature map is generated that can distinguish multiple operating applications, and the key features are converted into the multi-scenario feature vector.

[0035] Furthermore, converting the key features into the multi-scenario feature vector includes:

[0036] Applying a feature dimensionality reduction method to reduce redundant information of the key features and extracting main feature components that can characterize changes in multi-application characteristics;

[0037] Based on the multi-application feature map, encoding and combining the main feature components to generate application characteristic vectors corresponding to the application category labels;

[0038] Based on the changes in time series information during the operation of the main axis, the application feature vector is time-series fused using a time series sliding window method to generate a preliminary multi-scenario feature vector containing time correlation;

[0039] Smoothing transition changes between vectors is performed on the preliminary multi-scene feature vector to generate the multi-scene feature vector.

[0040] Furthermore, a multi-dimensional coupling fatigue analysis is performed on the multi-scenario feature vectors, including:

[0041] Extracting damage characteristic parameters that affect spindle fatigue damage based on the multi-scenario feature vectors, quantifying the contribution of the damage characteristic parameters to fatigue damage accumulation, and generating a multi-dimensional feature contribution matrix of spindle fatigue damage;

[0042] According to the multidimensional feature contribution matrix, a nonlinear interaction relationship is established to describe the plurality of damage characteristic parameters, so as to characterize the fatigue damage accumulation effect;

[0043] Based on the fatigue damage accumulation effect, the change in fatigue damage accumulation rate caused by application switching is quantified to generate a fatigue response curve of the spindle under dynamic operating conditions;

[0044] Based on the fatigue response curve and the multi-scenario characteristic vector, the fatigue damage accumulation trend of the main shaft in various operating scenarios is analyzed, the fatigue damage accumulation value is calculated, and the change trend of the fatigue accumulation rate is predicted and verified.

[0045] Furthermore, calculating the fatigue damage accumulation value includes:

[0046] Performing a step-by-step cumulative analysis on the multi-scenario feature vectors, and generating basic analysis data describing fatigue damage accumulation based on the changing trends under various operating conditions;

[0047] Dividing the operation process into multiple time intervals, performing independent damage assessment on the basic analysis data in each time interval, and generating interval fatigue damage accumulation values;

[0048] Adjust the cumulative weight of each time interval in the cumulative calculation process according to the changes in operating status to reflect the impact of various applications on the fatigue damage accumulation rate;

[0049] The cumulative weight and the interval fatigue damage cumulative value are combined to perform global cumulative integration to generate a fatigue damage cumulative value as a quantified result of the damage state in the current operation stage.

[0050] A life prediction system for a main shaft based on application analysis, the system comprising:

[0051] The modal perception module establishes a modal perception network, collects multi-source operation data during the spindle operation process, and generates a multi-source operation characteristic matrix;

[0052] The feature mapping module establishes a multi-application feature map based on the multi-source operation characteristic matrix and generates a multi-scenario feature vector describing the device operation status;

[0053] Fatigue analysis module, which performs multi-dimensional coupled fatigue analysis on multi-scenario feature vectors and outputs the fatigue damage accumulation value of the main shaft;

[0054] The life prediction module obtains and combines the historical motion data of the spindle according to the accumulated value of fatigue damage, establishes a heterogeneous parameter life prediction model, and outputs the remaining life prediction value based on the heterogeneous parameter life prediction model.

[0055] Furthermore, the lifespan prediction module includes:

[0056] The regularity statistics unit extracts fatigue damage accumulation patterns from historical spindle operation data based on various application scenarios and generates application fatigue correlation feedback as input to the heterogeneous parameter life prediction model;

[0057] The time series analysis unit establishes the fatigue damage accumulation curve under the current application scenario based on the fatigue damage accumulation value and application fatigue correlation feedback combined with the time series analysis method;

[0058] The critical prediction unit simulates the fatigue accumulation trend of the main shaft under various applications based on the fatigue damage accumulation curve, and predicts the fatigue critical point according to the trend direction and trend intersection time sequence;

[0059] Comprehensive calculation unit, integrating fatigue critical points, and calculating the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set. The comprehensive remaining life is the predicted life under the current application;

[0060] The scenario superposition unit calculates and outputs the remaining life prediction value based on the comprehensive remaining life and the main axis single application and multi-application superposition conditions.

[0061] The technical solution of the present invention can achieve the following technical effects:

[0062] It solves the problems of single data source, insufficient prediction accuracy under complex working conditions, difficulty in modeling nonlinear fatigue damage, and lack of dynamic feedback and correction capabilities in traditional spindle life prediction methods, and realizes accurate prediction of the comprehensive remaining life of the spindle under single working conditions and multiple working conditions.

[0063] 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

[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 any creative work.

[0065] Figure 1 The figure is a flow chart of the life prediction method of the main shaft based on application analysis;

[0066] Figure 2 Schematic diagram of the process for predicting fatigue critical point;

[0067] Figure 3 This is a schematic diagram of the structure of multi-application feature mapping;

[0068] Figure 4 Schematic diagram of the structure for multi-dimensional coupled fatigue analysis. DETAILED DESCRIPTION

[0069] The technical solutions in the embodiments of the present invention will be clearly and completely described 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.

[0070] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0071] Embodiment 1;

[0072] like Figure 1 As shown, the present application provides a life prediction method based on application analysis of the main shaft, the method comprising:

[0073] S10: Establish a modal perception network to collect multi-source operation data during the spindle operation process and generate a multi-source operation characteristic matrix;

[0074] S20: Based on the multi-source operation characteristic matrix, a multi-application feature map is established, and a multi-scenario feature vector describing the device operation status is generated;

[0075] S30: Perform multi-dimensional coupled fatigue analysis on the multi-scenario feature vectors and output the fatigue damage accumulation value of the main shaft;

[0076] S40: According to the fatigue damage accumulation value, the historical motion data of the spindle is acquired and combined to establish a heterogeneous parameter life prediction model, and a remaining life prediction value is output based on the heterogeneous parameter life prediction model.

[0077] Specifically, first, by placing sensors (such as acceleration sensors, temperature sensors, vibration sensors, etc.) at different positions of the spindle, multi-source operation data generated by the spindle during operation are collected in real time, including vibration signals, temperature changes, speed, load and other operation status indicators; the collected multi-source operation data are preprocessed (such as denoising, standardization), and a multi-source operation characteristic matrix is ​​generated according to the time series characteristics of the multi-source operation data. The multi-source operation characteristic matrix contains characteristic information under different operating conditions, such as vibration, temperature and other time series data, which can comprehensively describe the dynamic operation status of the spindle; based on the obtained multi-source operation characteristic matrix, data analysis methods (such as principal component analysis (PCA), cluster analysis, etc.) are used to classify different working conditions, extract key features related to the spindle operation status, establish multi-application feature mapping through key features, and map the feature vectors under different operation states. The multi-scenario feature vector is finally generated. The multi-scenario feature vector can describe the state of the spindle under different operating environments and reflect the working condition characteristics under multiple scenarios; a multi-dimensional coupled fatigue analysis is performed on the multi-scenario feature vector, and a fatigue damage accumulation model (such as Miner's law) is used in combination with the multi-scenario feature vector of the spindle to perform a quantitative analysis of fatigue damage. By analyzing the contribution of each operating state to the fatigue damage of the spindle, the fatigue damage accumulation value of the spindle is output, which can help determine the fatigue state of the spindle under the current working condition and provide basic data for the remaining life prediction; according to the fatigue damage accumulation value and combined with the historical motion data of the spindle, a heterogeneous parameter life prediction model is established. The heterogeneous parameter life model adopts multi-factor analysis, regression analysis and other methods, combined with fatigue damage, historical operating data and working condition characteristics, to accurately predict the remaining life of the spindle. Through the prediction results of the heterogeneous parameter model, the remaining life prediction value of the spindle under current and future working conditions is output.

[0078] The technical solution of the present invention solves the problems of traditional spindle life prediction methods such as single data source, insufficient prediction accuracy under complex working conditions, difficulty in modeling nonlinear fatigue damage, and lack of dynamic feedback and correction capabilities, and realizes accurate prediction of the comprehensive remaining life of the spindle under single working conditions and multiple working conditions.

[0079] Furthermore, the remaining life prediction value is output based on the heterogeneous life prediction model, including:

[0080] Extract fatigue damage accumulation patterns from historical spindle operation data based on various application scenarios, and generate application fatigue correlation feedback as input to the heterogeneous parameter life prediction model;

[0081] Based on the fatigue damage accumulation value and application fatigue correlation feedback, combined with the time series analysis method, the fatigue damage accumulation curve under the current application scenario is established;

[0082] Based on the fatigue damage accumulation curve, the fatigue accumulation trend of the main shaft under various applications is simulated, and the fatigue critical point is predicted according to the trend direction and the trend intersection time sequence;

[0083] Integrate the fatigue critical points and calculate the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set. The comprehensive remaining life is the predicted life under the current application;

[0084] According to the comprehensive remaining life, combined with the single application and multi-application superposition conditions of the spindle, the remaining life prediction value is calculated and output.

[0085] As a preferred embodiment of the above, first, historical operating data of the spindle in multiple application scenarios are collected and sorted. Based on these data, fatigue damage models (such as Miner's law or other nonlinear fatigue models) are used to extract the fatigue damage accumulation rules of the spindle under different working conditions (such as different loads, speeds, ambient temperatures, etc.). Through statistical analysis and pattern recognition, fatigue damage accumulation characteristics are generated for each application scenario, and application fatigue correlation feedback is generated as the input of the heterogeneous parameter life prediction model, reflecting the differences and regularities of spindle fatigue evolution in different application scenarios. According to the fatigue damage accumulation value and application fatigue correlation feedback, combined with time series analysis methods (such as sliding windows, time series modeling, etc.), a fatigue damage accumulation curve of the spindle in the current application scenario is established. The fatigue damage accumulation curve depicts the evolution trend of spindle fatigue damage with time or operating conditions under a specific application environment, can reflect the accumulation of fatigue damage in real time, and provide a basis for predicting future fatigue critical points. The fatigue damage accumulation curve is used to further simulate the fatigue damage accumulation change trend of the spindle in multiple application scenarios. By comparing and analyzing the fatigue damage accumulation curves of multiple application scenarios, the fatigue damage accumulation curves of multiple application scenarios are predicted. The method measures future trends in fatigue damage, particularly the fatigue critical point where trends converge (i.e., the critical moment when the spindle will suffer severe damage or failure). This process not only focuses on the local trends in each application scenario, but also analyzes the overall fatigue evolution of multiple applications to predict the occurrence of the fatigue critical point. After obtaining the fatigue critical points in different application scenarios, these critical points are integrated into an integrated point set. By analyzing each fatigue critical point and combining the changes in the spindle's operating conditions in current and future applications, the comprehensive remaining life of the spindle is calculated. The comprehensive remaining life refers to the predicted value of the spindle's remaining life considering the current application and multiple future application scenarios, and integrates the cumulative effect of fatigue damage on the spindle under different application conditions. Based on the integrated comprehensive remaining life, the remaining life prediction values ​​of the spindle are output under multiple application scenarios, combining the conditions of a single spindle application and the superposition of multiple applications. During this process, the prediction is dynamically adjusted according to changes in the actual operating environment and operating conditions to ensure that the prediction results can adapt to the actual fatigue evolution under different applications. In this way, the predicted value can not only reflect the life of the current application, but also adapt to the fatigue damage of the spindle in various possible future application environments.

[0086] Further, if Figure 2 As shown in the figure, the fatigue critical point is predicted based on the trend direction and the trend intersection time sequence, including:

[0087] Analyze the fatigue damage accumulation patterns of various spindles under multiple operating conditions, organize trends based on the fatigue damage accumulation patterns, and set fatigue critical point determination criteria;

[0088] A fatigue evolution model is established based on the application of fatigue correlation feedback and fatigue damage accumulation curve to simulate the nonlinear dynamic characteristics of spindle fatigue damage accumulation;

[0089] Taking multi-scenario feature vectors as input, the weights of fatigue influencing factors of the main shaft under various operating scenarios are analyzed, and the fatigue influencing factors are used as feedback to adjust the prediction accuracy of the fatigue evolution model;

[0090] Based on fatigue impact factors, for multiple multi-scenario feature vectors, the timing of the main axis reaching the fatigue critical point is predicted based on the trend trend and the corresponding application characteristics to generate simulation prediction results;

[0091] The simulation prediction results are compared with the fatigue critical point judgment conditions for dynamic correction to generate the fatigue critical points for current and future applications.

[0092] As a preferred embodiment of the above, the operating data of different spindles under various application conditions (such as different speeds, loads, temperatures, etc.) are collected and analyzed to extract the fatigue damage accumulation law of the spindle. For each application condition, the fatigue damage accumulation rate, damage characteristics and nonlinear characteristics of the accumulation process of the spindle are analyzed. These laws provide basic data for subsequent trend prediction. Based on the fatigue damage accumulation law under different working conditions, the fatigue damage development trend of each application scenario is further sorted out. Through trend analysis methods (such as regression analysis, Fourier analysis, etc.), the key change points and trend turning points in the fatigue damage accumulation process are identified. Combined with historical data and theoretical analysis, fatigue critical point judgment conditions are set. These conditions may include fatigue damage reaching a preset threshold, damage rate exceeding a critical value or a specific shape of the damage accumulation curve. Based on the extracted fatigue damage accumulation law, fatigue damage accumulation curve and set fatigue critical point judgment conditions, combined with fatigue correlation feedback, a fatigue damage evolution model of the spindle is established. The model should be able to simulate the nonlinear dynamic characteristics of fatigue damage of the spindle under multiple working conditions, taking into account working condition switching, load fluctuation and The impact of environmental changes on fatigue evolution is analyzed. Taking multi-scenario feature vectors as input, weight analysis is used to determine the weights of fatigue influencing factors (such as temperature, load, and speed) that affect spindle fatigue damage. Sensitivity analysis is used to quantify the contribution of each factor to fatigue damage accumulation. These fatigue influencing factors are used to dynamically adjust the fatigue evolution model to improve prediction accuracy. In particular, in application scenarios with large operating condition variations, real-time adjustment of the fatigue evolution model can reduce prediction deviations. Utilizing the extracted fatigue influencing factors and fatigue evolution model, the multi-scenario feature vectors under various operating scenarios are combined with trend trends to predict the timing of the spindle reaching the fatigue critical point under different application conditions. Specifically, by analyzing the cumulative trends under different operating conditions, the moment when fatigue damage reaches the critical value is identified, and corresponding simulation prediction results are generated. After the simulation prediction results are generated, they are compared with the set fatigue critical point judgment conditions and dynamically corrected based on the error size and deviation from the actual data. By comparing with the actual operating data, the fatigue evolution model is gradually revised to improve its prediction accuracy, ultimately generating the fatigue critical point for current and future applications.

[0093] Furthermore, the nonlinear dynamic characteristics of the spindle fatigue damage accumulation are simulated, including:

[0094] Through structured analysis of multi-source operation data, a fatigue damage accumulation simulation framework is constructed, and the multi-scenario feature vectors of the spindle operation are mapped to the fatigue damage accumulation simulation framework;

[0095] Simulate the nonlinear evolution of fatigue damage under multiple application superposition conditions in various operating scenarios, and describe the changes in coupling characteristics and fatigue accumulation rates in multiple applications;

[0096] According to the fatigue accumulation rate, the fatigue accumulation coefficient is calculated to adjust the fatigue damage accumulation simulation framework;

[0097] The cumulative laws of the main shaft in various stages are integrated, and the simulation results of each stage are integrated to generate the nonlinear trend of fatigue damage accumulation throughout the life cycle;

[0098] By comparing the historical operating data of the spindle with the fatigue damage critical point judgment conditions, the accuracy of the simulation results is verified and the deviations are corrected.

[0099] As a preferred embodiment of the above, the multi-source operating data of the spindle (such as vibration signal, temperature, load, speed, etc.) is subjected to structured analysis. The original data is converted into a format that meets the analysis requirements through data preprocessing, feature extraction and other methods. According to different fatigue damage models (such as Miner's law, nonlinear accumulation model, etc.), a fatigue damage accumulation simulation framework is constructed. The framework can be dynamically adjusted according to different working conditions of the spindle and adapt to the fatigue damage variation characteristics of the spindle under different application conditions. The multi-scenario feature vectors of the spindle are mapped to the framework so that the influence of different operating scenarios can be fully considered in the simulation; based on The fatigue damage accumulation simulation framework constructed is used to simulate the fatigue damage accumulation evolution process of the spindle under various operating scenarios. Taking into account the nonlinear characteristics of fatigue damage and the changes in the accumulation rate under different working conditions, nonlinear evolution models (such as dynamic system models, neural network models, etc.) are used to describe the changes in spindle fatigue damage. Specifically, multiple application superposition conditions need to be considered during the simulation process. The spindle is often in an environment where multiple working conditions are superimposed in actual operation. Therefore, it is necessary to simulate the superposition effect of fatigue damage and the interaction between various working conditions under multiple working conditions. The interaction between different application working conditions may affect the fatigue damage accumulation rate. For example, under high load High-speed operating conditions may cause higher fatigue damage rates than low-load operating conditions alone. Based on the fatigue accumulation rate, the parameters of the fatigue damage accumulation simulation framework are adjusted according to the current operating conditions at each operating stage to calculate the fatigue accumulation coefficient. The fatigue accumulation coefficient reflects the changing pattern of fatigue damage rates under different operating conditions. By dynamically adjusting the fatigue accumulation coefficient in the simulation framework, the fatigue damage evolution process of the spindle under different operating conditions can be more accurately described and the simulation accuracy can be improved. According to the changing trends under different operating conditions, the model is continuously optimized in combination with historical fatigue damage data. During the simulation process, the spindle goes through multiple operating stages, and the fatigue damage accumulation pattern of each stage may be different. By combining the fatigue damage accumulation results of each stage and integrating the accumulation patterns of each stage, a nonlinear trend of fatigue damage accumulation is generated for the entire life cycle of the spindle. The nonlinear trend of fatigue damage accumulation can describe the fatigue damage evolution path of the spindle at different life cycle stages. The accuracy of the simulation results is verified by comparing the fatigue damage accumulation trend with the historical operating data of the spindle. The historical operating data of the spindle can be used as a reference standard. By comparing with the actual fatigue damage conditions, the deviation of the simulation results is analyzed, and the model is corrected according to the fatigue damage conditions under actual operating conditions.

[0100] Further, if Figure 3 As shown, a multi-application feature map is established, including:

[0101] Based on the multi-source operation data during the spindle operation process, key features related to the spindle operation status are extracted to generate a multi-source operation characteristic matrix containing time series information;

[0102] Classify the multi-source operating characteristic matrix and generate corresponding application category labels based on the various operating states;

[0103] According to the statistical correlation analysis method between key features, the core characteristic parameters closely related to the spindle operation status are identified;

[0104] A mapping is established to map the multi-source running characteristic matrix to the application category label, a multi-application feature map is generated that can distinguish multiple running applications, and the key features are converted into multi-scenario feature vectors.

[0105] As a preferred embodiment of the above, key features closely related to the spindle operation state are extracted from the multi-source operation data of the spindle. The key features are feature information in the frequency domain, time domain and time-frequency domain extracted from the original data. The extracted features may include instantaneous frequency, vibration amplitude, energy distribution, instantaneous phase, etc., combined with the time series information of the spindle operation, and the key features are organized in chronological order to form a multi-source operation characteristic matrix. Each column of the multi-source operation characteristic matrix corresponds to a feature under a different operation state, and the row represents the value of each feature at a specific time point; the data in the multi-source operation characteristic matrix is ​​classified according to the working environment or working condition of the spindle, and different application category labels are generated. For example, according to different working conditions, the spindle operation can be divided into different categories such as normal operation, overload, abnormal vibration, high temperature, etc., and the data in different states are classified into different categories according to the change pattern of the operation data. Each application category label represents a specific state of the spindle operation (for example, operation under a specific load and speed condition); in the multi-source operation data of the spindle, some features may be closely related to the operation state of the spindle, while other features may have a weak relationship with fatigue damage or changes in working performance, so a unified classification is used. A correlation analysis method is used to identify and screen core features closely related to the health status of the spindle. Through correlation analysis, it is possible to identify which features have high correlation in different application states. As key features of the model input, they can effectively reflect the operating changes of the spindle under different working conditions. The multi-source operating characteristic matrix is ​​mapped to the application category label. The mapping process aims to distinguish different operating states and working conditions, output a multi-application feature map, and associate different operating states and working conditions with the operating characteristics of the spindle. During the training process, a mapping function with high discrimination is generated by learning the multi-source feature matrix and application labels. Based on the multi-application feature map, the identified key features are converted into multi-scenario feature vectors. Feature dimensionality reduction techniques (such as principal component analysis (PCA) and linear discriminant analysis (LDA)) are used to reduce the redundant information of the key features. The main components are extracted from the reduced key features to generate multi-scenario feature vectors that can characterize the changes in different operating application characteristics. Based on the diversity of application conditions and the changes in the spindle operating state, methods such as time series sliding windows are used to perform time series fusion on the multi-scenario feature vectors, which include the key characteristics of the spindle operation under different application conditions.

[0106] Furthermore, the key features are converted into multi-scenario feature vectors, including:

[0107] Apply feature dimensionality reduction methods to reduce redundant information of key features and extract main feature components that can characterize the changes in multi-application characteristics;

[0108] Based on the multi-application feature map, the main feature components are encoded and combined to generate application feature vectors corresponding to each application category label;

[0109] Based on the changes in time series information during the operation of the main axis, the time series sliding window method is used to perform time series fusion on the application feature vectors to generate preliminary multi-scenario feature vectors containing time correlation;

[0110] The transition changes between the preliminary multi-scene feature vectors are smoothed to generate a multi-scene feature vector.

[0111] As a preferred embodiment of the above embodiment, the key features extracted from multi-source operation data are reduced in dimensionality, and principal component analysis (PCA) is used to calculate the covariance matrix of the feature matrix, extract the principal components with high feature contribution rate, and retain the components whose cumulative contribution rate of feature variance reaches a set threshold, or linear discriminant analysis (LDA) is used to calculate the intra-class scatter and inter-class scatter matrix according to the application category label, optimize the classification interval, extract the discriminant features related to the category, and output the main feature components after dimensionality reduction; based on multi-application feature mapping, the main feature components are encoded and combined to generate application feature vectors; based on multi-application feature mapping, the main feature components are combined with the application category labels, and the application feature vectors are generated by encoding methods, a binary feature matrix is ​​created for each category label, and the main feature components are used as continuous value features and combined with the category label encoding to form a feature vector; the feature vector is normalized to ensure the consistency of the numerical range and avoid deviations caused by differences in feature scales. The output feature vector set covers the main axis operation status of different application scenarios; based on the changes in time series information, the time series sliding window method is used to The method is used to perform time series fusion. According to the time series changes of the application characteristic vector, the sliding window length and sliding step are set, and the weighted average of the characteristic vector in the window is calculated. The weight can be assigned according to the time distance or characteristic importance. The characteristic vectors in the window are accumulated and summed, the cumulative characteristics of the operating status are extracted, and the characteristic vector after window fusion is output. The time dynamic information is retained while eliminating short-term fluctuations. The generated preliminary multi-scenario feature vector contains time correlation and application status characteristics; the preliminary multi-scenario feature vector is smoothed and optimized to generate the final multi-scenario feature vector. The preliminary multi-scenario feature vector is smoothed, and the moving average method is used to calculate the mean of adjacent feature vectors to reduce short-term fluctuations, or the exponential smoothing method is used to weight the current characteristic vector and the historical smoothing result. The weight is controlled by the smoothing factor (such as 0.2). The smoothed vector is standardized to ensure that the characteristic value is within a unified range. The continuity and accuracy of the vector are confirmed by verifying and comparing the original data with the optimization results. The final generated multi-scenario feature vector fully describes the main shaft operating status and dynamic change trend under multiple application scenarios.

[0112] Further, if Figure 4 As shown in Figure 2, multi-dimensional coupled fatigue analysis is performed on multi-scenario feature vectors, including:

[0113] Based on multi-scenario feature vectors, the damage characteristic parameters that affect the main shaft fatigue damage are extracted, the contribution of the damage characteristic parameters to the fatigue damage accumulation is quantified, and a multi-dimensional feature contribution matrix of the main shaft fatigue damage is generated;

[0114] Based on the multidimensional characteristic contribution matrix, a nonlinear interaction relationship between multiple damage characteristic parameters is established to characterize the cumulative effect of fatigue damage.

[0115] Based on the fatigue damage accumulation effect, the change in fatigue damage accumulation rate caused by application switching is quantified, and the fatigue response curve of the spindle under dynamic operating conditions is generated;

[0116] Based on the fatigue response curve and multi-scenario characteristic vectors, the fatigue damage accumulation trend of the main shaft under various operating scenarios is analyzed, the fatigue damage accumulation value is calculated, and the changing trend of the fatigue accumulation rate is predicted and verified.

[0117] As a preferred embodiment of the above, based on the multi-scenario feature vector, characteristic parameters closely related to fatigue damage are extracted, including but not limited to operating indicators such as speed, vibration amplitude, temperature, and load. Through sensitivity analysis or correlation analysis, the contribution of each characteristic parameter to the accumulation of fatigue damage is calculated. For example, statistical methods (such as regression analysis, weight distribution model) are used to determine the weight values ​​of different parameters to fatigue damage, and a multidimensional feature contribution matrix is ​​constructed. The rows of the multidimensional feature contribution matrix represent the characteristic parameters in the multi-scenario feature vector, and the columns represent the influencing indicators of fatigue damage. The matrix element values ​​are the contributions of the characteristic parameters; machine learning methods (such as neural networks, support vector machines) or mathematical modeling methods (such as high-order polynomial fitting, nonlinear regression models) are used to establish an interaction relationship model between damage characteristic parameters to characterize the complex nonlinear effects between parameters. Based on the nonlinear interaction relationship, the comprehensive cumulative effect of multiple characteristic parameters on fatigue damage is simulated to describe the dynamics of damage accumulation. dynamic evolution process; according to the changes in operating characteristic parameters under different working conditions (such as load increase and speed fluctuation), the changes in fatigue damage accumulation rate caused by working condition switching are quantified, the amplitude and time constant of the damage rate change are calculated, and the fatigue response curve of the spindle under dynamic operating conditions is drawn by combining the damage rate change with time series data. The fatigue response curve describes the speed and trend of fatigue damage accumulation under different operating scenarios; based on the fatigue response curve, the fatigue damage accumulation trend of the spindle under various operating scenarios is analyzed, and the change pattern of the damage rate (such as accelerated accumulation or slowed accumulation) is extracted. According to the fatigue response curve, the accumulated damage in different time intervals is integrated, and the fatigue damage accumulation value is calculated as a quantitative indicator of fatigue damage under the current operating state; the future change trend of the fatigue damage accumulation rate is predicted using the time series analysis method, and the damage evolution path under future operating scenarios is obtained. The predicted results are compared with the actual operating data of the spindle to verify the accuracy of the prediction.

[0118] Furthermore, the calculation of fatigue damage accumulation includes:

[0119] Perform a step-by-step cumulative analysis of multi-scenario feature vectors, and generate basic analysis data describing fatigue damage accumulation based on the changing trends under various operating conditions;

[0120] The operation process is divided into multiple time intervals, and the basic analysis data in each time interval are independently evaluated for damage to generate interval fatigue damage accumulation values;

[0121] Adjust the cumulative weight of each time interval in the cumulative calculation process according to the changes in operating status to reflect the impact of various applications on the fatigue damage accumulation rate;

[0122] The cumulative weight and interval fatigue damage cumulative value are integrated globally to generate the fatigue damage cumulative value as the quantitative result of the damage state in the current operation stage.

[0123] As a preferred embodiment of the above, based on the multi-scenario feature vector, the fatigue damage change trend of the main shaft under various operating conditions is gradually analyzed, the characteristic parameters related to fatigue damage (such as speed, load, vibration amplitude, etc.) are extracted, and the basic analysis data is constructed in combination with the time series information. The contribution of each characteristic parameter to fatigue damage is gradually accumulated to obtain the basic analysis data describing the cumulative change of fatigue damage; the operation process is divided into multiple time intervals, and an independent damage assessment is performed on each time interval. According to the time series characteristics of the main shaft operation state, the entire operation process is divided into multiple time intervals (such as fixed-length time intervals or adaptive time intervals based on changes in the operation state). In each time interval, the basic analysis data is independently evaluated, and the fatigue damage accumulation value in the interval is quantified. The cumulative value of fatigue damage in each time interval is calculated using mathematical methods (such as time-weighted integration or the average value of characteristic parameters within the interval); the cumulative weight of each time interval is adjusted during the cumulative calculation process. The cumulative weight of different time intervals is calculated according to the changes in the operating state to reflect the different effects of the operating conditions of each interval on the fatigue damage accumulation rate. The cumulative weight can be dynamically adjusted based on factors such as the importance of characteristic parameters, the impact of working condition switching, and changes in load levels. For example, the weight of the high-load operating state can be set to a higher value, while the weight of the low-load operating state can be set to a lower value to accurately reflect the actual damage contribution; the cumulative weight and the interval fatigue damage cumulative value are combined to generate a global cumulative result. The fatigue damage cumulative value of each time interval is weighted according to the weight to generate a global cumulative result. The formula is:

[0124]

[0125] in, D total is the global cumulative value of fatigue damage,D i is the cumulative value of the i-th time interval, W i is the weight of the corresponding interval, N is the total number of time intervals, and the global cumulative value of fatigue damage is obtained by integrating the cumulative weight and the cumulative value of interval damage, which is used as the quantitative result of the fatigue state in the current operation stage.

[0126] Embodiment 2;

[0127] Based on the same inventive concept as the life prediction method based on the application analysis main axis in the aforementioned embodiment, the present invention further provides a life prediction system based on the application analysis main axis, the system comprising:

[0128] The modal perception module establishes a modal perception network, collects multi-source operation data during the spindle operation process, and generates a multi-source operation characteristic matrix;

[0129] The feature mapping module establishes a multi-application feature map based on the multi-source operation characteristic matrix and generates a multi-scenario feature vector describing the device operation status;

[0130] Fatigue analysis module, which performs multi-dimensional coupled fatigue analysis on multi-scenario feature vectors and outputs the fatigue damage accumulation value of the main shaft;

[0131] The life prediction module obtains and combines the historical motion data of the spindle according to the accumulated value of fatigue damage, establishes a heterogeneous parameter life prediction model, and outputs the remaining life prediction value based on the heterogeneous parameter life prediction model.

[0132] The above-mentioned adjustment system in the present invention can effectively implement the life prediction method of the main shaft based on application analysis, and the technical effects that can be achieved are as described in the above-mentioned embodiments and will not be repeated here.

[0133] Specifically, the lifespan prediction module includes:

[0134] The regularity statistics unit extracts fatigue damage accumulation patterns from historical spindle operation data based on various application scenarios and generates application fatigue correlation feedback as input to the heterogeneous parameter life prediction model;

[0135] The time series analysis unit establishes the fatigue damage accumulation curve under the current application scenario based on the fatigue damage accumulation value and application fatigue correlation feedback combined with the time series analysis method;

[0136] The critical prediction unit simulates the fatigue accumulation trend of the main shaft under various applications based on the fatigue damage accumulation curve, and predicts the fatigue critical point according to the trend direction and trend intersection time sequence;

[0137] Comprehensive calculation unit, integrating fatigue critical points, and calculating the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set. The comprehensive remaining life is the predicted life under the current application;

[0138] The scenario superposition unit calculates and outputs the remaining life prediction value based on the comprehensive remaining life and the main axis single application and multi-application superposition conditions.

[0139] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the corresponding optimization effects of the method in Example 1, which will not be repeated here.

[0140] Although the present application has been described with reference to specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and drawings are merely illustrative of the present application as defined herein and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, the present application is intended to include such modifications and variations as fall within the scope of the present application and its equivalents.

Claims

1. A life prediction method for a main shaft based on application analysis, characterized in that: The method comprises: Establish a modal perception network to collect multi-source operation data during the spindle operation and generate a multi-source operation characteristic matrix; Based on the multi-source operation characteristic matrix, a multi-application feature map is established, and a multi-scenario feature vector describing the operation status of the device is generated; Performing multi-dimensional coupled fatigue analysis on the multi-scenario feature vectors, and outputting a fatigue damage accumulation value of the main shaft; According to the fatigue damage accumulation value, obtaining and combining the historical operation data of the spindle, establishing a different parameter life prediction model, and outputting a remaining life prediction value based on the different parameter life prediction model, including: Extracting fatigue damage accumulation patterns from the spindle's historical operating data based on multiple application scenarios, and generating application fatigue correlation feedback as input to the heterogeneous parameter life prediction model; According to the fatigue damage accumulation value and the application fatigue correlation feedback, combined with the time series analysis method, a fatigue damage accumulation curve in the current application scenario is established; Based on the fatigue damage accumulation curve, the fatigue accumulation trend of the main shaft under various applications is simulated, and the fatigue critical point is predicted according to the trend direction and the trend intersection time sequence; Integrate the fatigue critical points and calculate the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set, wherein the comprehensive remaining life is the predicted life under the current application; According to the comprehensive remaining life, combined with the single application and multi-application superposition conditions of the main spindle, a remaining life prediction value is calculated and output; Build multi-application feature maps, including: Extracting key features related to the spindle operation state based on the multi-source operation data during the spindle operation process, and generating a multi-source operation characteristic matrix containing time series information; Classifying the multi-source operating characteristic matrix and generating corresponding application category labels according to the various operating states; Identify core characteristic parameters closely related to the spindle operating state based on a statistical correlation analysis method between the key characteristics; A mapping is established to map the multi-source operating characteristic matrix to the application category label, a multi-application feature map is generated that can distinguish multiple operating applications, and the key features are converted into the multi-scenario feature vector.

2. The life prediction method based on application analysis of main shaft according to claim 1 is characterized in that: Predict fatigue critical points based on trend direction and trend intersection timing, including: Analyze the fatigue damage accumulation rules of various spindles under multiple application operating conditions, organize the trend according to the fatigue damage accumulation rules and set the fatigue critical point judgment conditions; Establishing a fatigue evolution model based on the applied fatigue correlation feedback and the fatigue damage accumulation curve to simulate the nonlinear dynamic characteristics of the main shaft fatigue damage accumulation; Taking the multi-scenario feature vector as input, analyzing the weights of fatigue influencing factors of the main shaft under various operating scenarios, and using the fatigue influencing factors as feedback to adjust the prediction accuracy of the fatigue evolution model; According to the fatigue impact factor, for a plurality of the multi-scenario feature vectors, a timing of the main shaft reaching the fatigue critical point is predicted according to the trend and corresponding application characteristics to generate a simulation prediction result; The simulation prediction results are compared with the fatigue critical point judgment conditions to perform dynamic correction and generate fatigue critical points for current applications and future applications.

3. The life prediction method based on application analysis of main shaft according to claim 2 is characterized in that: Simulate the nonlinear dynamic characteristics of spindle fatigue damage accumulation, including: By structured analysis of the multi-source operation data, a fatigue damage accumulation simulation framework is constructed, and the multi-scenario feature vectors of the spindle operation are mapped to the fatigue damage accumulation simulation framework; Simulate the nonlinear evolution of fatigue damage under multiple application superposition conditions in various operating scenarios, and describe the changes in coupling characteristics and fatigue accumulation rates in multiple applications; Calculating a fatigue accumulation coefficient according to the fatigue accumulation rate to adjust the fatigue damage accumulation simulation framework; The cumulative laws of the main shaft in various stages are integrated, and the simulation results of each stage are integrated to generate the nonlinear trend of fatigue damage accumulation throughout the life cycle; By comparing the historical operating data of the spindle with the fatigue damage critical point judgment conditions, the accuracy of the simulation results is verified and the deviation is corrected.

4. The life prediction method based on application analysis of main shaft according to claim 1 is characterized in that: Converting the key features into the multi-scenario feature vector includes: Applying a feature dimensionality reduction method to reduce redundant information of the key features and extracting main feature components that can characterize changes in multi-application characteristics; Based on the multi-application feature map, encoding and combining the main feature components to generate application characteristic vectors corresponding to the application category labels; Based on the changes in time series information during the operation of the main axis, the application feature vector is time-series fused using a time series sliding window method to generate a preliminary multi-scenario feature vector containing time correlation; Smoothing transition changes between vectors is performed on the preliminary multi-scene feature vector to generate the multi-scene feature vector.

5. The life prediction method based on application analysis of main shaft according to claim 1 is characterized in that: Performing a multi-dimensional coupled fatigue analysis on the multi-scenario feature vectors includes: Extracting damage characteristic parameters that affect spindle fatigue damage based on the multi-scenario feature vectors, quantifying the contribution of the damage characteristic parameters to fatigue damage accumulation, and generating a multi-dimensional feature contribution matrix of spindle fatigue damage; According to the multidimensional feature contribution matrix, a nonlinear interaction relationship is established to describe the plurality of damage characteristic parameters, so as to characterize the fatigue damage accumulation effect; Based on the fatigue damage accumulation effect, the change in fatigue damage accumulation rate caused by application switching is quantified to generate a fatigue response curve of the spindle under dynamic operating conditions; Based on the fatigue response curve and the multi-scenario characteristic vector, the fatigue damage accumulation trend of the main shaft in various operating scenarios is analyzed, the fatigue damage accumulation value is calculated, and the change trend of the fatigue accumulation rate is predicted and verified.

6. The life prediction method based on application analysis of main shaft according to claim 5 is characterized in that: Calculating the fatigue damage accumulation value includes: Performing a step-by-step cumulative analysis on the multi-scenario feature vectors, and generating basic analysis data describing fatigue damage accumulation based on the changing trends under various operating conditions; Dividing the operation process into multiple time intervals, performing independent damage assessment on the basic analysis data in each time interval, and generating interval fatigue damage accumulation values; Adjust the cumulative weight of each time interval in the cumulative calculation process according to the changes in operating status to reflect the impact of various applications on the fatigue damage accumulation rate; The cumulative weight and the interval fatigue damage cumulative value are combined to perform global cumulative integration to generate a fatigue damage cumulative value as a quantified result of the damage state in the current operation stage.

7. The life prediction system of the main shaft based on application analysis is characterized by: According to the life prediction method of the main shaft based on application analysis as claimed in claim 1, the system includes: The modal perception module establishes a modal perception network, collects multi-source operation data during the spindle operation, and generates a multi-source operation characteristic matrix; The feature mapping module establishes a multi-application feature map based on the multi-source operation characteristic matrix and generates a multi-scenario feature vector describing the device operation status; Fatigue analysis module, which performs multi-dimensional coupled fatigue analysis on multi-scenario feature vectors and outputs the fatigue damage accumulation value of the main shaft; The life prediction module obtains and combines the historical operation data of the spindle according to the accumulated fatigue damage value to establish a heterogeneous life prediction model, and outputs the remaining life prediction value based on the heterogeneous life prediction model; The life prediction module includes: The regularity statistics unit extracts fatigue damage accumulation patterns from historical spindle operation data based on various application scenarios and generates application fatigue correlation feedback as input to the heterogeneous parameter life prediction model; The time series analysis unit establishes the fatigue damage accumulation curve under the current application scenario based on the fatigue damage accumulation value and application fatigue correlation feedback combined with the time series analysis method; The critical prediction unit simulates the fatigue accumulation trend of the main shaft under various applications based on the fatigue damage accumulation curve, and predicts the fatigue critical point according to the trend direction and trend intersection time sequence; Comprehensive calculation unit, integrating fatigue critical points, and calculating the comprehensive remaining life of the main shaft in current and future applications based on the integrated point set. The comprehensive remaining life is the predicted life under the current application; The scenario superposition unit calculates and outputs the remaining life prediction value based on the comprehensive remaining life and the main axis single application and multi-application superposition conditions.

Images