Machine tool monitoring method, system and device based on industrial internet of things

Abstract

This application discloses a machine tool monitoring method, system, and device based on the Industrial Internet of Things (IIoT), relating to the technical field of IIoT. The method includes: segmenting the vibration signal to obtain a set of vibration signal segments; performing empirical mode decomposition (EMD) on each vibration signal segment to obtain corresponding intrinsic mode function (IMF) components; for each machining cycle, constructing a corresponding component vibration coordination vector based on the IMF components and combining them to form a component vibration coordination matrix; determining a component vibration coordination index based on the component vibration coordination matrix; for each transmission component, constructing a corresponding component state vector based on the IMF components and combining them to form a component state matrix; determining a component performance index based on the component state matrix; and determining whether the target machine tool has malfunctioned based on the component vibration coordination index and the component performance index. This application improves the flexibility of machine tool monitoring.

Description

Technical Field

[0001] This application relates to the technical field of industrial Internet of Things (IIoT), and in particular to machine tool monitoring methods, systems, and devices based on IIoT. Background Technology

[0002] As core equipment in the manufacturing industry, machine tools' operating status directly affects production efficiency and product quality. Traditional machine tool status monitoring mainly relies on periodic manual inspections, experience-based judgment, or single-point sensor alarms with fixed thresholds. These methods have significant limitations: manual inspections are inefficient, dependent on personal experience, and cannot achieve continuous monitoring; alarms based on fixed thresholds lack flexibility, are difficult to adapt to dynamic operating conditions such as changes in machining tasks and tool wear, and are prone to false alarms or missed alarms.

[0003] With the development of Industrial Internet of Things (IIoT) technology, online monitoring via vibration sensors has become an important method. Currently, vibration signal-based monitoring methods mostly focus on the analysis of single key components (such as spindles) or single characteristic parameters (such as overall vibration intensity). While these methods can detect some faults, they often have the following shortcomings: First, they lack a comprehensive assessment of the collaborative working state of multiple transmission components in machine tools, making it difficult to identify systemic faults caused by misalignment between components. Second, feature extraction methods are not adaptable enough to non-stationary signals, potentially missing early and weak fault characteristics. Third, diagnostic logic is mostly based on single-index judgments, limiting the ability to identify complex fault modes, and the timeliness and accuracy of early warnings need to be improved.

[0004] Therefore, how to develop a machine tool monitoring method that can deeply integrate multi-source information, adaptively extract features, and achieve early and accurate warnings under the industrial Internet of Things architecture has become an urgent technical problem to be solved. Summary of the Invention

[0005] To improve the accuracy of machine tool monitoring, this application provides a machine tool monitoring method, system, and equipment based on the Industrial Internet of Things.

[0006] Firstly, this application provides a machine tool monitoring method based on the Industrial Internet of Things, employing the following technical solution:

[0007] A machine tool monitoring method based on the Industrial Internet of Things (IIoT) is applied to an IIoT system, which includes a management platform, a sensor network platform, and an object platform connected in sequence. The method is executed by the management platform and includes:

[0008] The vibration signals of each transmission component during each processing cycle are obtained during the current batch processing task of the target machine tool. The vibration signals are then segmented to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier. Each element corresponding to each index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window.

[0009] Empirical mode decomposition is performed on each of the vibration signal segments to obtain the corresponding intrinsic mode function components;

[0010] For each processing cycle, a corresponding component vibration coordination vector is constructed based on the intrinsic mode function components, and the vectors are combined to form a component vibration coordination matrix; and a component vibration coordination index is determined based on the component vibration coordination matrix.

[0011] For each transmission component, a corresponding component state vector is constructed based on the intrinsic mode function components, and the vectors are combined to form a component state matrix; and the component performance index is determined based on the component state matrix.

[0012] The determination of whether the target machine tool has malfunctioned is based on the component vibration coordination index and the component performance index.

[0013] By employing the above technical solution, the vibration signals of each transmission component during each processing cycle of the target machine tool in the current batch processing task are first acquired. The vibration signals are then segmented to obtain a set of vibration signal segments. This set is indexed by the transmission component identifier and the processing cycle identifier, with each index containing multiple vibration signal segments. Each vibration signal segment corresponds to a time slot window. Empirical mode decomposition (EMD) is then performed on each vibration signal segment to obtain the corresponding intrinsic mode function (EMF) components. For each processing cycle, a corresponding component vibration coordination vector is constructed based on the EMF components, and these vectors are combined to form a component vibration coordination matrix. Furthermore, component vibration coordination indices are determined based on the component vibration coordination matrix. Then, for each transmission component, a corresponding component state vector is constructed based on the EMF components, and these vectors are combined to form a component state matrix. Finally, component performance indices are determined based on the component state matrix. Finally, the component vibration coordination indices and component performance indices are used to determine whether the target machine tool has experienced a fault. In this invention, by performing time-series and component-based structured segmentation of vibration signals, the original data stream is transformed into analysis units with clear spatiotemporal identifiers, achieving effective organization and utilization of monitoring data. Then, the empirical mode decomposition method is applied to adaptively extract the intrinsic mode function components in the signal, thereby accurately separating and enhancing the characteristic information representing different fault sources. Subsequently, a component vibration coordination matrix and a component state matrix are constructed. From the two complementary dimensions of system coordinated working state and individual performance degradation trajectory, a comprehensive assessment of the machine tool health status is completed. Finally, by integrating the above two types of indicators for decision-making, it is possible not only to issue early warnings of system coordination disorder, but also to achieve highly reliable fault diagnosis through dual verification, thereby improving the accuracy and intelligence level of machine tool fault monitoring.

[0014] Optionally, the step of segmenting the vibration signal to obtain a set of vibration signal segments includes:

[0015] Analyze the CNC program that drives the target machine tool to execute the current batch machining task;

[0016] Based on the motion instructions and machining parameters in the CNC program, the machining process type for each machining stage is determined.

[0017] Based on the first preset table, according to the processing technology type, time slot lengths are allocated to each processing stage to obtain a variable-length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between the processing technology type data and the time slot length sequence data.

[0018] Based on the variable-length time slot timing sequence, the vibration signals corresponding to each transmission component in each processing cycle are segmented to obtain a set of vibration signal segments.

[0019] By adopting the above technical solution, in order to obtain a set of vibration signal segments, the CNC program that drives the target machine tool to execute the current batch of processing tasks is analyzed. Then, based on the motion instructions and processing parameters in the CNC program, the processing technology type of each processing stage in the processing process is determined. Then, based on the first preset table, the time slot length is allocated to each processing stage according to the processing technology type to obtain a variable length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between processing technology type data and time slot length sequence data. Then, based on the variable length time slot sequence, the vibration signal corresponding to each transmission component in each processing cycle is segmented to obtain a set of vibration signal segments.

[0020] Optionally, the step of constructing a corresponding component vibration cooperative vector based on the intrinsic mode function components for each processing cycle includes:

[0021] For each processing cycle, the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle are obtained;

[0022] For each time slot window, the corresponding process physical characteristics are determined based on the motion command and processing parameters corresponding to that time slot window.

[0023] Based on the second preset table, the target frequency range of each time slot window is determined according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between the process physical characteristic data and the frequency range data.

[0024] Based on the target frequency range, determine the effective components corresponding to each time slot window from the target intrinsic mode function components;

[0025] Feature extraction is performed on each of the effective components to obtain the corresponding candidate state features;

[0026] Based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features;

[0027] Based on the merged features, a component vibration coordination vector is generated.

[0028] By adopting the above technical solution, in order to construct the component vibration cooperative vector, for each processing cycle, the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle are obtained. Then, for each time slot window, the corresponding process physical characteristics are determined according to the motion command and processing parameters corresponding to that time slot window. Then, based on the second preset table, the target frequency range of each time slot window is determined according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between process physical characteristic data and frequency range data. Then, based on the target frequency range, the effective components corresponding to each time slot window are determined from the target intrinsic mode function components. Then, feature extraction is performed on each effective component to obtain the corresponding candidate state features. Then, based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features. Finally, based on the merged features, the component vibration cooperative vector is generated.

[0029] Optionally, the step of linearly merging the candidate state features based on Bayes' theorem to obtain merged features includes:

[0030] For a single time slot window, a prior probability distribution model corresponding to each of the candidate state features is obtained, and the likelihood probability of each of the candidate state features in each time slot window is determined based on the prior probability distribution model, wherein the prior probability distribution model is obtained based on historical normal vibration signal samples.

[0031] Based on the Bayesian probability formula, the likelihood probability of each candidate state feature is converted into a posterior probability, and the posterior probability is normalized to obtain the fusion weight corresponding to each candidate state feature.

[0032] Based on the fusion weight, the candidate state features within the current time slot window are linearly merged to obtain the merged features corresponding to the current time slot window.

[0033] By adopting the above technical solution, in order to obtain the merged features, for a single time slot window, the prior probability distribution model corresponding to each candidate state feature is obtained, and the likelihood probability of each candidate state feature in each time slot window is determined according to the prior probability distribution model. The prior probability distribution model is obtained based on historical normal vibration signal samples. Then, based on the Bayesian probability formula, the likelihood probability of each candidate state feature is converted into a posterior probability, and the posterior probability is normalized to obtain the fusion weight corresponding to each candidate state feature. Then, according to the fusion weight, the candidate state features in the current time slot window are linearly merged to obtain the merged features corresponding to the current time slot window.

[0034] Optionally, the step of generating a component vibration cooperative vector based on the merged features includes:

[0035] For each processing cycle, each time slot window within that processing cycle is grouped according to the corresponding processing technology type to obtain at least one technology type group;

[0036] For each process type group, statistical features are extracted from the merging features of each time slot window within the group to obtain group statistical features;

[0037] The statistical features of each process type group are combined in a preset order to form the component vibration coordination vector corresponding to the processing cycle.

[0038] By adopting the above technical solution, in order to generate the component vibration coordination vector, for each processing cycle, each time slot window in the processing cycle is grouped according to the corresponding processing technology type to obtain at least one process type group. Then, for each process type group, the combined features of each time slot window in the group are statistically extracted to obtain group statistical features. Then, the group statistical features of each process type group are combined according to a preset arrangement order to form the component vibration coordination vector corresponding to the processing cycle.

[0039] Optionally, the step of determining the component vibration coordination index based on the component vibration coordination matrix includes:

[0040] Based on the historical normal vibration signal samples, a corresponding reference vibration cooperative vector is constructed and combined to form a reference vibration cooperative matrix;

[0041] Principal component analysis was performed on the reference vibration coordination matrix to obtain the principal component vector;

[0042] Calculate the cosine similarity between each row vector in the component vibration coordination matrix and the principal component vector to obtain a similarity sequence;

[0043] The information entropy of the similarity sequence is calculated, and the information entropy is used as a component vibration coordination index.

[0044] By adopting the above technical solution, in order to determine the component vibration coordination index, a corresponding benchmark vibration coordination vector is constructed based on historical normal vibration signal samples, and combined to form a benchmark vibration coordination matrix. Then, principal component analysis is performed on the benchmark vibration coordination matrix to obtain the principal component vector. Then, the cosine similarity between each row vector in the component vibration coordination matrix and the principal component vector is calculated to obtain a similarity sequence. Then, the information entropy of the similarity sequence is calculated, and the information entropy is used as the component vibration coordination index.

[0045] Optionally, the step of constructing the corresponding component state vector based on the intrinsic mode function components includes:

[0046] For each transmission component, for each processing cycle, feature extraction is performed on the target intrinsic mode function components to obtain the state features of the transmission component corresponding to each time slot window in that processing cycle.

[0047] For each transmission component, for different processing cycles, the state characteristics of a single transmission component corresponding to all time slot windows in a single processing cycle are statistically aggregated to obtain the periodic state characteristics of the transmission component in that processing cycle.

[0048] For each transmission component, the periodic state characteristics of the transmission component in each processing cycle are arranged according to the time sequence of the processing cycle to generate a component state vector.

[0049] By adopting the above technical solution, in order to construct the corresponding component state vector, for each transmission component, for each processing cycle, feature extraction is performed on the target intrinsic mode function components to obtain the state features of the transmission component corresponding to each time slot window in the processing cycle. Then, for each transmission component, for different processing cycles, the state features corresponding to all time slot windows of a single transmission component in a single processing cycle are statistically aggregated to obtain the periodic state features of the transmission component in the processing cycle. Finally, for each transmission component, the periodic state features of the transmission component in each processing cycle are arranged according to the time order of the processing cycle to generate the component state vector.

[0050] Optionally, the step of determining the component performance index based on the component state matrix includes:

[0051] Based on the historical normal vibration signal samples, a corresponding reference component state matrix is ​​constructed;

[0052] Principal component analysis is performed on the state matrix of the reference component to obtain the set of reference principal component vectors and the variance contribution rate corresponding to each reference principal component vector;

[0053] Principal component analysis is performed on the component state matrix to obtain the target principal component vector set;

[0054] Calculate the cosine of the angle between the principal component vectors with corresponding indices in the benchmark principal component vector set and the target principal component vector set respectively to obtain the similarity of m principal components, where m is a positive integer determined according to a preset cumulative variance contribution rate threshold;

[0055] The principal component similarities are weighted and fused using the variance contribution rate as the weight to obtain a weighted similarity, and the component performance indicators are determined based on the weighted similarity.

[0056] By adopting the above technical solution, in order to determine the component performance indicators, a corresponding benchmark component state matrix is ​​constructed based on historical normal vibration signal samples. Then, principal component analysis is performed on the benchmark component state matrix to obtain the benchmark principal component vector set and the variance contribution rate corresponding to each benchmark principal component vector. Then, principal component analysis is performed on the component state matrix to obtain the target principal component vector set. Then, the cosine value of the angle between the principal component vectors with corresponding indices in the benchmark principal component vector set and the target principal component vector set is calculated to obtain m principal component similarities, where m is a positive integer determined according to a preset cumulative variance contribution rate threshold. Then, the principal component similarities are weighted and fused with variance contribution rate as the weight to obtain a weighted similarity, and the component performance indicators are determined based on the weighted similarity.

[0057] Secondly, this application also provides a machine tool monitoring system based on the Industrial Internet of Things, which adopts the following technical solution:

[0058] A machine tool monitoring system based on the Industrial Internet of Things (IIoT) includes a management platform, a sensor network platform, and an object platform that are sequentially connected in communication. The management platform is configured with:

[0059] The vibration signal segmentation module is used to acquire the vibration signals of each transmission component in each processing cycle during the current batch processing task of the target machine tool, and to segment the vibration signals to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier. Each element corresponding to each index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window.

[0060] The empirical mode decomposition module is used to perform empirical mode decomposition on each of the vibration signal segments to obtain the corresponding intrinsic mode function components;

[0061] The component vibration coordination module is used to construct a corresponding component vibration coordination vector based on the intrinsic mode function components for each processing cycle, and combine them to form a component vibration coordination matrix; and to determine the component vibration coordination index based on the component vibration coordination matrix.

[0062] The component state module is used to construct a corresponding component state vector for each transmission component based on the intrinsic mode function components, and combine them to form a component state matrix; and to determine the component performance index based on the component state matrix.

[0063] The fault determination module is used to determine whether the target machine tool has malfunctioned based on the component vibration coordination index and the component performance index.

[0064] Thirdly, this application also provides a computer device, which adopts the following technical solution:

[0065] A computer device includes a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the computer program to implement the method described in the first aspect.

[0066] In summary, this application includes at least the following beneficial technical effects: First, the vibration signals of each transmission component during each processing cycle of the target machine tool in the current batch processing task are acquired, and the vibration signals are segmented to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier, and each index corresponds to an element containing multiple vibration signal segments. Each vibration signal segment corresponds to a time slot window. Then, empirical mode decomposition is performed on each vibration signal segment to obtain the corresponding intrinsic mode function components. For each processing cycle, a corresponding component vibration coordination vector is constructed based on the intrinsic mode function components, and these vectors are combined to form a component vibration coordination matrix. Furthermore, a component vibration coordination index is determined based on the component vibration coordination matrix. Then, for each transmission component, a corresponding component state vector is constructed based on the intrinsic mode function components, and these vectors are combined to form a component state matrix. Finally, a component performance index is determined based on the component state matrix. Finally, whether the target machine tool has malfunctioned is determined based on the component vibration coordination index and the component performance index. In this invention, by performing time-series and component-based structured segmentation of vibration signals, the original data stream is transformed into analysis units with clear spatiotemporal identifiers, achieving effective organization and utilization of monitoring data. Then, the empirical mode decomposition method is applied to adaptively extract the intrinsic mode function components in the signal, thereby accurately separating and enhancing the characteristic information representing different fault sources. Subsequently, a component vibration coordination matrix and a component state matrix are constructed. From the two complementary dimensions of system coordinated working state and individual performance degradation trajectory, a comprehensive assessment of the machine tool health status is completed. Finally, by integrating the above two types of indicators for decision-making, it is possible not only to issue early warnings of system coordination disorder, but also to achieve highly reliable fault diagnosis through dual verification, thereby improving the accuracy and intelligence level of machine tool fault monitoring. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of the overall process of an embodiment of this application.

[0068] Figure 2 This is a structural diagram of one application scenario of the system in this application embodiment.

[0069] Figure 3 This is a structural diagram of another application scenario of the system according to an embodiment of this application.

[0070] Figure 4 This is a structural block diagram of the computer device described in this application. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0072] This application discloses a machine tool monitoring method based on the Industrial Internet of Things (IIoT).

[0073] Reference Figure 1 A machine tool monitoring method based on the Industrial Internet of Things (IIoT) is applied to an IIoT system. The IIoT system includes a management platform, a sensor network platform, and an object platform that are sequentially connected via communication. The method is executed by the management platform and includes:

[0074] Step S11: Obtain the vibration signals of each transmission component in each processing cycle during the current batch processing task of the target machine tool, and segment the vibration signals to obtain a set of vibration signal segments.

[0075] The vibration signal segment set is indexed by the transmission component identifier and the processing cycle identifier. Each index corresponds to an element containing multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window.

[0076] It should be noted that in step S11, the management platform obtains the vibration sensing signals of each transmission component of the target machine tool on the object platform during multiple processing cycles when the target machine tool is performing the current batch processing task through the sensor network platform. The management platform performs initial segmentation of the original vibration signal according to the processing cycle, and further divides the signal in each cycle into multiple consecutive time slot windows in the time domain, thereby generating a set of vibration signal segments with the transmission component identifier and the processing cycle identifier as dual indices. Each element under the index contains multiple vibration signal segments corresponding to different time slots. This structured data organization method meets the requirements of the refined time-frequency analysis and state tracking in subsequent steps.

[0077] Step S12: Perform empirical mode decomposition on each vibration signal segment to obtain the corresponding intrinsic mode function components.

[0078] It should be noted that in step S12, the management platform performs empirical mode decomposition (EMD) on each vibration signal segment. Empirical mode decomposition is an adaptive time-frequency analysis method that can decompose the non-stationary original vibration signal into a series of intrinsic mode function components arranged from high frequency to low frequency, representing the intrinsic modes of the signal, thereby effectively separating vibration components originating from different physical processes or fault sources.

[0079] Step S13: For each processing cycle, construct the corresponding component vibration coordination vector based on the intrinsic mode function components, and combine them to form the component vibration coordination matrix; and determine the component vibration coordination index based on the component vibration coordination matrix.

[0080] It should be noted that in step S13, for each machining cycle, the management platform constructs a feature vector reflecting the overall vibration mode of each component in the current cycle, i.e., a component vibration coordination vector, based on the intrinsic mode function components of all transmission components within that cycle. Then, the coordination vectors of all components within the same cycle are combined to form the component vibration coordination matrix for that cycle. This matrix encodes the dynamic coupling relationship between the transmission components of the machine tool at a specific machining moment. The management platform calculates a comprehensive component vibration coordination index by analyzing the stability, eigenvalue distribution, or correlation between internal vectors of this matrix. This index reflects the overall health of the transmission chain's coordinated operation at the system level and is sensitive to systemic degradation caused by misalignment, abnormal load, etc.

[0081] Step S14: For each transmission component, construct the corresponding component state vector based on the intrinsic mode function components, and combine them to form a component state matrix; and determine the component performance index based on the component state matrix.

[0082] It should be noted that in step S14, for each transmission component, the management platform constructs a feature vector reflecting the component's own state for each cycle based on its intrinsic mode function components in multiple consecutive processing cycles. This feature vector is called the component state vector. Arranging these state vectors in chronological order forms the component state matrix, which contains information about the evolution trajectory of the component state over time. By analyzing this matrix, such as calculating the temporal trend of key features or measuring the distance between the current state and the historical normal state benchmark, the management platform determines the component performance index. This index directly reflects the development degree of local defects such as bearing wear and gear pitting.

[0083] Step S15: Determine whether the target machine tool has malfunctioned based on the component vibration coordination index and component performance index.

[0084] It should be noted that in step S15, the management platform comprehensively considers the vibration coordination index of the components and the performance index of each transmission component. Through decision-making mechanisms such as rule engines, fuzzy logic, or machine learning classifiers, it performs information fusion and joint judgment on the two types of indicators. For example, when the performance index of a component shows its own condition deteriorating, and at the same time, the system-level coordination index also shows anomalies, a fault related to that component can be determined with high confidence. Furthermore, when the coordination index first shows a slight shift while all individual performance indicators are still within the normal threshold, an early warning of systemic imbalance can be achieved. This method, through dual verification and complementary analysis, significantly improves the accuracy, reliability, and timeliness of fault diagnosis and early warning.

[0085] In the above implementation, vibration signals of each transmission component during each processing cycle of the target machine tool are first acquired during the execution of the current batch processing task. These vibration signals are then segmented to obtain a set of vibration signal segments. Each segment is indexed by the transmission component identifier and the processing cycle identifier, with each index containing multiple vibration signal segments. Each segment corresponds to a time slot window. Empirical mode decomposition (EMD) is then performed on each segment to obtain the corresponding intrinsic mode function (IMF) components. For each processing cycle, a corresponding component vibration coordination vector is constructed based on the IMF components, and these vectors are combined to form a component vibration coordination matrix. Furthermore, a component vibration coordination index is determined based on the component vibration coordination matrix. Then, for each transmission component, a corresponding component state vector is constructed based on the IMF components, and these vectors are combined to form a component state matrix. Finally, a component performance index is determined based on the component state matrix. Finally, the component vibration coordination index and component performance index are used to determine whether the target machine tool has malfunctioned. In this invention, by performing time-series and component-based structured segmentation of vibration signals, the original data stream is transformed into analysis units with clear spatiotemporal identifiers, achieving effective organization and utilization of monitoring data. Then, the empirical mode decomposition method is applied to adaptively extract the intrinsic mode function components in the signal, thereby accurately separating and enhancing the characteristic information representing different fault sources. Subsequently, a component vibration coordination matrix and a component state matrix are constructed. From the two complementary dimensions of system coordinated working state and individual performance degradation trajectory, a comprehensive assessment of the machine tool health status is completed. Finally, by integrating the above two types of indicators for decision-making, it is possible not only to issue early warnings of system coordination disorder, but also to achieve highly reliable fault diagnosis through dual verification, thereby improving the accuracy and intelligence level of machine tool fault monitoring.

[0086] As a further implementation of the method, the step of segmenting the vibration signal to obtain a set of vibration signal segments includes:

[0087] Step S21: Analyze the CNC program that drives the target machine tool to execute the current batch machining task.

[0088] Step S22: Based on the motion instructions and machining parameters in the CNC program, determine the machining process type for each machining stage during the machining process.

[0089] Step S23: Based on the first preset table, allocate time slot lengths for each processing stage according to the processing technology type to obtain a variable length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between processing technology type data and time slot length sequence data.

[0090] It should be noted that the core of step S23 lies in implementing an adaptive signal segmentation strategy. The first preset table is pre-established based on a large number of process experiments and data analysis. It stores the mapping relationship between different machining process types (such as fine milling, rough boring, high-speed drilling, tapping, etc.) and the optimal analysis time slot length. For example, for rough machining with drastic changes in vibration characteristics or tapping processes with obvious intermittent impacts, a shorter time slot length is mapped to capture transient details. For fine machining processes with stable vibration, a longer time slot is mapped to stably extract periodic features. Through table lookup matching, the system can dynamically generate a set of variable-length time slot sequences with non-fixed time slot lengths for machining cycles containing different process stages. This design breaks the limitations of traditional fixed time window segmentation, making subsequent signal segments optimally matched with the current actual physical machining process in the time domain scale. This ensures that the vibration modes contained in each signal segment are more physically consistent, thereby providing higher quality and more targeted input for subsequent empirical mode decomposition, effectively improving the accuracy of feature extraction and the sensitivity of fault diagnosis.

[0091] Step S24: Based on the variable length time slot timing sequence, the vibration signals corresponding to each transmission component in each processing cycle are segmented to obtain a set of vibration signal segments.

[0092] In the above embodiments, in order to obtain a set of vibration signal segments, the CNC program that drives the target machine tool to execute the current batch of processing tasks is parsed. Then, based on the motion instructions and processing parameters in the CNC program, the processing technology type of each processing stage in the processing process is determined. Then, based on a first preset table, the time slot length is allocated to each processing stage according to the processing technology type to obtain a variable length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between processing technology type data and time slot length sequence data. Then, based on the variable length time slot sequence, the vibration signal corresponding to each transmission component in each processing cycle is segmented to obtain a set of vibration signal segments.

[0093] As a further implementation of the method, the step of constructing a corresponding component vibration cooperative vector based on the intrinsic modal function components for each processing cycle includes:

[0094] Step S31: For each processing cycle, obtain the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle.

[0095] Step S32: For each time slot window, determine the corresponding process physical characteristics based on the motion command and processing parameters corresponding to that time slot window.

[0096] It should be noted that in step S32, "determining the process physical characteristics corresponding to the time slot window based on the motion command and machining parameters corresponding to the time slot window" means that the system parses the CNC code segment synchronized with the time interval of the time slot window, and calculates or maps the process physical characteristics that characterize the current machining state, such as the theoretical cutting force range, the main excitation frequency band, or the expected vibration energy level, by combining the machining parameters such as spindle speed, feed rate, and cutting depth with the motion command type (such as linear interpolation or circular interpolation) through physical models or empirical formulas. This provides a physical basis for subsequent frequency selection.

[0097] Step S33: Based on the second preset table, determine the target frequency range for each time slot window according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between process physical characteristic data and frequency range data.

[0098] It should be noted that in step S33, "based on the second preset table, determine the target frequency range of each time slot window according to the process physical characteristics," the key lies in using a pre-established mapping knowledge base. This second preset table is constructed based on cutting mechanics, rotor dynamics theory, and historical data mining. It clearly associates specific process physical characteristics such as "high-speed fine milling" and "large depth of cut rough boring" with one or more target frequency ranges that are most likely to contain core state information (e.g., spindle bearing characteristic frequency band, tool passage frequency harmonic band). Through table lookup matching, the system can dynamically and specifically lock the characteristic frequency intervals that should be analyzed for each time slot window.

[0099] Step S34: Based on the target frequency range, determine the effective components corresponding to each time slot window from the target intrinsic mode function components.

[0100] It should be noted that in step S34, "determining the effective components corresponding to each time slot window from the target intrinsic mode function components according to the target frequency range" is the core operation for achieving accurate feature selection. The system compares the dominant frequency or energy concentration band of each IMF component with the target frequency range determined in step S33. Only those IMFs whose main frequency components fall within or are significantly close to the target frequency range are determined to be effective components of that time slot window. This process is essentially a filtering based on physical mechanisms. It automatically filters out noise components, background vibrations, or interference signals from other components that are irrelevant to the current processing, ensuring that the subsequently extracted features are highly focused on the true state response of key components under the current process, thereby greatly improving the signal-to-noise ratio of state characterization and the specificity of fault diagnosis.

[0101] Step S35: Extract features from each effective component to obtain the corresponding candidate state features.

[0102] Step S36: Based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features.

[0103] It should be noted that in step S36, "based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features," the core of which lies in achieving an adaptive weighted fusion. This method does not simply average the features of all time slot windows, but rather dynamically assigns a confidence weight to the candidate state features of each window based on the Bayesian inference framework. Specifically, the system uses the statistical distribution of feature values ​​in historical normal data as prior knowledge, and the feature values ​​observed in each time slot window within the current processing cycle as new evidence. It calculates the posterior probability that each window feature belongs to the "normal state" under the current observation using Bayes' theorem. This posterior probability is used as the weight of the window feature. The final merged feature is a weighted linear combination of all candidate state features of all time slot windows, weighted by their posterior probabilities. This process makes the feature contributions from time slot windows with stable physical processes and high signal quality greater, while automatically weakening the feature contributions from windows affected by accidental interference or transient processes, thereby generating a more stable and reliable synthetic feature that characterizes the overall vibration mode of the component within this processing cycle.

[0104] Step S37: Generate a component vibration cooperative vector based on the merged features.

[0105] In the above implementation, in order to construct the component vibration cooperative vector, for each processing cycle, the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle are obtained. Then, for each time slot window, the process physical characteristics corresponding to that time slot window are determined according to the motion command and processing parameters corresponding to that time slot window. Then, based on a second preset table, the target frequency range of each time slot window is determined according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between process physical characteristic data and frequency range data. Then, based on the target frequency range, the effective components corresponding to each time slot window are determined from the target intrinsic mode function components. Then, feature extraction is performed on each effective component to obtain the corresponding candidate state features. Then, based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features. Finally, based on the merged features, the component vibration cooperative vector is generated.

[0106] As a further implementation of the method, based on Bayes' theorem, the step of linearly merging the candidate state features to obtain merged features includes:

[0107] Step S41: For a single time slot window, obtain the prior probability distribution model corresponding to each candidate state feature, and determine the likelihood probability of each candidate state feature in each time slot window based on the prior probability distribution model. The prior probability distribution model is obtained based on historical normal vibration signal samples.

[0108] Step S42: Based on the Bayesian probability formula, the likelihood probability of each candidate state feature is converted into a posterior probability, and the posterior probability is normalized to obtain the fusion weight corresponding to each candidate state feature.

[0109] Step S43: Based on the fusion weight, linearly merge the candidate state features within the current time slot window to obtain the merged features corresponding to the current time slot window.

[0110] In the above implementation, in order to obtain the merged features, for a single time slot window, the prior probability distribution model corresponding to each candidate state feature is obtained, and the likelihood probability of each candidate state feature in each time slot window is determined according to the prior probability distribution model. The prior probability distribution model is obtained based on historical normal vibration signal samples. Then, based on the Bayesian probability formula, the likelihood probability of each candidate state feature is converted into a posterior probability, and the posterior probability is normalized to obtain the fusion weight corresponding to each candidate state feature. Then, according to the fusion weight, the candidate state features in the current time slot window are linearly merged to obtain the merged features corresponding to the current time slot window.

[0111] As a further implementation of the method, the step of generating a component vibration cooperative vector based on the merged features includes:

[0112] Step S51: For each processing cycle, group the time slot windows within that processing cycle according to the corresponding processing technology type to obtain at least one technology type group.

[0113] It should be noted that the core purpose of "grouping the time slot windows within the processing cycle according to the corresponding processing technology type" in step S51 is to achieve the structured organization of features and process association. Based on the processing technology type (such as fine milling, rough boring) corresponding to each time slot window recorded in the signal segmentation stage (see step S23), the system groups time slot windows that perform the same or similar processes within the same cycle into the same process type group. This grouping strategy reorganizes discrete time-series features (merged features) into categories with clear process semantics, enabling subsequent analysis to distinguish and focus on the differentiated effects of different process conditions on the vibration mode of the component.

[0114] Step S52: For each process type group, perform statistical feature extraction on the merged features of each time slot window within the group to obtain the group statistical features.

[0115] It should be noted that in step S52, for each process type group, statistical features are extracted from the combined features of all time slot windows within the group. For example, the average value, standard deviation, kurtosis, skewness and other statistical quantities of the combined features within the group are calculated to form group statistical features that characterize the overall integrity, stability and distribution characteristics of the vibration mode of the component under this specific process type. This step condenses the detailed information of multiple time slots into more representative statistical indicators, effectively reducing the impact of data dimensionality and random fluctuations.

[0116] Step S53: Combine the group statistical features of each process type group according to a preset arrangement order to form the component vibration coordination vector corresponding to the processing cycle.

[0117] In the above implementation, in order to generate the component vibration coordination vector, for each processing cycle, each time slot window in the processing cycle is grouped according to the corresponding processing technology type to obtain at least one process type group. Then, for each process type group, the combined features of each time slot window in the group are statistically extracted to obtain group statistical features. Then, the group statistical features of each process type group are combined according to a preset arrangement order to form the component vibration coordination vector corresponding to the processing cycle.

[0118] As a further implementation of the method, the step of determining the component vibration coordination index based on the component vibration coordination matrix includes:

[0119] Step S61: Based on historical normal vibration signal samples, construct the corresponding reference vibration cooperative vector and combine them to form the reference vibration cooperative matrix.

[0120] Step S62: Perform principal component analysis on the reference vibration coordination matrix to obtain the principal component vector.

[0121] Step S63: Calculate the cosine similarity between each row vector in the component vibration coordination matrix and the principal component vector to obtain the similarity sequence.

[0122] Step S64: Calculate the information entropy of the similarity sequence and use the information entropy as a component vibration coordination index.

[0123] In the above implementation, in order to determine the component vibration coordination index, a corresponding benchmark vibration coordination vector is constructed based on historical normal vibration signal samples, and the vectors are combined to form a benchmark vibration coordination matrix. Then, principal component analysis is performed on the benchmark vibration coordination matrix to obtain the principal component vector. Then, the cosine similarity between each row vector in the component vibration coordination matrix and the principal component vector is calculated to obtain a similarity sequence. Then, the information entropy of the similarity sequence is calculated, and the information entropy is used as the component vibration coordination index.

[0124] As a further implementation of the method, the step of constructing the corresponding component state vector based on the intrinsic mode function components includes:

[0125] Step S71: For each transmission component, for each processing cycle, feature extraction is performed on the target intrinsic mode function components to obtain the state features of the transmission component corresponding to each time slot window in the processing cycle.

[0126] Step S72: For each transmission component, for different processing cycles, statistically aggregate the state characteristics of a single transmission component corresponding to all time slot windows in a single processing cycle to obtain the periodic state characteristics of the transmission component in that processing cycle.

[0127] Step S73: For each transmission component, arrange the periodic state characteristics of the transmission component in each processing cycle according to the time sequence of the processing cycle to generate a component state vector.

[0128] In the above implementation, in order to construct the corresponding component state vector, for each transmission component, for each processing cycle, feature extraction is performed on the target intrinsic mode function components to obtain the state features of the transmission component corresponding to each time slot window in the processing cycle. Then, for each transmission component, for different processing cycles, the state features corresponding to all time slot windows of a single transmission component in a single processing cycle are statistically aggregated to obtain the periodic state features of the transmission component in the processing cycle. Finally, for each transmission component, the periodic state features of the transmission component in each processing cycle are arranged according to the time order of the processing cycle to generate the component state vector.

[0129] As a further implementation of the method, the step of determining the component performance index based on the component state matrix includes:

[0130] Step S81: Based on historical normal vibration signal samples, construct the corresponding reference component state matrix.

[0131] Step S82: Perform principal component analysis on the benchmark component state matrix to obtain the benchmark principal component vector set and the variance contribution rate corresponding to each benchmark principal component vector.

[0132] Step S83: Perform principal component analysis on the component state matrix to obtain the target principal component vector set.

[0133] Step S84: Calculate the cosine of the angle between the principal component vectors with corresponding indices in the benchmark principal component vector set and the target principal component vector set, respectively, to obtain the similarity of m principal components, where m is a positive integer determined according to the preset cumulative variance contribution rate threshold.

[0134] In step S85, the principal component similarity is weighted and fused using the variance contribution rate as the weight to obtain the weighted similarity, and the component performance index is determined based on the weighted similarity.

[0135] In the above implementation, in order to determine the component performance indicators, a corresponding benchmark component state matrix is ​​constructed based on historical normal vibration signal samples. Then, principal component analysis is performed on the benchmark component state matrix to obtain the benchmark principal component vector set and the variance contribution rate corresponding to each benchmark principal component vector. Then, principal component analysis is performed on the component state matrix to obtain the target principal component vector set. Then, the cosine value of the angle between the principal component vectors with corresponding indices in the benchmark principal component vector set and the target principal component vector set is calculated to obtain m principal component similarities, where m is a positive integer determined according to a preset cumulative variance contribution rate threshold. Then, the principal component similarities are weighted and fused with variance contribution rate as the weight to obtain a weighted similarity, and the component performance indicators are determined based on the weighted similarity.

[0136] This application also discloses a machine tool monitoring system based on the Industrial Internet of Things.

[0137] refer to Figure 2 A machine tool monitoring system based on the Industrial Internet of Things (IIoT) includes a management platform, a sensor network platform, and an object platform that are connected in sequence via communication. The management platform is configured with:

[0138] The vibration signal segmentation module is used to acquire the vibration signals of each transmission component in each processing cycle during the current batch processing task of the target machine tool, and to segment the vibration signals to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier. Each element corresponding to each index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window.

[0139] The Empirical Mode Decomposition (EMD) module is used to perform EMD on each vibration signal segment to obtain the corresponding intrinsic mode function components.

[0140] The component vibration coordination module is used to construct a corresponding component vibration coordination vector based on the intrinsic mode function components for each processing cycle, and combine them to form a component vibration coordination matrix; and to determine the component vibration coordination index based on the component vibration coordination matrix.

[0141] The component state module is used to construct the corresponding component state vector for each transmission component based on the intrinsic mode function components, and combine them to form a component state matrix; and to determine the component performance index based on the component state matrix.

[0142] The fault determination module is used to determine whether a fault has occurred in the target machine tool based on the component vibration coordination index and component performance index.

[0143] The overall framework of another application scenario of the machine tool monitoring system based on the Industrial Internet of Things in this application is as follows: Figure 3 As shown, it can include a user platform, service platform, management platform, sensor network platform, and object platform that interact sequentially, forming a five-platform architecture based on the Industrial Internet of Things. The sensor network platform includes n sensor network sub-platforms, each with its own sensor sub-database. The service platform includes a main service database, n service sub-databases, and n service sub-platforms.

[0144] Specifically, in the aforementioned application scenario, the machine tool monitoring method based on the Industrial Internet of Things (IIoT) includes a management platform configured to: acquire vibration signals of each transmission component during each processing cycle of the target machine tool while executing the current batch of processing tasks; segment the vibration signals to obtain a set of vibration signal segments, wherein the set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier, and each element corresponding to the index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window; perform empirical mode decomposition on each vibration signal segment to obtain the corresponding intrinsic mode function (IMF) components; for each processing cycle, construct the corresponding component vibration coordination vector based on the IMF components and combine them to form a component vibration coordination matrix; determine the component vibration coordination index based on the component vibration coordination matrix; construct the corresponding component state vector for each transmission component based on the IMF components and combine them to form a component state matrix; determine the component performance index based on the component state matrix; and determine whether the target machine tool has malfunctioned based on the component vibration coordination index and the component performance index.

[0145] By establishing a complete closed-loop information operation logic through the interaction between various functional platforms of the industrial Internet of Things-based machine tool monitoring system based on the above three or five platforms, the orderly operation of sensing and control information is ensured, thereby realizing intelligent equipment management.

[0146] The machine tool monitoring system based on the Industrial Internet of Things of the present invention can implement any of the machine tool monitoring methods based on the Industrial Internet of Things, and the specific working process of the machine tool monitoring system based on the Industrial Internet of Things of the present invention can refer to the corresponding process in the above-mentioned machine tool monitoring methods based on the Industrial Internet of Things.

[0147] This application also discloses a computer device.

[0148] refer to Figure 4 A computer device includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement any of the above-described machine tool monitoring methods based on the Industrial Internet of Things.

[0149] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A machine tool monitoring method based on the Industrial Internet of Things, characterized in that, Applied to an industrial Internet of Things (IIoT) system, the IIoT system includes a management platform, a sensor network platform, and an object platform that are sequentially and communicatively connected. The method is executed by the management platform and includes: The vibration signals of each transmission component during each processing cycle are obtained during the current batch processing task of the target machine tool. The vibration signals are then segmented to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier. Each element corresponding to each index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window. Empirical mode decomposition is performed on each of the vibration signal segments to obtain the corresponding intrinsic mode function components; For each processing cycle, a corresponding component vibration coordination vector is constructed based on the intrinsic mode function components, and the vectors are combined to form a component vibration coordination matrix; and a component vibration coordination index is determined based on the component vibration coordination matrix. For each transmission component, a corresponding component state vector is constructed based on the intrinsic mode function components, and the vectors are combined to form a component state matrix; and the component performance index is determined based on the component state matrix. Determine whether the target machine tool has malfunctioned based on the component vibration coordination index and the component performance index; The step of segmenting the vibration signal to obtain a set of vibration signal segments includes: Analyze the CNC program that drives the target machine tool to execute the current batch machining task; Based on the motion instructions and machining parameters in the CNC program, the machining process type for each machining stage is determined. Based on the first preset table, according to the processing technology type, time slot lengths are allocated to each processing stage to obtain a variable-length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between the processing technology type data and the time slot length sequence data. Based on the variable-length time slot timing sequence, the vibration signal corresponding to each transmission component in each processing cycle is segmented to obtain a set of vibration signal segments; The step of constructing a corresponding component vibration cooperative vector based on the intrinsic mode function components for each processing cycle includes: For each processing cycle, the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle are obtained; For each time slot window, the corresponding process physical characteristics are determined based on the motion command and processing parameters corresponding to that time slot window. Based on the second preset table, the target frequency range of each time slot window is determined according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between the process physical characteristic data and the frequency range data. Based on the target frequency range, determine the effective components corresponding to each time slot window from the target intrinsic mode function components; Feature extraction is performed on each of the effective components to obtain the corresponding candidate state features; Based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features; Based on the merged features, a component vibration coordination vector is generated.

2. The machine tool monitoring method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of linearly merging the candidate state features based on Bayes' theorem to obtain merged features includes: For a single time slot window, a prior probability distribution model corresponding to each of the candidate state features is obtained, and the likelihood probability of each of the candidate state features in each time slot window is determined based on the prior probability distribution model, wherein the prior probability distribution model is obtained based on historical normal vibration signal samples. Based on the Bayesian probability formula, the likelihood probability of each candidate state feature is converted into a posterior probability, and the posterior probability is normalized to obtain the fusion weight corresponding to each candidate state feature. Based on the fusion weight, the candidate state features within the current time slot window are linearly merged to obtain the merged features corresponding to the current time slot window.

3. The machine tool monitoring method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of generating a component vibration cooperative vector based on the merged features includes: For each processing cycle, each time slot window within that processing cycle is grouped according to the corresponding processing technology type to obtain at least one technology type group; For each process type group, statistical features are extracted from the merging features of each time slot window within the group to obtain group statistical features; The statistical features of each process type group are combined in a preset order to form the component vibration coordination vector corresponding to the processing cycle.

4. The machine tool monitoring method based on the Industrial Internet of Things according to claim 2, characterized in that, The step of determining the component vibration coordination index based on the component vibration coordination matrix includes: Based on the historical normal vibration signal samples, a corresponding reference vibration cooperative vector is constructed and combined to form a reference vibration cooperative matrix; Principal component analysis was performed on the reference vibration coordination matrix to obtain the principal component vector; Calculate the cosine similarity between each row vector in the component vibration coordination matrix and the principal component vector to obtain a similarity sequence; The information entropy of the similarity sequence is calculated, and the information entropy is used as a component vibration coordination index.

5. The machine tool monitoring method based on the Industrial Internet of Things according to claim 1, characterized in that, The step of constructing the corresponding component state vector based on the intrinsic mode function components includes: For each transmission component, for each processing cycle, feature extraction is performed on the target intrinsic mode function components to obtain the state features of the transmission component corresponding to each time slot window in that processing cycle. For each transmission component, for different processing cycles, the state characteristics of a single transmission component corresponding to all time slot windows in a single processing cycle are statistically aggregated to obtain the periodic state characteristics of the transmission component in that processing cycle. For each transmission component, the periodic state characteristics of the transmission component in each processing cycle are arranged according to the time sequence of the processing cycle to generate a component state vector.

6. The machine tool monitoring method based on the Industrial Internet of Things according to claim 4, characterized in that, The step of determining the component performance index based on the component state matrix includes: Based on the historical normal vibration signal samples, a corresponding reference component state matrix is ​​constructed; Principal component analysis is performed on the state matrix of the reference component to obtain the set of reference principal component vectors and the variance contribution rate corresponding to each reference principal component vector; Principal component analysis is performed on the component state matrix to obtain the target principal component vector set; Calculate the cosine of the angle between the principal component vectors with corresponding indices in the benchmark principal component vector set and the target principal component vector set respectively to obtain the similarity of m principal components, where m is a positive integer determined according to a preset cumulative variance contribution rate threshold; The principal component similarities are weighted and fused using the variance contribution rate as the weight to obtain a weighted similarity, and the component performance indicators are determined based on the weighted similarity.

7. A machine tool monitoring system based on the Industrial Internet of Things, characterized in that, It includes a management platform, a sensor network platform, and an object platform that are connected in sequence. The management platform is configured with: The vibration signal segmentation module is used to acquire the vibration signals of each transmission component in each processing cycle during the current batch processing task of the target machine tool, and to segment the vibration signals to obtain a set of vibration signal segments. The set of vibration signal segments is indexed by the transmission component identifier and the processing cycle identifier. Each element corresponding to each index contains multiple vibration signal segments, and each vibration signal segment corresponds to a time slot window. The empirical mode decomposition module is used to perform empirical mode decomposition on each of the vibration signal segments to obtain the corresponding intrinsic mode function components; The component vibration coordination module is used to construct a corresponding component vibration coordination vector based on the intrinsic mode function components for each processing cycle, and combine them to form a component vibration coordination matrix; and to determine the component vibration coordination index based on the component vibration coordination matrix. The component state module is used to construct a corresponding component state vector for each transmission component based on the intrinsic mode function components, and combine them to form a component state matrix; and to determine the component performance index based on the component state matrix. The fault determination module is used to determine whether the target machine tool has malfunctioned based on the component vibration coordination index and the component performance index. The step of segmenting the vibration signal to obtain a set of vibration signal segments includes: Analyze the CNC program that drives the target machine tool to execute the current batch machining task; Based on the motion instructions and machining parameters in the CNC program, the machining process type for each machining stage is determined. Based on the first preset table, according to the processing technology type, time slot lengths are allocated to each processing stage to obtain a variable-length time slot sequence. The first preset table includes processing technology type data and time slot length sequence data. The first preset table is used to represent the mapping relationship between the processing technology type data and the time slot length sequence data. Based on the variable-length time slot timing sequence, the vibration signal corresponding to each transmission component in each processing cycle is segmented to obtain a set of vibration signal segments; The step of constructing a corresponding component vibration cooperative vector based on the intrinsic mode function components for each processing cycle includes: For each processing cycle, the target intrinsic mode function components corresponding to the vibration signal segments of a single transmission component in each time slot window within that processing cycle are obtained; For each time slot window, the corresponding process physical characteristics are determined based on the motion command and processing parameters corresponding to that time slot window. Based on the second preset table, the target frequency range of each time slot window is determined according to the process physical characteristics. The second preset table includes process physical characteristic data and frequency range data, and is used to represent the mapping relationship between the process physical characteristic data and the frequency range data. Based on the target frequency range, determine the effective components corresponding to each time slot window from the target intrinsic mode function components; Feature extraction is performed on each of the effective components to obtain the corresponding candidate state features; Based on Bayes' theorem, the candidate state features of each time slot window are linearly merged to obtain merged features; Based on the merged features, a component vibration coordination vector is generated.

8. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method of any one of claims 1 to 6.

Images