Intelligent tool holder monitoring method and system

Abstract

This invention relates to the field of mechanical monitoring technology, and in particular to an intelligent tool holder monitoring method and system. The method includes: acquiring at least two types of heterogeneous physical quantities from the tool holder structural state response in real time; extracting the damage-sensitive characteristics of each heterogeneous physical quantity and fusing them to generate a comprehensive structural degradation index; and outputting a tool holder health monitoring map corresponding to the processing stage when the comprehensive structural degradation index exceeds an adaptive threshold. This effectively solves the problems of limited information, difficulty in identifying early degradation, and lack of visualization of processing stages in existing electrode slitting tool holder monitoring, achieving accurate perception of structural degradation and intelligent output of a health status map.

Description

Technical Field

[0001] This invention relates to the field of mechanical monitoring technology, and in particular to an intelligent tool holder monitoring method and system. Background Technology

[0002] In the lithium battery manufacturing process, the electrode slitting process is a crucial step in ensuring the dimensional accuracy and edge quality of the electrodes. As a core structural component of the slitting equipment, the electrode slitting cutter holder is subjected to high-frequency reciprocating loads, dynamic impacts, and minute displacements, making it highly susceptible to structural deterioration issues such as loosening, cracking, or stiffness degradation. Once the cutter holder malfunctions, it can easily lead to electrode burrs, increased dust, or even equipment damage, severely impacting subsequent processes and product yield.

[0003] Currently, health monitoring of electrode slitting tool holders mostly relies on vibration signals or manual inspections, which has two main problems: First, the monitoring information is limited and it is difficult to accurately identify early structural deterioration. The characteristic dimensions of a single physical quantity are limited and it is difficult to reflect the true degradation behavior of the tool holder structure under complex coupled working conditions. Second, there is a lack of a status presentation mechanism for specific processing stages. Existing systems mostly use fixed threshold alarms and fail to generate intuitive health maps by combining the stage characteristics of the slitting process, making it difficult to support real-time diagnosis and maintenance decisions. Summary of the Invention

[0004] This invention provides an intelligent tool holder monitoring method and system, thereby effectively solving the problems pointed out in the background art.

[0005] This invention, by acquiring at least two types of heterogeneous physical quantities (such as vibration, displacement, and temperature) in real time and extracting their respective damage-sensitive features, can more comprehensively reflect the structural state changes of the electrode slitting tool holder under complex working conditions. This overcomes the limitations of traditional single-signal monitoring and effectively improves the ability to identify early structural loosening, cracks, and other minor deteriorations. When the comprehensive structural deterioration index exceeds the adaptive threshold, it can output a health monitoring map corresponding to the current processing stage. This allows operators not only to obtain alarm information indicating "abnormality" but also to intuitively grasp the current health status of the tool holder and its evolution trend, supporting more targeted maintenance and adjustment strategies and enhancing the practicality and intelligence of the monitoring system.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A smart tool holder monitoring method, comprising: Real-time acquisition of at least two types of heterogeneous physical quantities in the structural state response of the tool holder; Damage-sensitive features of each heterogeneous physical quantity are extracted and fused to generate a comprehensive structural degradation index; When the comprehensive index of structural deterioration exceeds the adaptive threshold, the tool holder health monitoring map for the corresponding processing stage is output.

[0007] Furthermore, the method also includes: mapping the output tool holder health monitoring map to the structural deterioration comprehensive index to a state level, and encoding the mapping result into a visual output format.

[0008] Furthermore, the damage-sensitive characteristics of each of the heterogeneous physical quantities are extracted and fused to generate a comprehensive structural degradation index, including: For each type of heterogeneous physical quantity, damage-sensitive features related to structural response changes are extracted; The extracted features are processed to be fusionable; Based on the discriminative ability of each type of damage-sensitive feature in historical damage state identification, the corresponding feature weights are determined. The standardized damage-sensitive features are weighted and fused with their corresponding weights to generate a comprehensive index for measuring the degree of structural degradation.

[0009] Furthermore, based on the discriminative power of each type of damage-sensitive feature in historical damage state identification, corresponding feature weights are determined, including: A labeled training sample set containing multiple typical degradation modes is constructed, and the sample set covers a multidimensional distribution of processing conditions, tool holder type and damage level; A multi-round feature evaluation and elimination mechanism is adopted, and the stability and discriminative ability scores of features are obtained based on cross-validation. An initial weight vector is generated based on the evaluation score, and redundant feature weights are suppressed. The initial weight vector is iteratively optimized on the validation set to minimize the discrimination error of the structural degradation index, thus forming the final fused weight configuration.

[0010] Furthermore, an initial weight vector is generated based on the evaluation score, and redundant feature weights are suppressed, including: The evaluation scores of each damage-sensitive feature are normalized into an initial weight vector; Construct a feature redundancy matrix based on the correlation coefficient between features to identify highly redundant feature pairs; For feature pairs with high redundancy, a regularization factor is introduced to compress their weights.

[0011] Furthermore, when the comprehensive structural deterioration index exceeds the adaptive threshold, a tool holder health monitoring map corresponding to the machining stage is output, including: Based on a preset processing stage identification algorithm and combined with real-time processing parameters, the current processing stage of the tool holder is determined. The corresponding atlas template for the processing stage is invoked, and combined with the current structural deterioration comprehensive index, to generate multi-dimensional atlas information reflecting the health status of the tool holder.

[0012] Furthermore, the atlas template construction process corresponding to the processing stage includes: Establish a feature vector library for the machining stage, wherein the feature vectors include cutting parameters, tool paths, and load spectrum. Based on historical health data clustering analysis, a set of graph configuration rules matching the feature vectors is generated; A dynamic mapping mechanism for map elements is defined to transform the comprehensive index of structural degradation into morphological parameters of the visualized elements.

[0013] Furthermore, it also includes: Real-time acquisition of work order data from the MES system; extraction of key production parameters affecting tool holder health. A health-production coupling model is established to dynamically map the comprehensive structural deterioration index into a health status code that can be recognized by the MES system. When the health status code reaches the preset warning level, a set of production decision instructions is output to the MES system.

[0014] A smart tool holder monitoring system, the system comprising: The physical quantity acquisition module acquires at least two types of heterogeneous physical quantities from the tool holder structure state response in real time. The index fusion module extracts the damage-sensitive features of each heterogeneous physical quantity and fuses them to generate a comprehensive structural degradation index. The graph output module outputs a tool holder health monitoring graph for the corresponding processing stage when the comprehensive index of structural deterioration exceeds the adaptive threshold.

[0015] Furthermore, the index fusion module includes: For each type of heterogeneous physical quantity, damage-sensitive features related to structural response changes are extracted; The extracted features are processed to be fusionable; Based on the discriminative ability of each type of damage-sensitive feature in historical damage state identification, the corresponding feature weights are determined. The standardized damage-sensitive features are weighted and fused with their corresponding weights to generate a comprehensive index for measuring the degree of structural degradation.

[0016] The technical solution of this invention can achieve the following technical effects: It effectively solves the problems of limited information, difficulty in identifying early degradation, and lack of visualization of processing stages in the existing electrode slitting tool holder monitoring, and realizes accurate perception of structural degradation and intelligent output of health status map. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an intelligent tool holder monitoring method. Figure 2 A flowchart illustrating the process of extracting damage-sensitive features and fusing them to generate a comprehensive structural degradation index; Figure 3 A flowchart illustrating the process of determining the corresponding feature weights; Figure 4 A schematic diagram illustrating the process of generating an initial weight vector and suppressing redundant feature weights; Figure 5 A flowchart illustrating the process of outputting tool holder health monitoring charts for the corresponding machining stages; Figure 6 This is a flowchart illustrating the process of constructing the atlas template corresponding to the processing stage. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1 like Figure 1 As shown, the present invention provides an intelligent tool holder monitoring method, the method comprising: S1: Real-time acquisition of at least two heterogeneous physical quantities in the structural state response of the tool holder; Specifically, the collected physical quantities may include, but are not limited to, vibration, strain, temperature, acoustic emission, electrical signal fluctuations, and sound energy. These response quantities originate from different physical mechanisms, possess complementarity and multidimensionality, and can more comprehensively reflect the structural changes and potential deterioration trends of the tool holder during dynamic machining. For example, vibration response is often used to detect problems such as loosening, impact, and structural resonance, and has sensitivity to dynamic anomalies; strain response reflects the stress distribution and local stiffness changes of the structure, which helps to identify microcracks and structural deformation; temperature response reveals the heat accumulation and local friction state during operation, and is often used to identify thermal fatigue or local overload; acoustic emission response is used to detect micro-destructive activities inside the material, and has characteristics such as high frequency and transient changes; electrical signal disturbances during machining can indirectly reflect changes in cutting load and mechanical transmission state through current or voltage fluctuations. The aforementioned physical quantities are synchronously acquired using a unified time reference. Combined with sliding window and short-time processing strategies, instantaneous values, rates of change, frequency distributions, and other dynamic response characteristics of each channel are extracted. To adapt to the characteristic changes at different processing stages, multiple processing parameter reference values ​​can be preset before acquisition to determine the current operating state and adjust the sampling frequency or processing method, making the acquisition process adaptable and targeted. In practical applications, when the tool holder is in a high-load slitting stage, certain physical quantities (such as vibration, temperature rise, and strain) will exhibit regular changes. If an increase in vibration response amplitude, a slow drift in strain signals, and a stable rise in temperature data are detected during this stage, it can be preliminarily determined that the tool holder may have entered a slightly deteriorated state. This method, based on the parallel acquisition of multiple physical quantities, allows structural anomalies to be identified promptly without relying on a single threshold, providing a rich information foundation for subsequent data fusion and health assessment.

[0022] S2: Extract the damage-sensitive features of each heterogeneous physical quantity and fuse them to generate a comprehensive structural degradation index; Specifically, since different physical response quantities reflect different aspects of the tool holder structure during processing—for example, vibration signals can reveal structural loosening or impact characteristics, strain response reflects stress concentration or local deformation, and temperature changes may indicate increased friction or thermal fatigue—it is difficult to fully perceive the true deterioration state of the structure by relying on a single signal. By extracting representative damage-sensitive features from various heterogeneous physical quantities, potential abnormal signs can be captured from multiple perspectives. Subsequently, these features are normalized and fusion weights are assigned based on their distinguishing ability in historical state identification, thus constructing a comprehensive structural deterioration index that reflects the overall degradation level. This comprehensive index, as a unified expression of the tool holder's operating state, not only solves the problems of inconsistent scales and different modes of change among multi-source information but also provides a basis for subsequent health status assessment, deterioration trend analysis, and graph visualization. This processing effectively enhances the system's ability to identify early structural damage, contributing to more accurate dynamic monitoring and maintenance decisions.

[0023] S3: When the comprehensive index of structural deterioration exceeds the adaptive threshold, output the tool holder health monitoring map corresponding to the processing stage.

[0024] Specifically, the structural degradation comprehensive index, as a health metric fused with multi-source features, reflects the operational stability and degradation degree of the tool holder structure. When the index exceeds a pre-set adaptive threshold, it means that the structural state may have transitioned from normal to mild, moderate, or even severe degradation. At this point, by outputting a health monitoring map that matches the current processing stage, the index changes can be linked to the processing rhythm, process parameters, and historical states, thereby achieving a clear presentation of the degradation state and trend tracing. The map format can be selected according to different stages, choosing appropriate display dimensions such as time trend curves, feature heatmaps, or level indicators, making the status information more intuitive. Through this process, not only is the interpretability of abnormal states enhanced, but health monitoring also gains stage perception and visual expression capabilities, providing effective support for achieving efficient operation and maintenance and intelligent decision-making.

[0025] This invention effectively solves the problems of limited information, difficulty in identifying early degradation, and lack of visualization of processing stages in existing electrode slitting tool holder monitoring, and achieves accurate perception of structural degradation and intelligent output of health status maps.

[0026] As a preferred embodiment of the above, the method further includes: mapping the output tool holder health monitoring map to the structural deterioration comprehensive index, and encoding the mapping result into a visual output format.

[0027] Specifically, after the structural deterioration comprehensive index breaks through the adaptive threshold and outputs the corresponding health monitoring map of the processing stage, a state level mapping and visualization coding mechanism is further introduced to achieve clear classification and intuitive presentation of the monitoring results. The process first establishes a mapping relationship between the comprehensive structural deterioration index and the structural health level. To this end, multiple health status level intervals can be defined based on historical working condition data and typical damage cases. For example, the comprehensive index can be divided into four levels: "normal," "slightly deteriorated," "moderately deteriorated," and "severely deteriorated." Alternatively, it can be expanded to a more refined status level division according to actual needs. These level intervals can be set in combination with the characteristics of the tool holder material, the evolution law of response characteristics, and the safety tolerance of the processing technology, and can be dynamically adjusted according to new data to achieve adaptive optimization of the mapping logic. After the level determination is completed, each level is mapped to a corresponding visual coding element. For example, different colors, icons, or layer depths are used to color or label the health monitoring spectrum: when it is determined to be "normal," the spectrum background is green; slightly deteriorated is yellow; moderately deteriorated is orange; and severely deteriorated is red. At the same time, arrows, flashing dots, or prompt symbols are superimposed on key areas of the spectrum to indicate the direction of change and the intensity of the trend, so as to enhance the intuitiveness and readability of the expression. In practical applications, if the comprehensive index of the cutter holder is in a critically rising state at a certain cutting stage, it is mapped to the "moderate deterioration" level and displayed in real time through specific color bands and markings on the graph, so that the operator can grasp the current health risk level as soon as possible and make a handling judgment in combination with the processing rhythm.

[0028] As a preferred embodiment of the above, such as Figure 2 As shown, step S2 involves extracting the damage-sensitive characteristics of each heterogeneous physical quantity and fusing them to generate a comprehensive structural degradation index, including: S21: For each type of heterogeneous physical quantity, extract damage-sensitive features related to changes in structural response; S22: Perform fusion-enhancing processing on the extracted features; S23: Determine the corresponding feature weights based on the discriminative ability of various damage-sensitive features in historical damage state identification; S24: The standardized damage-sensitive features are weighted and fused with their corresponding weights to generate a comprehensive index for measuring the degree of structural degradation.

[0029] Specifically, for each type of heterogeneous physical quantity, damage-sensitive features closely related to structural response changes are extracted. For example, for vibration signals, frequency domain energy ratio, root mean square acceleration value, and dominant frequency drift, representing energy changes and impact responses, can be extracted; for strain responses, indices such as strain amplitude, slope change, and hysteresis characteristics can be selected; and for temperature signals, features such as temperature rise rate, maximum thermal gradient, and thermal steady-state offset can be selected. These features were obtained through signal processing and time window analysis, demonstrating strong structural state characterization capabilities. After initial feature extraction, to overcome fusion obstacles such as scale inconsistencies and large statistical distribution differences among multi-source features, all features need to be fusion-compatible. Preferred methods include normalizing or standardizing each feature to ensure comparability on a uniform numerical scale; aligning the time dimensions of different acquisition channels to ensure each feature reflects the structural state during the same processing period; and correcting abnormal fluctuations by applying methods such as mean filtering and moving averages to improve data stability. Next, based on the performance of various damage-sensitive features in historical identification tasks, fusion weights are assigned to them to reflect the relative contribution of different features to the degradation judgment results. The weights can be initially set based on the distinguishing ability between features and state levels. Finally, the standardized damage-sensitive features and their corresponding weights are weighted and fused to generate a comprehensive structural degradation index. This index effectively measures the current overall structural health of the tool holder, with its value gradually increasing as damage worsens, exhibiting continuity, trend, and interpretability. In practical applications, when the tool holder is in a high-load operation phase, if multiple characteristic indicators deviate from the normal range simultaneously, the comprehensive index will rise rapidly, providing a basis for subsequent map generation and maintenance intervention.

[0030] As a preferred embodiment of the above, such as Figure 3 As shown, step S23 involves determining the corresponding feature weights based on the discriminative power of various damage-sensitive features in historical damage state identification, including: S231: Construct a labeled training sample set containing multiple typical degradation modes. The sample set covers a multidimensional distribution of processing conditions, tool holder types, and damage levels. S232: Employs a multi-round feature evaluation and elimination mechanism, and obtains feature stability and discriminative ability scores based on cross-validation. S233: Generate an initial weight vector based on the evaluation score and suppress redundant feature weights; S234: Iteratively optimize the initial weight vector on the validation set to minimize the discrimination error of the structural degradation index and form the final fused weight configuration.

[0031] Specifically, a labeled training sample set is first constructed, covering various typical tool holder degradation modes and possessing multi-dimensional feature distributions such as processing conditions, tool holder type, and damage level to ensure broad and representative sample coverage. Sample data sources can include long-term monitoring records from actual production processes, manually set controlled damage test data, and simulated loading data. Each sample includes the corresponding value of the structural degradation comprehensive index and a known state label for subsequent supervised evaluation. Based on this, a multi-round feature evaluation and elimination mechanism is employed to quantify the discriminative ability of various damage-sensitive features. During the evaluation process, a cross-validation strategy is used to repeatedly divide the training and test subsets, calculating the stability and discriminative ability of each feature in the state classification task across multiple rounds. Statistical scoring methods, such as information gain, mutual information, Fisher's discrimination rate, or tree-based importance scoring, are preferred to ensure sufficient interpretability and stability of the evaluation results. For features with large performance fluctuations or consistently low scores across multiple rounds of evaluation, the evaluation process is optimized. Features with low performance can be removed to reduce redundant noise interference in subsequent fusion calculations. Based on the above evaluation results, an initial weight vector containing all retained features is generated, with weight values ​​proportional to the discriminative ability score. To further improve the effectiveness of fusion, a redundancy suppression mechanism is introduced into the initial weight vector to avoid fusion imbalance caused by information overlap. Finally, based on the initial weight vector, it is iteratively optimized on the validation set. The optimization objective is to minimize the state discrimination error caused by the fused comprehensive index. Gradient descent, genetic algorithms, or Bayesian inference-based tuning methods can be used. In each iteration, the classification accuracy or mean square error of the fusion index under the current weight configuration is evaluated, and the weight distribution is adjusted according to the optimization objective until convergence is achieved. Finally, a stable, interpretable, and discriminative fusion weight configuration is obtained for subsequent calculation of the structural degradation comprehensive index.

[0032] As a preferred embodiment of the above, such as Figure 4 As shown, step S233 involves generating an initial weight vector based on the evaluation score and suppressing redundant feature weights, including: S2331: Normalize the evaluation scores of each damage-sensitive feature into an initial weight vector; S2332: Construct a feature redundancy matrix based on the correlation coefficient between features to identify highly redundant feature pairs; S2333: For feature pairs with high redundancy, a regularization factor is introduced to compress their weights.

[0033] Specifically, the discriminative ability scores obtained during the aforementioned feature evaluation process are first normalized to ensure all scores fall within a uniform numerical range. This normalization result serves as the basis for the initial weight vector. Methods such as maximum-minimum normalization or Z-score standardization can be used to ensure numerical comparability among features and prevent undue amplification of a particular feature due to its large score dimension during fusion. Subsequently, considering the potential for strong linear or nonlinear redundancy in some features, a feature correlation analysis step is introduced to avoid adverse effects on weight allocation due to information overlap. This involves calculating the pairwise correlation coefficients between features based on all sample data, such as Pearson correlation coefficient, Spearman rank correlation coefficient, or mutual information measure, and constructing a feature redundancy matrix accordingly. This matrix reflects the redundancy of each pair of features. The redundancy between features is assessed by setting a redundancy threshold to filter out highly redundant feature pairs. For identified redundant feature pairs, instead of directly removing a feature, a regularization factor is introduced to compress its component in the initial weight vector to reduce its impact on the fusion result. This compression can be achieved using L2 norm constraints, correlation-weighted penalty terms, or adaptive adjustment factors. For example, for a pair of highly correlated features, while keeping the total weight unchanged, the weights of the two features are redistributed according to their independent contributions, so that redundant information is reasonably weakened in the fusion process. In addition, if a feature exhibits high redundancy with multiple other features, its overall weight can be further reduced or multiple highly overlapping feature expressions can be merged through a weighted average method, thereby simplifying the feature structure and improving the compactness and interpretability of the fusion expression.

[0034] As a preferred embodiment of the above, such as Figure 5 As shown, in step S3, when the comprehensive structural deterioration index exceeds the adaptive threshold, the tool holder health monitoring map for the corresponding processing stage is output, including: S31: Based on a preset processing stage identification algorithm and combined with real-time processing parameters, determine the current processing stage of the tool holder; S32: Call the map template corresponding to the processing stage and combine it with the current structural deterioration comprehensive index to generate multi-dimensional map information reflecting the health status of the tool holder.

[0035] Specifically, the current working condition of the tool holder is first determined by a processing stage identification algorithm. This algorithm can be based on pre-set processing parameter thresholds, such as spindle speed, depth of cut, feed rate, and processing duration, combined with real-time acquired data to perform condition matching and judgment. For example, in a practical application, if the current processing parameters indicate that the tool holder is in a high-speed thin-material slitting stage, the system determines it to be in the "fine slitting" stage; if the parameters indicate that the tool holder is in a low-speed thick-material cutting state, it is determined to be in the "coarse slitting" stage. The above stage division can be pre-set according to rules or trained from historical processing data, and supports dynamic updates to adapt to different equipment types or tool holder configurations. After the processing stage identification is completed, the map template matching the stage is called. The board is a pre-built visualization layout unit that includes typical structural response characteristics, health level classification information, and graphic style parameters for this stage. Combined with the current comprehensive structural deterioration index, it is mapped to the health level partition in the template, and various auxiliary information is superimposed to generate a complete map. This map may include: the time-series change curve of the comprehensive structural deterioration index, the fluctuation trajectory diagram of key sensor features, health level indicator bars, abnormal trend markers, and the joint change relationship diagram between processing parameters and structural indicators. It can also dynamically adjust the map refresh frequency and display priority according to the processing rhythm control logic. For example, when the tool holder enters an unstable region, the index trend curve and health level indicator are enlarged first to guide the operator to pay attention to risk changes.

[0036] As a preferred embodiment of the above, such as Figure 6 As shown, the process of constructing the atlas template corresponding to the processing stage includes: A10: Establish a feature vector library for the machining stage. The feature vectors include cutting parameters, tool paths, and load spectrum. A20: Generate a set of graph configuration rules that match the feature vectors based on cluster analysis of historical health data; A30: Defines a dynamic mapping mechanism for atlas elements, which transforms the comprehensive index of structural deterioration into morphological parameters of visual elements.

[0037] Specifically, firstly, a feature vector library for each machining stage is established. This feature vector distinguishes different machining stages and serves as the prerequisite for calling atlas templates. The feature vector consists of multiple dimensions, including but not limited to typical cutting parameters such as spindle speed, depth of cut, and feed rate; tool path information such as tool trajectory and path change rate in space; and load spectrum data reflecting load variation patterns (such as average load, load fluctuation frequency, and load cycle count). These features can be extracted from actual machining data and categorized according to different working condition labels to form a digital representation of the machining stage. After obtaining sufficient historical machining samples, cluster analysis is performed based on archived health monitoring data. Clustering algorithms based on density or contour coefficient optimization (such as DBSCAN or K-means++) are preferred to divide machining states with similar feature vectors into several categories, forming typical machining stage patterns. In each clustering result, the evolution characteristics of its structural degradation comprehensive index and state response characteristics are statistically analyzed. Based on the actual health performance of this type of machining state, corresponding atlas configuration rules are extracted. This includes specifying which indicator curves, characteristic waveforms, color levels, anomaly identification methods, and layer arrangement logic should be displayed in the graph. For example, during the high-speed precision cutting stage, high-frequency vibration characteristic curves and temperature rise trends should be prioritized; while during the roughing stage, the graph should highlight strain response and load change ranges to improve diagnostic targeting. Finally, to ensure that the comprehensive structural degradation index can be accurately mapped to specific visual elements in different graph templates, a dynamic mapping mechanism for graph elements needs to be defined. This mechanism associates the index value range with the morphological parameters of the graph elements. For example, the comprehensive index can be mapped to color saturation, background color gradient, icon flashing frequency, bar chart length, curve thickness, or transparency. The preferred approach is to divide the index value into multiple level ranges and set corresponding visual parameter response ranges, so that the graph presents obvious visual changes under different degrees of degradation. For example, when the index is in the mild degradation range, the graph background is presented with a yellow gradient, accompanied by thicker waveform curves; if it is in the severe degradation range, the graph background turns red, and flashing warning signs and layer boundary intensification are superimposed to enhance the visual warning effect.

[0038] As a preferred embodiment of the above, it further includes: B10: Real-time acquisition of work order data from the MES system to extract key production parameters that affect tool holder health; B20: Establish a health-production coupling model to dynamically map the comprehensive structural deterioration index into a health status code that can be recognized by the MES system; B30: When the health status code reaches the preset warning level, output a set of production decision instructions to the MES system.

[0039] Specifically, the system first acquires work order information and production plan data from the MES system in real time via a data interface. This includes current product specifications, order batches, target output, planned processing rhythm, tool change schedule, and processing parameter settings. From this information, key production parameters closely related to tool holder health are extracted, such as processing frequency, number of tool changes per unit time, average processing time, tool type and number, and target tolerance level. These parameters are used to construct a set of health influencing factors. These factors directly reflect the impact of the current work order on the tool holder structure's load-bearing capacity, fatigue accumulation, and abnormal sensitivity, providing background support for subsequent health status assessments. Subsequently, based on the interaction between these production parameters and the comprehensive structural deterioration index, a health-production coupling model is established. This model maps the real-time monitored comprehensive structural deterioration index into a set of health status codes that can be identified and invoked by the MES system. The preferred approach is to establish a multi-level status code system, such as 0 representing "healthy and normal," 1 representing "mild warning," 2 representing "moderate abnormality," and 3 representing "severe deterioration requiring shutdown." Each status code level corresponds to a specific range of structural deterioration index values. The status code trigger boundaries are dynamically adjusted based on the current work order's process tolerance requirements and tool change strategy. For example, for high-precision, high-speed orders, the status code level threshold can be appropriately tightened to increase alertness; while for low-speed roughing tasks, more lenient trigger conditions can be set to improve tool holder utilization efficiency. When the comprehensive structural deterioration index evolves to a certain level and triggers the corresponding health status code, i.e. It can trigger the output of a set of control instructions to the MES system. The set of instructions includes, but is not limited to, strategic suggestions such as adjusting feed rate, prompting early tool change, pausing the current work order scheduling, switching to a standby tool holder, and reducing the machining load level. The specific content can be automatically combined according to different health status levels. For example, when the tool holder is detected to be in a moderately deteriorated state, the system can send a prompt to the MES to "keep the current batch running, but arrange for tool replacement in advance"; if it enters a severely deteriorated state, it can immediately output a high-priority shutdown instruction such as "terminate the current work order and switch to standby equipment" to prevent structural damage from causing quality accidents or equipment damage.

[0040] Example 2 Based on the same inventive concept as the intelligent tool holder monitoring method in the foregoing embodiments, the present invention also provides an intelligent tool holder monitoring system, the system comprising: The physical quantity acquisition module acquires at least two types of heterogeneous physical quantities from the tool holder structure state response in real time. The index fusion module extracts the damage-sensitive characteristics of each heterogeneous physical quantity and fuses them to generate a comprehensive structural degradation index. The graph output module outputs a tool holder health monitoring graph for the corresponding processing stage when the comprehensive structural deterioration index exceeds the adaptive threshold.

[0041] As a preferred embodiment of the above, the index fusion module includes: For each type of heterogeneous physical quantity, damage-sensitive features related to structural response changes are extracted; The extracted features are processed to be fusionable; Based on the discriminative power of various damage-sensitive features in historical damage state identification, the corresponding feature weights are determined. The standardized damage-sensitive features are weighted and fused with their corresponding weights to generate a comprehensive index for measuring the degree of structural degradation.

[0042] The monitoring system described above in this invention can effectively realize the intelligent tool holder monitoring method, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0043] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.

[0044] Although this application has been described in conjunction with specific features and embodiments, it will be apparent that various modifications and combinations can be made therein without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A smart tool holder monitoring method, characterized by, include: Real-time acquisition of at least two types of heterogeneous physical quantities in the structural state response of the tool holder; The damage-sensitive features of each heterogeneous physical quantity are extracted and fused to generate a comprehensive structural degradation index, including: for each type of heterogeneous physical quantity, extracting damage-sensitive features related to changes in structural response; The extracted features are processed to be fusionable; Based on the discriminative ability of each type of damage-sensitive feature in historical damage state identification, the corresponding feature weights are determined. The standardized damage-sensitive features are weighted and fused with their corresponding weights to generate a comprehensive structural degradation index for measuring the degree of structural degradation. When the comprehensive structural deterioration index exceeds the adaptive threshold, a tool holder health monitoring map corresponding to the machining stage is output, including: Based on a preset processing stage identification algorithm and combined with real-time processing parameters, the current processing stage of the tool holder is determined. The process of constructing the atlas template corresponding to the processing stage, and combining it with the current structural deterioration comprehensive index, generates multidimensional atlas information reflecting the health status of the tool holder. The process includes: Establish a feature vector library for the machining stage, wherein the feature vectors include cutting parameters, tool paths, and load spectrum. Based on historical health data clustering analysis, a set of graph configuration rules matching the feature vectors is generated; A dynamic mapping mechanism for map elements is defined to transform the comprehensive index of structural degradation into morphological parameters of the visualized elements.

2. The intelligent tool holder monitoring method according to claim 1, characterized in that, The method further includes: mapping the output tool holder health monitoring map to the structural deterioration comprehensive index to a state level, and encoding the mapping result into a visual output format.

3. The intelligent tool holder monitoring method according to claim 1, characterized in that, According to various types of damage sensitivity The discriminative power of sensory features in identifying historical damage states, and the determination of corresponding feature weights, including: A labeled training sample set containing multiple typical degradation modes is constructed, and the sample set covers a multidimensional distribution of processing conditions, tool holder type and damage level; A multi-round feature evaluation and elimination mechanism is adopted, and the stability and discriminative ability scores of features are obtained based on cross-validation. An initial weight vector is generated based on the evaluation score, and redundant feature weights are suppressed. The initial weight vector is iteratively optimized on the validation set to minimize the discrimination error of the structural degradation index, thus forming the final fused weight configuration.

4. The intelligent tool holder monitoring method according to claim 3, characterized in that, An initial weight vector is generated based on the evaluation score, and redundant feature weights are suppressed, including: The evaluation scores of each damage-sensitive feature are normalized into an initial weight vector; Construct a feature redundancy matrix based on the correlation coefficient between features to identify highly redundant feature pairs; For feature pairs with high redundancy, a regularization factor is introduced to compress their weights.

5. The intelligent tool holder monitoring method according to claim 1, characterized in that, Also includes: Real-time acquisition of work order data from the MES system; extraction of key production parameters affecting tool holder health. A health-production coupling model is established to dynamically map the comprehensive structural deterioration index into a health status code that can be recognized by the MES system. When the health status code reaches the preset warning level, a set of production decision instructions is output to the MES system.

6. An intelligent tool holder monitoring system, characterized in that, The system is used to perform the intelligent tool holder monitoring method as described in any one of claims 1-5, the system comprising: The physical quantity acquisition module acquires at least two types of heterogeneous physical quantities from the tool holder structure state response in real time. The index fusion module extracts the damage-sensitive features of each heterogeneous physical quantity and fuses them to generate a comprehensive structural degradation index. The graph output module outputs a tool holder health monitoring graph for the corresponding processing stage when the comprehensive index of structural deterioration exceeds the adaptive threshold.

Images