Method and system for intelligent analysis of high-speed motor operation data

Abstract

The application provides a high-speed motor operation data intelligent analysis method and system, and relates to the technical field of data processing.The method comprises the following steps: performing a preprocessing operation on basic operation information and surrounding working condition factors to form initial integrated information; performing boundary contour calculation on a multidimensional data point set for the initial integrated information, and extracting an outer convex boundary point set of data distribution; determining a coverage range of data distribution based on the outer convex boundary point set, and establishing a minimum convex polygon verification interval containing all data points; performing adaptive hierarchical cutting on the verification interval according to a data density gradient and a working condition change amplitude to form a plurality of cutting sub-regions with different characteristic densities; extracting geometric features, distribution densities and boundary curvatures of each cutting sub-region as characteristic parameters, and mapping and converting the characteristic parameters into stability evaluation parameters and abnormality evaluation parameters of corresponding verification points.The application realizes accurate evaluation, real-time abnormality identification and dynamic adaptation of the running state of a high-speed motor.

Description

Technical Field

[0001] This invention relates to the fields of industrial automation and motor control technology, and in particular to a method and system for intelligent analysis of high-speed motor operating data. Background Technology

[0002] In the fields of high-end equipment such as new energy vehicles and high-speed precision machine tools, high-speed motors have become key power components due to their core advantages of high power density and high operating efficiency. Real-time monitoring and early warning of abnormalities in their operating status are directly related to the safety and service life of the equipment. However, high-speed motors operate at ultra-high speeds and face complex scenarios with frequent load fluctuations and dynamic changes in environmental temperature and humidity. The multi-dimensional operating data has strong coupling and nonlinear characteristics, and traditional analysis methods are difficult to accurately depict the correlation patterns of the data.

[0003] To ensure the reliability of drive motor operation, an automobile manufacturer adopted a monitoring method based on fixed thresholds. By collecting relevant operating parameters of the motor and environmental factor data, and after simple normalization, the data is directly compared with preset thresholds to detect anomalies. However, under special environmental conditions and continuous high-speed driving, the sudden change in environmental factors and the sudden increase in motor load combined to fail to identify the coordinated abnormal changes between key operating parameters. This led to a failure of accelerated component wear after the motor was slightly overloaded for a long time. This exposed the technical defects of traditional methods, such as the lack of accurate extraction of the boundary features of multi-dimensional data space, the use of fixed cutting methods and static weight configuration, which cannot adapt to dynamic changes in operating conditions, resulting in lagging anomaly identification and high false alarm and false negative rates. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for intelligent analysis of high-speed motor operation data, so as to achieve accurate assessment of the operating status of high-speed motor, real-time anomaly identification and dynamic adaptation.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a method for intelligent analysis of high-speed motor operating data, the method comprising:

[0007] Collect basic operating information of the motor and surrounding operating conditions;

[0008] Preprocessing operations are performed on basic operational information and surrounding working conditions to form initial integrated information;

[0009] Based on the initial integrated information, the boundary contour of the multidimensional data point set is calculated to extract the convex boundary point set of the data distribution. Based on the convex boundary point set, the coverage of the data distribution is determined, and a minimum convex polygon verification interval containing all data points is established. The verification interval is adaptively layered and cut according to the data density gradient and the magnitude of the change in working conditions to form multiple cut sub-regions with different feature densities. The geometric features, distribution density, and boundary curvature of each cut sub-region are extracted as feature parameters, and the feature parameters are mapped and converted into stability evaluation parameters and anomaly evaluation parameters for the corresponding verification points.

[0010] Path optimization analysis is performed on the stability evaluation parameters and anomaly evaluation parameters of each verification point to determine the final verification path from the starting sub-region of the verification interval to the target sub-region, and the global evaluation output corresponding to the final verification path is converted into correction coefficients.

[0011] The weight configuration scheme in the dynamic weight integration mechanism is calibrated based on the correction coefficient to generate improved initial integration information;

[0012] The improved initial integrated information is collaboratively analyzed, the differences are compared and deeply integrated with the surrounding operating conditions, and the correction completion information is output.

[0013] Furthermore, preprocessing operations are performed on the basic operational information and surrounding operating conditions to form initial integrated information, including:

[0014] The system collects basic operating information of the motor and surrounding operating conditions from the data source, and performs format standardization processing on the collected data to ensure data format consistency, resulting in data with a unified format.

[0015] Perform missing value imputation and outlier removal operations on data after it has been standardized to improve data integrity;

[0016] The processed data is dimensionally aligned and time-series synchronized to construct a multidimensional data point set;

[0017] The multidimensional data point set is normalized to eliminate the influence of dimensions and form initial integrated information.

[0018] Furthermore, based on the initial integrated information, boundary contour calculations are performed on the multidimensional data point set to extract the convex boundary point set of the data distribution; based on the convex boundary point set, the coverage area of ​​the data distribution is determined, and a minimum convex polygon verification interval containing all data points is established, including:

[0019] Receive initial integrated information, perform dimensional structure analysis and distribution feature identification on the multidimensional data point set after normalization of the initial integrated information, and generate analytical results that characterize the spatial form of the data.

[0020] Based on the analytical results, the incremental convex hull algorithm is used to perform boundary contour calculation on the multidimensional data point set. The convex boundary point set that completely represents the extensional shape of the data distribution is identified and extracted from the analytical results through point-by-point iteration.

[0021] By using the set of convex boundary points, the spatial distance and angular relationship between each boundary point are calculated. Based on the spatial geometric characteristics of the set of convex boundary points, the overall coverage of the data distribution is determined, and a spatial boundary description framework that accurately describes the characteristics of the data boundary is constructed.

[0022] Based on the spatial boundary description framework, a geometric optimization strategy is adopted to connect key vertices and optimize the shape of the set of convex boundary points in the framework, generating the minimum convex polygon that strictly contains all data points, and defining the minimum convex polygon as the verification interval.

[0023] Furthermore, the verification interval is adaptively segmented according to the data density gradient and the magnitude of operating condition changes, forming multiple segmented sub-regions with different feature densities. The geometric features, distribution density, and boundary curvature of each segmented sub-region are extracted as feature parameters, and these feature parameters are mapped and converted into stability evaluation parameters and anomaly evaluation parameters for the corresponding verification points, including:

[0024] The system receives the verification interval, performs density distribution scanning and operating condition change analysis on the multidimensional data points within the verification interval, calculates the data density gradient distribution map and the operating condition change amplitude matrix, and obtains an adaptive parameter set to guide the stratification and segmentation.

[0025] Based on the adaptive parameter set, an adaptive hierarchical cutting operation is performed on the verification interval. According to the local characteristics of the data density gradient change trend and the working condition change magnitude, the cutting threshold and cutting direction are dynamically adjusted to divide the verification interval into multiple cutting sub-regions with consistent internal data distribution characteristics and different feature density levels.

[0026] For each of the formed sub-regions, the geometric morphological features, data distribution density features, and boundary morphological features of each sub-region are calculated to obtain a multi-dimensional set of feature parameters.

[0027] The set of feature parameters is transformed into stability evaluation parameters for the corresponding verification points under the current working conditions through feature weight allocation and nonlinear mapping relationship.

[0028] Based on stability evaluation parameters, and combined with historical anomaly pattern database and real-time operating condition comparison analysis, the feature parameters are further mapped and converted into anomaly evaluation parameters for corresponding verification points.

[0029] Furthermore, path optimization analysis is performed on the stability evaluation parameters and anomaly evaluation parameters of each verification point to determine the final verification path from the starting sub-region to the target sub-region of the verification interval. The global evaluation output corresponding to the final verification path is then converted into correction coefficients, including:

[0030] The system receives stability evaluation parameters and anomaly evaluation parameters from each verification point, performs spatial mapping and weight allocation on the evaluation parameters, and constructs the state transition cost matrix between each sub-region within the verification interval.

[0031] Based on the constructed state transition cost matrix, starting from the initial sub-region of the verification interval, and taking into account the continuity constraints of the stability evaluation parameters and the jump threshold of the anomaly evaluation parameters, the verification path that minimizes the global evaluation cost is searched to obtain a set of candidate verification paths.

[0032] The candidate verification path set is evaluated from multiple dimensions. The global stability score and anomaly accumulation degree of each candidate path are calculated. The optimal verification path is selected by combining path length, computational complexity and real-time requirements, and the final verification path from the starting sub-region to the target sub-region is determined.

[0033] Based on the final verification path, the evaluation parameters of all verification points on the path are fused in time and aggregated in space to calculate the overall stability index and anomaly risk index of the path, and form a global evaluation output.

[0034] The global evaluation output is input into a preset conversion function. Based on the degree of deviation between the global evaluation output and the preset benchmark value, the conversion parameters are dynamically adjusted to linearly map the global evaluation output to a correction coefficient between 0 and 1.

[0035] Furthermore, the weight configuration scheme in the dynamic weight integration mechanism is calibrated based on the correction coefficient to generate improved initial integration information, including:

[0036] Receive correction coefficients, perform numerical analysis and calibration direction identification on the correction coefficients, and determine the magnitude and direction parameters of the weight adjustment;

[0037] Based on the amplitude and direction parameters, an adaptive calibration operation is performed on the original weight configuration scheme in the dynamic weight integration mechanism. According to the magnitude and sign characteristics of the correction coefficient, the weight allocation ratio of each data dimension is dynamically adjusted to obtain the calibrated weight configuration scheme.

[0038] By using a calibrated weight configuration scheme, the initial integrated information is reconstructed with weights, and the multidimensional data point set is re-integrated and calculated according to the calibrated weight ratios to eliminate the influence of biases in the original data and obtain improved initial integrated information with optimized weight distribution.

[0039] Furthermore, the improved initial integrated information is collaboratively analyzed, compared, and deeply integrated with surrounding operating conditions to output correction completion information, including:

[0040] Receive the improved initial integration information and simultaneously acquire the real-time surrounding operating conditions corresponding to the initial integration information to construct a comprehensive dataset containing motor operating status and environmental conditions;

[0041] Based on the comprehensive dataset, a spatiotemporal correlation analysis was performed on the motor operating characteristics and surrounding operating conditions in the improved initial integrated information to identify the coupling relationship between operating status and changes in operating conditions, and to obtain collaborative analysis results.

[0042] Perform a difference comparison operation on the collaborative analysis results, calculate the deviation between the current operating status and the historical benchmark status under the same working conditions, quantify the degree of abnormal fluctuation of motor operating parameters, and generate a difference comparison report.

[0043] Based on the difference comparison report, the improved initial integrated information and surrounding operating conditions are deeply integrated. The operating condition interference factors are eliminated by multi-source data fusion algorithm, the essential characteristics of motor operation are extracted, and the corrected operating status evaluation data is obtained.

[0044] The corrected operating status assessment data is standardized and packaged, and timestamps, operating condition identifiers, and confidence level labels are added. The output includes correction completion information containing comprehensive assessment results of motor operating status and correction suggestions.

[0045] Secondly, the high-speed motor operation data intelligent analysis system includes:

[0046] The acquisition module is used to collect basic operating information of the motor and surrounding operating conditions.

[0047] The integration module is used to perform preprocessing operations on basic operational information and surrounding working conditions to form initial integrated information.

[0048] The calculation module is used to calculate the boundary contour of the multidimensional data point set based on the initial integrated information, and extract the convex boundary point set of the data distribution; determine the coverage of the data distribution based on the convex boundary point set, and establish the minimum convex polygon verification interval containing all data points; adaptively divide the verification interval into multiple sub-regions with different feature densities according to the data density gradient and the magnitude of the change in working conditions; extract the geometric features, distribution density and boundary curvature of each sub-region as feature parameters, and map the feature parameters into the stability evaluation parameters and anomaly evaluation parameters of the corresponding verification points;

[0049] The optimization module is used to perform path optimization analysis on the stability evaluation parameters and anomaly evaluation parameters of each verification point, determine the final verification path from the starting sub-region of the verification interval to the target sub-region, and convert the global evaluation output corresponding to the final verification path into correction coefficients.

[0050] The calibration module is used to calibrate the weight configuration scheme in the dynamic weight integration mechanism based on the correction coefficient, and generate improved initial integration information.

[0051] The processing module is used to perform collaborative analysis, difference comparison and deep integration of the improved initial integrated information with surrounding operating conditions and factors, and output correction completion information.

[0052] Thirdly, a computing device includes:

[0053] One or more processors;

[0054] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0055] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0056] The above-described solution of the present invention has at least the following beneficial effects:

[0057] By employing refined preprocessing techniques such as multi-dimensional data acquisition and format unification, missing value filling, outlier removal, dimension alignment and time-series synchronization, and normalization; spatial feature extraction techniques using incremental convex hull algorithm to extract the set of external convex boundary points and construct the minimum convex polygon verification interval; adaptive layering and cutting techniques based on data density gradient and operating condition change amplitude; evaluation parameter conversion techniques for feature weight allocation and nonlinear mapping; path optimization techniques for constructing state transition cost matrix and path search; dynamic weight calibration techniques based on correction coefficients; and spatiotemporal correlation analysis, difference comparison, and deep fusion techniques for multi-source data, this approach overcomes the technical problems of traditional intelligent analysis methods for high-speed motor operation data, such as lack of accurate characterization of multi-dimensional data spatial boundary features, inability to adapt to dynamic changes in operating conditions due to fixed cutting methods and static weight configurations, resulting in delayed anomaly identification and high false alarm / missed alarm rates. This achieves the technical effects of dynamically adapting to complex operating scenarios of high-speed motors, accurately extracting the essential characteristics of motor operation, identifying operating anomalies in real time and accurately, improving the reliability and timeliness of motor operation monitoring, and providing strong technical support for the safe and stable operation of high-end equipment. Attached Figure Description

[0058] Figure 1This is a flowchart illustrating the intelligent analysis method for high-speed motor operating data provided in an embodiment of the present invention.

[0059] Figure 2 This is a schematic diagram of a high-speed motor operation data intelligent analysis system provided in an embodiment of the present invention. Detailed Implementation

[0060] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0061] like Figure 1 As shown, embodiments of the present invention propose an intelligent analysis method for high-speed motor operating data, the method comprising the following steps:

[0062] Step 1: Collect basic operating information of the motor and surrounding operating conditions;

[0063] Step 2: Perform preprocessing operations on the basic operational information and surrounding working conditions to form initial integrated information;

[0064] Step 3: Based on the initial integrated information, perform boundary contour calculation on the multidimensional data point set to extract the convex boundary point set of the data distribution; determine the coverage of the data distribution based on the convex boundary point set, and establish the minimum convex polygon verification interval containing all data points; adaptively divide the verification interval into multiple sub-regions with different feature densities according to the data density gradient and the magnitude of operating condition changes; extract the geometric features, distribution density, and boundary curvature of each sub-region as feature parameters, and map the feature parameters into the stability evaluation parameters and anomaly evaluation parameters of the corresponding verification points;

[0065] Step 4: Perform path optimization analysis on the stability evaluation parameters and anomaly evaluation parameters of each verification point to determine the final verification path from the starting sub-region of the verification interval to the target sub-region, and convert the global evaluation output corresponding to the final verification path into correction coefficients.

[0066] Step 5: Based on the correction coefficient, calibrate the weight configuration scheme in the dynamic weight integration mechanism to generate improved initial integration information;

[0067] Step 6: Perform collaborative analysis, difference comparison and in-depth integration of the improved initial integrated information with surrounding operating conditions and factors, and output correction completion information.

[0068] In this embodiment of the invention, a series of technical means are employed, including multi-dimensional motor basic operation information and surrounding operating condition factors collaborative acquisition, data preprocessing, multi-dimensional data point set boundary contour calculation and external convex boundary point set extraction, minimum convex polygon verification interval construction, adaptive layered cutting based on data density gradient and operating condition change amplitude, cutting sub-region feature parameter extraction and stability and anomaly evaluation parameter mapping transformation, evaluation parameter path optimization analysis and correction coefficient generation, dynamic weight configuration scheme calibration, improved initial integrated information and surrounding operating condition factors collaborative analysis and difference comparison and deep integration. Therefore, the technical problems of inaccurate characterization of multi-dimensional data space boundary features, difficulty in adapting to dynamic changes in operating conditions due to fixed cutting methods and static weight configuration, and resulting in delayed anomaly identification and high false alarm and false negative rates are overcome in traditional high-speed motor operation data analysis methods. This achieves the technical effects of dynamically adapting to complex high-speed motor operation scenarios, accurately capturing the essential characteristics of motor operation status, real-time and efficient identification of operation anomalies, improving the accuracy and reliability of motor operation monitoring, and providing strong technical support for the safe and stable operation of high-end equipment.

[0069] In a preferred embodiment of the present invention, step 1 above may include:

[0070] Step 1.1: Acquire basic operational information of the high-speed motor in real time through a distributed sensor network to generate a basic data stream with quality labels. Specifically, this includes deploying a distributed sensor network composed of multiple sensors of different types in key operating parts such as the high-speed motor's shaft windings, bearings, and cooling system, as well as the surrounding environment. Each sensor monitors various basic operational information of the motor in real time according to a preset acquisition frequency, including data such as motor speed, current, voltage, winding temperature, bearing temperature, output power, operating time, number of start-stop cycles, and power supply stability. When acquiring each piece of data, the sensor network simultaneously records the working status of the corresponding sensor, acquisition time, transmission delay, and data integrity. Based on the recorded information, each piece of acquired data is evaluated for quality, and different quality levels are assigned as quality labels. The basic operational information with quality labels is integrated in chronological order to form a continuous basic data stream.

[0071] Step 1.2: Based on the quality identifier of the basic data stream, dynamically prioritize the data sources, prioritizing the collection of high-quality data from high-reliability sensors. Simultaneously, low-quality data sources are marked and compensated, forming a set of basic operational information filtered by reliability. Simultaneously, surrounding operating conditions corresponding to this set of basic operational information are acquired. Specifically, this includes: receiving the basic data stream with quality identifiers, extracting the quality identifier information corresponding to each data point, and dynamically prioritizing all data sources based on the quality level represented by the quality identifier and the historical operational reliability data of the corresponding sensor. Data sources corresponding to sensors with high quality levels and strong historical operational reliability are set as high priority. Level; During data acquisition, high-quality data from high-priority data sources is prioritized to ensure the reliability of core data. For data sources with low-quality indicators, information such as sensor number, installation location, and data anomaly type is automatically recorded and marked. Simultaneously, combined with concurrent data from high-priority sensors of the same type and the historical normal operation data trend of the low-quality data source, reasonable supplementation methods are used to compensate for the low-quality data, fill data gaps, or correct abnormal data. After screening and compensation processing, a basic operation information set is formed. During this process, surrounding operating conditions within the same time period as the basic operation information set are synchronously collected through associated timestamps.

[0072] Step 1.3: Assess the environmental interference of surrounding operating conditions. Based on the interference assessment results, pre-screen and classify the operating conditions according to their importance, generating a set of effective operating conditions that are strongly correlated with the basic operating information. Specifically, this includes: collecting synchronously acquired surrounding operating conditions, and combining the operating characteristics of the high-speed motor and historical fault data to analyze the degree of interference of each operating condition factor on the motor's basic operating information. For example, determining whether factors such as sudden changes in ambient temperature and abnormal fluctuations in humidity will cause false changes in motor operating data. Based on this, conduct an environmental interference assessment for each surrounding operating condition factor, and determine the interference level based on the assessment results. As a result, operating condition factors with extremely high interference and no substantial reference value for judging the motor's operating status were eliminated, completing the pre-screening operation. Subsequently, the importance of each operating condition factor was classified based on the degree of correlation between it and the motor's basic operating information. For example, operating condition factors that directly affect the motor's operating load and heat dissipation effect, such as load fluctuations and the working status of the cooling system, were classified as high importance factors, while operating condition factors that have a relatively small impact on motor operation, such as ambient air pressure, were classified as low importance factors. Finally, the high-importance and medium-importance operating condition factors after pre-screening were integrated to form an effective set of operating condition factors that are strongly correlated with the basic operating information.

[0073] In this embodiment of the invention, by employing distributed sensor networks to collect basic data streams with quality labels in real time, dynamically prioritizing data sources based on quality labels and marking and compensating low-quality data sources, and assessing the environmental interference of surrounding operating conditions and performing pre-screening and importance grading based on the results, the technical problems of insufficient reliability of collected data due to lack of data quality screening in traditional motor data acquisition processes, chaotic data source priorities affecting data acquisition efficiency, and excessive environmental interference information in operating conditions with weak correlation to basic operational information are overcome. This achieves the technical effects of improving the credibility and relevance of collected data, ensuring strong correlation between basic operational information and operating conditions, and guaranteeing the accuracy and efficiency of the overall analysis process.

[0074] In a preferred embodiment of the present invention, step 2 above may include:

[0075] Step 2.1 involves collecting basic operating information of the motor and surrounding operating conditions from data sources, and standardizing the format of the collected data to ensure consistency. Specifically, this includes comprehensively collecting basic operating information and surrounding operating conditions from data sources corresponding to the basic operating information set (after reliability screening) and the effective operating condition factor set. Due to differences in the types of data acquisition devices from different data sources, the collected data may differ in format. Some data is stored directly in numerical form, some data includes additional identification information, and some data has inconsistent field naming and arrangement. These differences can affect the efficiency and accuracy of subsequent data processing. To address this issue, a unified data format standard needs to be established, clearly defining the field names, data types, numerical precision, storage format, and timestamp recording specifications. Following this standard, each piece of basic operating information and surrounding operating condition factor data is processed individually, converting data of different formats to the unified format, modifying non-standard field names, adjusting the data arrangement order, and standardizing the number of digits retained for numerical precision, ensuring that all collected data is completely consistent in format.

[0076] Step 2.2 involves performing missing value imputation and outlier removal on the standardized data to improve data integrity. Specifically, after standardizing the data format, a comprehensive integrity check is performed. This involves traversing all data records and fields to identify data entries and specific fields with missing values. Missing values ​​may arise from temporary sensor malfunctions, data transmission interruptions, etc. Appropriate imputation methods are adopted based on the characteristics and variation patterns of different data types. For operational data with temporal continuity, imputation is performed by referring to the normal data variation trends in adjacent time periods. For data strongly correlated with operating conditions, imputation is performed by combining historical data under similar operating conditions during the same period to ensure that the imputed data conforms to the actual operating patterns of the motor. Simultaneously, outlier detection is performed on the standardized data. Based on the rated operating parameter range of the motor, the fluctuation range of historical normal operating data, and relevant physical principles, it is determined whether the data exceeds a reasonable range. For outlier data that significantly deviates from the normal value range, exhibits drastic changes with adjacent data, or does not conform to the motor's operating logic, a detailed investigation is conducted to determine the cause. Once confirmed as invalid outlier data, it is removed to avoid interference from outlier data with subsequent analysis results, thereby improving data integrity and reliability.

[0077] Step 2.3 involves dimensional alignment and temporal synchronization of the processed data to construct a multidimensional data point set. Specifically, after missing value imputation and outlier removal, inconsistencies in dimensions and temporal synchronization still exist. Some data records may lack key dimensional information, and different data sources have varying collection frequencies, resulting in inconsistent timestamps. This affects the accuracy of multidimensional data correlation analysis. Therefore, dimensional alignment is first performed to identify the core data dimensions required for motor operation analysis, including basic operational information and all key indicators of surrounding operating conditions. Dimensional verification is then conducted for each data record, and missing key dimensional information is supplemented. Redundant and irrelevant dimensions are removed to ensure that all data records contain complete core analytical dimensions. Then, time-series synchronization is performed. A unified timeline is established using a preset time interval as a standard. Timestamp information for each data record is extracted, and data with different collection frequencies are mapped onto this timeline. For data with a collection frequency higher than the standard time interval, records at key time points are selected and retained. For data with a collection frequency lower than the standard time interval, reasonable interpolation is performed based on the data's changing trends to ensure that all data remain synchronized in the time dimension. Each time point corresponds to complete multi-dimensional data, thus constructing a structurally complete and time-consistent multi-dimensional data point set.

[0078] Step 2.4 involves normalizing the multidimensional data point set to eliminate the influence of dimensions and form initial integrated information. Specifically, this includes: In the constructed multidimensional data point set, different dimensions represent different physical meanings and have different dimensions. For example, the units for rotational speed data, temperature data, and current data are all different. Data with different dimensions differ significantly in numerical value. Direct analysis can lead to larger dimensions excessively influencing the analysis results, failing to objectively reflect the actual role of each dimension. To eliminate the adverse effects of dimensions, the multidimensional data point set needs to be normalized. First, the value range of each data dimension is determined, and the maximum and minimum values ​​of all data within that dimension are calculated. Then, according to a unified processing rule, all data within each dimension are transformed to the same numerical range, making data with different dimensions and numerical ranges comparable. Through this processing, all dimensions are on the same order of magnitude, objectively reflecting the changing patterns and relative importance of each dimension, eliminating the interference of dimensional differences on subsequent analysis, and ultimately forming initial integrated information that can be directly used for subsequent spatial feature extraction and intelligent analysis.

[0079] In this embodiment of the invention, a series of preprocessing techniques are employed to standardize the format of the collected data, fill in missing values ​​and remove outliers, align dimensions and synchronize time series, and normalize multidimensional data point sets. This overcomes the technical problems of format chaos, insufficient data integrity, asynchronous dimensions and time series, and interference from dimensional differences in analysis that exist in traditional data preprocessing. As a result, the invention achieves the goals of standardizing data format, improving data integrity and consistency, eliminating the influence of dimensional differences, and constructing high-quality multidimensional data point sets.

[0080] In a preferred embodiment of the present invention, step 3 above may include:

[0081] Step 3.1: Receive initial integrated information. Perform dimensional structure analysis and distribution feature identification on the normalized multidimensional data point set of the initial integrated information to generate an analytical result representing the spatial form of the data. Specifically, this includes: receiving the normalized initial integrated information, which contains a multidimensional data point set with a unified structure and consistent time sequence. This data point set covers the basic operating information of the motor and related dimensions of surrounding operating conditions. To accurately grasp the spatial distribution characteristics of the data, firstly, perform dimensional structure analysis on the multidimensional data point set, comprehensively sorting out the number of dimensions contained in the data point set, the physical meaning of each dimension, and the intrinsic relationship between dimensions, clarifying the role of different dimensions in representing the motor's operating state; on this basis, carry out distribution feature identification work, by analyzing the value range, central tendency, dispersion, and aggregation state of data points in the multidimensional space of each dimension, determining whether the data exhibits clustering phenomena and whether it presents a specific distribution pattern, and simultaneously identifying abnormal distribution areas and key distribution feature points in the data; integrate the dimensional correlation information obtained from the dimensional structure analysis with the distribution state information obtained from the distribution feature identification to generate an analytical result that can comprehensively represent the spatial form of the data.

[0082] Step 3.2: Based on the analytical results, the incremental convex hull algorithm is used to perform boundary contour calculation on the multidimensional data point set. The algorithm identifies and extracts the complete set of convex boundary points representing the extensional shape of the data distribution from the analytical results through a point-by-point iterative approach. Specifically, this includes: First, an empty convex hull structure is initialized as the basis for iteratively constructing the convex hull; then, data points are selected sequentially from the multidimensional data point set according to a preset order, and each selected data point is added to the current convex hull structure. By analyzing the spatial positional relationship between the data point and the existing convex hull vertices, it is determined whether the data point is located outside the current convex hull; if the data point is outside the convex hull, the existing convex hull structure is adjusted, deleting the original convex hull vertices occluded by the data point and adding the data point as a new convex hull vertex; if the data point is inside the convex hull or on its boundary, the existing convex hull structure remains unchanged. By using this point-by-point iterative approach, the convex hull structure is continuously updated and optimized until all data points in the multidimensional data point set have been traversed. Finally, the set of convex boundary points that can completely represent the extensional shape of the data distribution is identified and extracted from the analysis results.

[0083] Step 3.3: Using the convex boundary point set, calculate the spatial distances and angular relationships between each boundary point. Based on the spatial geometric characteristics of the convex boundary point set, determine the overall coverage of the data distribution and construct a spatial boundary description framework that accurately describes the characteristics of the data boundaries. Specifically, this includes: after obtaining the convex boundary point set, calculating the spatial distances between each boundary point; by traversing all vertices in the convex boundary point set, sequentially calculating the straight-line distances between adjacent boundary points in multidimensional space, and simultaneously calculating the spatial distances between non-adjacent boundary points, to comprehensively understand the distance distribution between boundary points; based on this, calculate the angular relationships between each boundary point, taking each edge... Using the boundary point as the vertex, the angle between the vertex and the line connecting the two adjacent boundary points is calculated to determine the direction and curvature of the line connecting the boundary points. Combining the spatial geometric characteristics of the convex boundary point set, namely the coordinate position of the boundary points, the distance relationship between them, and the angular relationship, the maximum and minimum values ​​of the data distribution in each dimension are determined, thereby delineating the overall coverage of the data distribution and clarifying the area occupied by the data in multidimensional space. By integrating the coordinate information of the convex boundary point set, the spatial distance data between each boundary point, the angular relationship data, and the determined overall coverage information, a spatial boundary description framework that accurately describes the characteristics of the data boundary is constructed.

[0084] Step 3.4: Based on the spatial boundary description framework, a geometric optimization strategy is used to connect key vertices and optimize the shape of the convex boundary point set in the framework, generating a minimum convex polygon that strictly contains all data points. This minimum convex polygon is defined as the validation interval. Specifically, this includes: based on the constructed spatial boundary description framework, a geometric optimization strategy is used to select key vertices of the convex boundary point set in the framework, prioritizing boundary points with large spatial distances, significant angle changes, and a decisive role in the boundary contour as key vertices. Key vertices can effectively represent the core boundary shape of the data distribution; subsequently, the selected key vertices are optimized according to spatial geometric rules. The connection sequence is used to ensure that the connected line segments form a continuous polygonal outline that can roughly encompass all data points. Based on this, shape optimization is performed by adjusting the connection order of key vertices to eliminate intersecting and redundant line segments in the outline. At the same time, the vertex positions of the polygon are fine-tuned to minimize the area of ​​the polygon as much as possible and strictly contain all data points in the multidimensional data point set to avoid data points being missed outside the polygon. After key vertex connection and shape optimization, the smallest convex polygon that strictly contains all data points is finally generated. The smallest convex polygon is formally defined as the verification interval for subsequent layer cutting and feature extraction.

[0085] In this embodiment of the invention, a series of technical means are employed to analyze the dimensional structure and identify the distribution features of the normalized multidimensional data point set, extract the convex boundary point set point by point through incremental convex hull algorithm, construct a spatial boundary description framework by calculating the spatial distance and angle relationship of boundary points, and generate the minimum convex polygon verification interval by using geometric optimization strategy. Therefore, the technical problems of inaccurate characterization of traditional multidimensional data space boundary features, ambiguous definition of data coverage, lack of scientific validity of verification interval and inability to fully encompass data distribution are overcome. Thus, the invention achieves accurate characterization of data space morphology and extension features, clarifies the overall coverage of data distribution, and constructs an efficient verification interval that strictly includes all data points.

[0086] In a preferred embodiment of the present invention, step 3 above may include:

[0087] Step 3.5: Receive the verification interval. Perform density distribution scanning and operating condition change analysis on the multidimensional data points within the verification interval. Calculate the data density gradient distribution map and the operating condition change amplitude matrix to obtain an adaptive parameter set for guiding layered segmentation. Specifically, this includes: receiving the generated minimum convex polygon verification interval, which strictly contains all multidimensional data points and accurately reflects the data space boundary characteristics. To achieve layered segmentation adapted to dynamic changes in operating conditions, firstly, perform density distribution scanning on the multidimensional data points within the verification interval. Divide the verification interval into a preset uniform grid, count the number of data points contained in each grid, calculate the data density of each region based on the difference in the number of data points between grids, and then generate a data density distribution map based on the density difference between adjacent grids. Based on the density gradient distribution map, the trend of data density from high to low within the verification interval is clearly presented. Simultaneously, by combining the operating condition factor data from the multi-dimensional data point set, the operating condition changes are analyzed based on time series analysis. The changes in the values ​​of operating condition factors within different time periods are extracted, and the changes in operating condition factors between adjacent time periods are calculated. All changes in operating condition factors are categorized and organized according to time sequence and dimension, constructing an operating condition change amplitude matrix to comprehensively quantify the dynamic change characteristics of operating condition factors. Finally, by integrating the density change patterns reflected in the data density gradient distribution map and the operating condition fluctuation characteristics reflected in the operating condition change amplitude matrix, key parameters such as the cutting threshold range, cutting direction reference, and sub-region number suggestions are selected to form an adaptive parameter set to guide layered cutting.

[0088] Step 3.6: Based on the adaptive parameter set, perform adaptive hierarchical segmentation on the verification interval. According to the local characteristics of the data density gradient change trend and the magnitude of operating condition changes, dynamically adjust the segmentation threshold and direction to divide the verification interval into multiple sub-regions with consistent internal data distribution characteristics and different feature density levels. Specifically, this includes: initiating the adaptive hierarchical segmentation operation on the verification interval based on the obtained adaptive parameter set; firstly, analyzing the segmentation threshold range in the adaptive parameter set, and combining it with the data density gradient distribution map to determine the initial segmentation threshold for different regions. For regions with large data density gradients, set a smaller segmentation threshold to achieve fine-grained segmentation; for regions with small data density gradients... A coarse-grained segmentation process is employed, using a relatively large segmentation threshold. Simultaneously, referencing the segmentation direction in the adaptive parameter set, and based on the local characteristics reflected by the operating condition variation matrix, if a region exhibits significant operating condition variation with a clear directionality, the segmentation direction is dynamically adjusted to align with the sensitive direction of operating condition variation. If the operating condition variation within a region lacks a clear directionality, segmentation is performed along the main direction of data distribution. During the segmentation process, the internal data distribution characteristics of each potential sub-region are monitored in real time. If a sub-region is found to have data density differences or operating condition variation exceeding preset standards, the segmentation threshold or direction is further adjusted for a secondary segmentation of that region. If the data distribution characteristics within the sub-region remain consistent, the segmentation operation for that region is stopped. Through this dynamic adjustment process, the validation interval is divided into multiple segmented sub-regions with consistent internal data distribution characteristics and different feature density levels.

[0089] Step 3.7: For each formed sub-region, calculate the geometric morphology features, data distribution density features, and boundary morphology features of each sub-region to obtain a multi-dimensional feature parameter set. Specifically, this includes: extracting multi-dimensional feature parameters for each formed sub-region; calculating the area, perimeter, and center coordinates of the sub-region by traversing the boundary vertex coordinates, and analyzing the irregularity of the sub-region boundaries to obtain parameters such as shape factors that characterize the geometric morphology of the sub-region; and calculating the data distribution density features by counting the total number of multi-dimensional data points contained in each sub-region. By combining the area of ​​the sub-region, the number of data points per unit area is calculated, and the clustering of data points within the sub-region is analyzed to obtain the uniformity parameter of data distribution, which comprehensively reflects the data density characteristics of the sub-region. In terms of boundary morphology calculation, the boundary contour of the sub-region is analyzed point by point to calculate the curvature of each point on the boundary line, and the boundary curvature distribution is obtained. At the same time, parameters such as the smoothness and frequency of concavity and convexity changes of the boundary line are statistically analyzed to fully capture the boundary morphology characteristics of the sub-region. The geometric morphology characteristic parameters, data distribution density characteristic parameters, and boundary morphology characteristic parameters of each sub-region are organized to form a multi-dimensional feature parameter set corresponding to each sub-region.

[0090] Step 3.8 involves converting the set of feature parameters, through feature weight allocation and nonlinear mapping, into stability evaluation parameters for the corresponding verification points under the current operating conditions, based on geometric features, distribution density, and boundary curvature. Specifically, this includes: after collecting the multi-dimensional feature parameter set of all segmented sub-regions, firstly, performing feature weight allocation; combining the operating characteristics of the high-speed motor and historical fault data, analyzing the influence of geometric features, distribution density features, and boundary curvature features on the motor's operational stability judgment; assigning higher weights to feature parameters more sensitive to stability changes, and lower weights to feature parameters with smaller impact. Low weights are used to ensure that the weight allocation conforms to the actual operation of the motor, rather than using a fixed weight ratio. Then, a nonlinear mapping relationship is established. The mapping relationship is calibrated based on the physical principle of motor operation and a large amount of historical data, which transforms feature parameters with different dimensions and different value ranges into a unified evaluation interval. Through the nonlinear mapping relationship, the geometric features, distribution density and boundary curvature feature parameters corresponding to each verification point are combined with the allocated feature weights to perform comprehensive calculations, and obtain a value that intuitively reflects the stability of the verification point under the current working conditions. This value is defined as the stability evaluation parameter.

[0091] Step 3.9: Based on the stability evaluation parameters, and combined with the historical anomaly pattern library and real-time operating condition comparison analysis, the feature parameters are further mapped and converted into anomaly evaluation parameters for the corresponding verification points. Specifically, this includes: based on the obtained stability evaluation parameters, calling the preset historical anomaly pattern library, which contains typical anomaly feature data of high-speed motors under different fault types and operating conditions, covering the stability evaluation parameter range and associated features corresponding to various common anomalies such as motor overload, bearing wear, and winding short circuit. The stability evaluation parameters and corresponding multi-dimensional feature parameters of the current verification point are compared one by one with the various anomaly pattern features in the historical anomaly pattern library to calculate the similarity between the current feature and the features of various anomaly patterns. At the same time, combined with real-time operating condition factor data, the differences between the current operating condition and the operating conditions corresponding to the historical anomaly patterns are analyzed, and the similarity calculation results are corrected to avoid misjudgment caused by different operating conditions. Based on the corrected similarity results, the feature parameters of the current verification point are further mapped and converted into anomaly evaluation parameters. The parameters quantify the degree to which the current operating state deviates from the normal state; the higher the value, the greater the anomaly risk.

[0092] In this embodiment of the invention, a series of technical means are employed to generate an adaptive parameter set by scanning the density distribution of multidimensional data points within the verification interval and analyzing changes in operating conditions. Based on this parameter set, the cutting threshold and direction are dynamically adjusted to perform adaptive layered cutting. Multidimensional feature parameters of each cutting sub-region are extracted, and these parameters are converted into stability evaluation parameters through feature weight allocation and nonlinear mapping. Furthermore, by combining a historical abnormal pattern library with real-time operating condition comparison, these parameters are further converted into anomaly evaluation parameters. This overcomes the technical problems of traditional technologies, such as fixed cutting methods that cannot adapt to data density gradients and changes in operating conditions, chaotic data distribution characteristics within sub-regions, one-sided feature parameter extraction, and lack of specificity in evaluation parameters, which make it difficult to accurately reflect the stability and abnormal state of motor operation. As a result, the invention achieves the technical effects of accurate layered cutting of the verification interval, ensuring the consistency of data features in each sub-region, obtaining comprehensive and targeted multidimensional feature parameters, generating stability and anomaly evaluation parameters that can accurately characterize the motor's operating state, and providing high-quality data support for path optimization analysis and dynamic weight calibration.

[0093] In a preferred embodiment of the present invention, step 4 above may include:

[0094] Step 4.1: Receive the stability evaluation parameters and anomaly evaluation parameters from each verification point. Perform spatial mapping and weight allocation on the evaluation parameters to construct the state transition cost matrix between sub-regions within the verification interval. Specifically, this includes: receiving the stability evaluation parameters and anomaly evaluation parameters corresponding to each verification point. These parameters can quantify the stability and anomaly risk of motor operation. To construct a matrix reflecting the transition cost between sub-regions, spatial mapping of the evaluation parameters is first performed, mapping the two evaluation parameters to different dimensions of a multi-dimensional space, so that the parameter states of each sub-region are presented in spatial coordinates, facilitating intuitive calculation of the transition relationship between sub-regions; subsequently, weight allocation is carried out. By combining the safety priority of high-speed motor operation and the analysis of historical fault data, the weight ratio of stability evaluation parameters and anomaly evaluation parameters is determined, with stability parameters having a higher weight than anomaly parameters. This is because stable operation is the foundation of motor safety, and anomaly risks need to be judged based on stable states, ensuring that the weight allocation aligns with actual operational needs. Based on the spatially mapped coordinate data and the allocated weights, the comprehensive cost of changes in stability parameters and anomaly parameters when moving from one sub-region to another is calculated. The transfer costs between all sub-regions are arranged in an orderly manner in rows and columns to construct a complete state transition cost matrix between each sub-region within the verification interval.

[0095] Step 4.2: Based on the constructed state transition cost matrix, starting from the initial sub-region of the verification interval, and comprehensively considering the continuity constraints of the stability evaluation parameters and the jump threshold of the anomaly evaluation parameters, a verification path that minimizes the global evaluation cost is searched to obtain a set of candidate verification paths. Specifically, based on the constructed state transition cost matrix, the initial sub-region of the verification interval is first determined. The sub-region is usually selected from the area with the highest data distribution density, the largest value of the stability evaluation parameter, and the smallest value of the anomaly evaluation parameter, i.e., the core data area corresponding to the normal operating state of the motor. Then, the verification path search process is started. During the search, the continuity constraints of the stability evaluation parameters are strictly followed, requiring... The stability parameter values ​​of adjacent sub-regions do not exceed a preset range to avoid deviations from the actual operating state of the motor due to abrupt changes in stability. At the same time, a jump threshold for the anomaly evaluation parameter is set. If the anomaly parameter values ​​of adjacent sub-regions change by abruptly exceeding this threshold, they are considered unreasonable transfer paths and are excluded to prevent the omission of potential abnormal associations. By traversing all transfer paths in the state transition cost matrix, the global evaluation cost of each path is calculated, which is the sum of the transfer costs of all sub-regions on the path. Multiple paths with the minimum global evaluation cost are selected. These paths cover the state change process from the starting sub-region to the target sub-region at the lowest cost, and finally form a candidate verification path set.

[0096] Step 4.3 involves a multi-dimensional evaluation of the candidate verification path set, calculating the global stability score and anomaly accumulation for each candidate path. Considering path length, computational complexity, and real-time requirements, the optimal verification path is selected, determining the final verification path from the starting sub-region to the target sub-region. Specifically, this includes: conducting a multi-dimensional evaluation of each path in the candidate verification path set; first, calculating the global stability score by averaging the stability evaluation parameters of all verification points along the path, with the score directly reflecting the overall stability level of the path; then calculating the anomaly accumulation by summing the anomaly evaluation parameters of all verification points along the path to quantify the total anomaly risk throughout the path; and simultaneously, counting the number of sub-regions contained in each path. Path length is considered; excessively long paths may cause analysis delays, while paths that are too short may miss critical states. The computational complexity of each path is analyzed, i.e., the computational resources and time consumed during path search, to ensure the feasibility of the path in engineering applications. Considering the requirements for real-time monitoring of high-speed motors, the real-time performance of each path is evaluated, i.e., whether the time from path determination to output results is within acceptable limits. Based on the above multi-dimensional indicators, a weighted scoring method is used to comprehensively score each candidate path, with global stability score and anomaly accumulation having the highest weights, followed by path length, computational complexity, and real-time requirements, with weights decreasing in that order. The path with the highest comprehensive score is selected as the final verification path from the starting sub-region to the target sub-region.

[0097] Step 4.4: Based on the final verification path, perform temporal fusion and spatial aggregation on the evaluation parameters of all verification points along the path, calculate the overall stability index and anomaly risk index of the path, and form a global evaluation output. Specifically, this includes: for the final verification path, performing temporal fusion and spatial aggregation on the stability evaluation parameters and anomaly evaluation parameters of all verification points along the path. During the temporal fusion process, considering the different degrees of influence of data at different time points on the current operating state, the parameters of recently collected verification points have higher weights, and the parameters of earlier collected points have lower weights. Parameters of the same dimension are fused in chronological order using a weighted average method to eliminate local biases caused by time fluctuations. During the spatial aggregation process, the parameters of the verification points are combined with the sub-paths they belong to. The spatial location characteristics of the sub-region are considered, with the verification point parameters in the central region having higher weights than those in the peripheral regions. This is because the data in the central region better represents the core characteristics of the sub-region. The spatially distributed parameters are aggregated by weighted summation to avoid evaluation bias caused by uneven spatial distribution. Based on the results of temporal fusion, the overall stability index of the path is calculated, which is the average value of the fused stability parameters, comprehensively reflecting the overall stability state of the motor corresponding to the path. Based on the results of spatial aggregation, the abnormal risk index of the path is calculated, which is the combined value of the maximum and average values ​​of the aggregated abnormality parameters, highlighting the risk impact of key abnormal points. The overall stability index and the abnormal risk index are integrated to form a comprehensive and objective global evaluation output that reflects the motor's operating status.

[0098] Step 4.5 involves inputting the global evaluation output into a preset conversion function. Based on the deviation between the global evaluation output and the preset benchmark value, the conversion parameters are dynamically adjusted to linearly map the global evaluation output to a correction coefficient between 0 and 1. Specifically, this includes: inputting the global evaluation output into a preset conversion function, which has been calibrated based on historical data from normal motor operation, to map the global evaluation result to correction coefficients that guide weight adjustments; first, determining the preset benchmark value for the global evaluation output, which corresponds to the ideal global evaluation result when the motor is operating normally under standard conditions, including the benchmark stability index and the benchmark anomaly risk index; then calculating the deviation between the current global evaluation output and the preset benchmark value. If the overall stability index is lower than the benchmark value and the anomaly risk index is higher than the benchmark value, the deviation is considered. The degree of deviation is considered a positive deviation, and the larger the deviation value, the more serious the deviation from the normal level of operation; conversely, a negative deviation indicates that the operation is better than the benchmark level. The conversion parameters of the conversion function are dynamically adjusted according to the degree of deviation. The larger the positive deviation, the more sensitive the conversion parameters are to the deviation; the larger the negative deviation, the more the conversion parameters are adjusted to make the correction coefficient tend to a stable intermediate value. Based on the adjusted conversion parameters, the global evaluation output is converted into a correction coefficient between 0 and 1 through a linear mapping method. The closer the correction coefficient is to 1, the larger the deviation of the operation, and the more significant the adjustment of the subsequent weight configuration is required; the closer the correction coefficient is to 0, the closer the operation is to the benchmark level, and the weight configuration does not need to be changed significantly. Finally, a correction coefficient that accurately reflects the deviation of the operation is generated.

[0099] In this embodiment of the invention, a series of technical means are employed to construct a state transition cost matrix by spatial mapping and weight allocation of stability and anomaly evaluation parameters, to search for candidate verification paths with the minimum global evaluation cost based on parameter continuity constraints and jump thresholds, to screen the optimal verification path through multi-dimensional evaluation, to generate a global evaluation output by temporal fusion and spatial aggregation of evaluation parameters on the path, and to dynamically adjust parameters and linearly map them into correction coefficients based on the deviation between the global evaluation output and the preset benchmark value. This overcomes the technical problems of traditional verification path selection lacking global cost consideration, insufficient path rationality due to lack of evaluation parameter constraints, one-sided global evaluation results, and correction coefficients failing to accurately reflect the deviation of motor operating state. Thus, the invention achieves the technical effects of determining the optimal verification path, improving the comprehensiveness and reliability of the global evaluation output, generating correction coefficients that accurately characterize operating state deviations, providing a scientific basis for dynamic weight calibration, and ensuring the accuracy of the overall analysis method.

[0100] In a preferred embodiment of the present invention, step 5 above may include:

[0101] Step 5.1: Receive correction coefficients, perform numerical analysis and calibration direction identification on the correction coefficients, and determine the magnitude and direction parameters of the weight adjustment. Specifically, this includes: receiving the generated correction coefficients between 0 and 1, which accurately reflect the degree of deviation between the current motor operating state and the normal reference state; firstly, performing comprehensive numerical analysis on the correction coefficients to clarify their specific values ​​and determine which segment within the 0-1 range they fall into. Different segments correspond to different deviation levels; the closer the value is to 1, the more severe the deviation, and the closer it is to 0, the closer the operating state is to the ideal reference; subsequently, performing calibration direction identification, combining the numerical characteristics of the correction coefficients and the physical logic of motor operation, if the correction coefficient... If the correction coefficient is greater than the preset intermediate threshold, it indicates a positive deviation in the current operating state, meaning that key operating parameters deviate from the normal range and the risk of anomalies is high. In this case, the calibration direction should be directed to increase the weight of data in dimensions sensitive to anomalies. If the correction coefficient is less than the intermediate threshold, it indicates that the deviation is small or the operating state is better than the baseline. The calibration direction should maintain the existing weights of core dimensions while fine-tuning the weights of secondary dimensions to optimize data representation. The magnitude parameter of weight adjustment is determined through numerical analysis. The larger the difference between the correction coefficient and the intermediate threshold, the larger the magnitude parameter and the stronger the weight adjustment. The direction parameter of weight adjustment is determined by identifying the calibration direction, clarifying which data dimensions need to increase weight and which dimensions need to decrease weight.

[0102] Step 5.2: Based on the amplitude and direction parameters, perform an adaptive calibration operation on the original weight configuration scheme in the dynamic weight integration mechanism. According to the magnitude and sign characteristics of the correction coefficients, dynamically adjust the weight allocation ratio of each data dimension to obtain the calibrated weight configuration scheme. Specifically, this includes: based on the determined amplitude and direction parameters, calling the original weight configuration scheme in the dynamic weight integration mechanism. This original scheme is based on the weight ratios of each data dimension set based on initial experience, which has the defect of not being able to adapt to dynamic changes in operating conditions. First, analyze the current weight ratios of each data dimension in the original weight configuration scheme, including the weight allocation of each dimension of basic motor operation information and each dimension of surrounding operating condition factors. Then, perform an adaptive calibration operation. According to the magnitude of the correction coefficients and the adjustment intensity determined by the amplitude parameters, dynamically adjust the weights of each data dimension. For dimensions that need increased weight as specified by the direction parameters, such as key operating parameters related to current abnormal risks like winding temperature and current, gradually increase their weight ratios according to the amplitude parameters. For dimensions that need decreased weight, such as environmental factors with strong interference or small impact under current operating conditions, appropriately reduce their weight ratios according to the amplitude parameters. During the adjustment process, the constraint that the sum of the weights of all data dimensions is 1 is strictly followed to ensure the rationality of the weight configuration. Simultaneously, considering the sign characteristics of the correction coefficients, if the deviation is positive, the weight adjustment of abnormally sensitive dimensions is significantly increased; if the deviation is negative, fine-tuning is the primary approach to avoid over-adjustment leading to data distortion. Through this dynamic adjustment, a calibrated weight configuration scheme adapted to the current operating status and changes in working conditions is ultimately obtained.

[0103] Step 5.3 involves reconstructing the initial integrated information using a calibrated weight configuration scheme. The multidimensional data point set is recalculated and integrated according to the calibrated weight ratios to eliminate the bias in the original data, resulting in improved initial integrated information with optimized weight distribution. Specifically, after obtaining the calibrated weight configuration scheme, the generated initial integrated information is retrieved. This information contains a pre-processed multidimensional data point set with a unified structure and consistent temporal sequence. However, due to the unreasonable original weight configuration, key features are not prominent and data biases are not eliminated. To address this issue, the initial integrated information is reconstructed using a calibrated weight configuration scheme. First, each data dimension is clearly defined. The corresponding calibrated weight ratio is used, and then each data record in the multidimensional data point set is re-integrated and calculated according to this ratio. During the calculation process, the value of each data dimension is multiplied by its corresponding calibrated weight, so that the key dimension data with higher weight occupies a more important proportion in the integration result, fully highlighting the core features closely related to the motor's operating status; the influence of secondary dimension data with lower weight is reduced, reducing irrelevant interference. Through this weighted reconstruction operation, the bias caused by unreasonable weight configuration in the original data is effectively eliminated, so that the integrated information can more accurately reflect the current actual operating status of the motor, and finally, an improved initial integrated information with optimized weight distribution is obtained.

[0104] In this embodiment of the invention, a series of technical means are employed to determine the magnitude and direction parameters of weight adjustment by numerically analyzing and calibrating the correction coefficients, to perform adaptive calibration on the original weight configuration scheme in the dynamic weight integration mechanism based on these parameters and to dynamically adjust the weight allocation ratio of each data dimension, and to reconstruct the initial integrated information by weighting the calibrated weight configuration scheme to eliminate the deviation of the original data. Therefore, the technical problems of the traditional fixed weight configuration scheme, inability to dynamically adjust the weight ratio of each data dimension according to the deviation of the motor operating state, resulting in unreasonable weight allocation and difficulty in eliminating the deviation of the original data are overcome. Thus, the technical effects of achieving dynamic adaptive optimization of weight configuration, effectively eliminating the influence of deviation in the original data, generating improved initial integrated information with optimized weight distribution, improving the accuracy of data in representing the motor operating state, and providing high-quality data support for the deep fusion of multi-source data are achieved.

[0105] In a preferred embodiment of the present invention, step 6 above may include:

[0106] Step 6.1: Receive the improved initial integrated information and simultaneously acquire the real-time surrounding operating conditions corresponding to the initial integrated information to construct a comprehensive dataset containing motor operating status and environmental conditions. Specifically, this includes: receiving the improved initial integrated information, which has undergone dynamic weight calibration and has an optimized weight distribution, enabling more accurate characterization of the core operating features of the motor; to achieve comprehensive correlation analysis between motor operating status and environmental conditions, the system simultaneously initiates the collection of real-time surrounding operating conditions, acquiring real-time data such as ambient temperature, humidity, air pressure, load fluctuations, cooling system operating status, and installation base vibration under the same time period and operating scenario corresponding to the improved initial integrated information through a distributed sensor network; during the collection process, the improved initial integrated information and real-time surrounding operating conditions are precisely matched according to the timestamp to ensure that each piece of motor operating data corresponds to the operating environment data at the same time point, avoiding analysis bias caused by spatiotemporal misalignment; subsequently, the matched improved initial integrated information and real-time surrounding operating conditions are integrated to construct a comprehensive dataset containing motor operating status parameters and environmental condition parameters.

[0107] Step 6.2: Based on the comprehensive dataset, perform spatiotemporal correlation analysis on the motor operating characteristics and surrounding operating conditions in the improved initial integrated information to identify the coupling relationship between operating status and changes in operating conditions, and obtain collaborative analysis results. Specifically, based on the constructed comprehensive dataset, the system conducts spatiotemporal correlation analysis on the motor operating characteristics and surrounding operating conditions. In the time dimension, the comprehensive dataset is divided into multiple time windows at fixed time intervals. The changing trends of motor operating characteristic parameters within each time window are analyzed, and the changes in surrounding operating conditions during the same period are analyzed accordingly. The synchronous change patterns of operating characteristics and operating conditions under the same time series are explored, such as the winding temperature when the ambient temperature rises. The analysis examines various aspects, including the rate of change of temperature and the response characteristics of current parameters during sudden load increases. Spatially, it considers the structural characteristics of the motor and the installation location of sensors to analyze the impact of operating conditions on the corresponding operating characteristics of different areas. Examples include the influence of cooling system outlet temperature on motor end winding temperature and the impact of base vibration on bearing operating parameters. Through statistical analysis and trend fitting, it identifies positive, negative, or nonlinear coupling relationships between operating states and changes in operating conditions. It clarifies how operating conditions lead to changes in operating parameters, the sensitivity of these changes, and the triggering conditions for coupling effects. The correlations, sensitivities, and triggering conditions are then compiled to form detailed collaborative analysis results.

[0108] Step 6.3 involves performing a difference comparison operation on the collaborative analysis results. This calculates the deviation between the current operating state and the historical baseline state under the same operating conditions, quantifies the degree of abnormal fluctuation in motor operating parameters, and generates a difference comparison report. Specifically, this includes: accessing the historical database, selecting historical normal operating data from the database that is consistent with or highly similar to the operating conditions of the current comprehensive dataset, and identifying these as historical baseline state data to ensure the fairness and effectiveness of the comparison; based on the collaborative analysis results, focusing on key parameters of motor operation, such as speed, current, voltage, winding temperature, and bearing temperature, calculating the deviation values ​​of these key parameters under the current operating state from the corresponding parameters under the historical baseline state, including numerical deviation and trend deviation. For numerical deviations, the difference and relative percentage between the current parameter value and the historical benchmark parameter value are directly calculated. For trend deviations, the degree of trend difference is quantified by comparing the slope of the parameter change curve over time, fluctuation frequency, and other characteristics. At the same time, combined with the coupling relationship identified by collaborative analysis, reasonable deviations caused by normal fluctuations in operating conditions are eliminated, and only abnormal deviations exceeding the normal range are retained. This quantifies the degree of abnormal fluctuations in motor operating parameters, clarifies the parameter type, fluctuation amplitude, duration, and corresponding operating condition triggering conditions of abnormal fluctuations, and compiles and summarizes the calculation results and analysis conclusions to form a difference comparison report containing a list of abnormal parameters, deviation values, fluctuation degrees, and triggering conditions.

[0109] Step 6.4: Based on the difference comparison report, deeply integrate the improved initial integrated information with surrounding operating condition factors. Use a multi-source data fusion algorithm to eliminate operating condition interference factors, extract the essential characteristics of motor operation, and obtain corrected operating status evaluation data. Specifically, this includes: initiating the deep integration of the improved initial integrated information with surrounding operating condition factors based on the difference comparison report; firstly, extracting operating condition interference-related information from the difference comparison report to clarify that abnormal deviations are interference deviations caused by fluctuations in operating condition factors, and are essential deviations caused by abnormal motor operating states; subsequently, using a multi-source data fusion algorithm, taking the motor operating characteristic data and operating condition factor data from the integrated dataset as input, and using the algorithm to decompose, filter, and reorganize the data. Interference factors under operating conditions are treated as noise items and separated and eliminated. For example, the winding temperature deviation caused by normal fluctuations in ambient temperature is deducted from the actual measured value, and the current fluctuation caused by load fluctuations is corrected to the equivalent current value under standard load. In the process of eliminating interference, the coupling relationship in the collaborative analysis results is combined to ensure the accuracy of the elimination operation and avoid mistakenly eliminating abnormal signals of the motor itself as interference. Through the processing of multi-source data fusion algorithms, the influence of interference factors under operating conditions on motor operating parameters is eliminated, and the essential characteristics reflecting the motor's own health status are accurately extracted, such as the true heating state of the windings after eliminating environmental influences and the inherent operating stability of the motor after eliminating load fluctuations. These essential characteristic data are organized into corrected operating status evaluation data.

[0110] Step 6.5: Standardize and encapsulate the corrected operating status assessment data, adding timestamps, operating condition identifiers, and confidence level labels. Output correction completion information containing the comprehensive assessment results of motor operating status and correction suggestions. Specifically, this includes: standardizing and encapsulating the corrected operating status assessment data. First, according to industry-standard data formats, the field names, data types, numerical precision, and storage formats of the assessment data are uniformly standardized to ensure data universality and readability. Then, a precise timestamp is added to each piece of assessment data, accurate to the millisecond level, clearly indicating the corresponding motor operating time point. Operating condition identifiers are added, classifying and labeling the data according to current operating conditions, such as load level, ambient temperature level, and operating mode. Finally, confidence level labels are added, specifying the confidence level. The tags are calculated based on factors such as the iterative convergence accuracy, data integrity, and operating condition matching degree of the multi-source data fusion algorithm, intuitively reflecting the reliability of the corrected evaluation data. After encapsulation, based on the corrected operating status evaluation data, combined with the historical abnormal pattern library and the motor's operating safety threshold, a comprehensive evaluation of the motor's current operating status is performed to determine whether the motor is in a normal operating state, a slightly abnormal state, or a severely abnormal state. Based on the evaluation results, combined with the difference comparison report and collaborative analysis results, targeted correction suggestions are generated, such as adjusting cooling power and optimizing load distribution for slightly abnormal situations, and suggesting stopping the machine to check bearing wear and checking winding insulation status for severely abnormal situations. Finally, the comprehensive evaluation results and correction suggestions are integrated with the encapsulated evaluation data to output complete correction completion information.

[0111] In this embodiment of the invention, by employing a series of technical means—including synchronously acquiring real-time surrounding operating condition factors and improved initial integrated information to construct a comprehensive dataset, performing spatiotemporal correlation analysis on motor operating characteristics and operating condition factors to identify coupling relationships, quantifying the deviation between the current and historical baseline states through difference comparison, using multi-source data fusion algorithms to eliminate operating condition interference and extract the essential characteristics of motor operation, and standardizing and encapsulating the corrected data to output completion information containing comprehensive evaluation results and correction suggestions—the invention overcomes the technical problems of traditional technologies, such as insufficient integration of operating condition factors and improved data for collaborative analysis, inaccurate identification of the coupling relationship between operating status and operating condition changes, difficulty in quantifying the degree of abnormal fluctuations, inability to effectively eliminate operating condition interference, and lack of standardized format and targeted correction guidance in the output results. This achieves the technical effects of accurately capturing the intrinsic correlation between motor operating status and operating condition changes, accurately quantifying the degree of abnormal fluctuations, effectively removing operating condition interference to highlight the essential characteristics of operation, outputting standardized and highly reliable operating status evaluation results and targeted correction suggestions, and providing scientific decision-making support for the precise maintenance and safe operation of high-speed motors.

[0112] like Figure 2As shown, embodiments of the present invention also provide a high-speed motor operating data intelligent analysis system, including:

[0113] The acquisition module is used to collect basic operating information of the motor and surrounding operating conditions.

[0114] The integration module is used to perform preprocessing operations on basic operational information and surrounding working conditions to form initial integrated information.

[0115] The calculation module is used to calculate the boundary contour of the multidimensional data point set based on the initial integrated information, and extract the convex boundary point set of the data distribution; determine the coverage of the data distribution based on the convex boundary point set, and establish the minimum convex polygon verification interval containing all data points; adaptively divide the verification interval into multiple sub-regions with different feature densities according to the data density gradient and the magnitude of the change in working conditions; extract the geometric features, distribution density and boundary curvature of each sub-region as feature parameters, and map the feature parameters into the stability evaluation parameters and anomaly evaluation parameters of the corresponding verification points;

[0116] The optimization module is used to perform path optimization analysis on the stability evaluation parameters and anomaly evaluation parameters of each verification point, determine the final verification path from the starting sub-region of the verification interval to the target sub-region, and convert the global evaluation output corresponding to the final verification path into correction coefficients.

[0117] The calibration module is used to calibrate the weight configuration scheme in the dynamic weight integration mechanism based on the correction coefficient, and generate improved initial integration information.

[0118] The processing module is used to perform collaborative analysis, difference comparison and deep integration of the improved initial integrated information with surrounding operating conditions and factors, and output correction completion information.

[0119] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent analysis of high-speed motor operating data, characterized in that, The method includes: Collect basic operating information of the motor and surrounding operating conditions; Preprocessing operations are performed on basic operational information and surrounding working conditions to form initial integrated information; Based on the initial integrated information, the boundary contour of the multidimensional data point set is calculated to extract the convex boundary point set of the data distribution. Based on the convex boundary point set, the coverage of the data distribution is determined, and a minimum convex polygon verification interval containing all data points is established. The verification interval is adaptively layered and cut according to the data density gradient and the magnitude of the change in working conditions to form multiple cut sub-regions with different feature densities. The geometric features, distribution density, and boundary curvature of each cut sub-region are extracted as feature parameters, and the feature parameters are mapped and converted into stability evaluation parameters and anomaly evaluation parameters for the corresponding verification points. Path optimization analysis is performed on the stability and anomaly assessment parameters of each verification point to determine the final verification path from the starting sub-region to the target sub-region of the verification interval. The global assessment output corresponding to the final verification path is converted into correction coefficients. This includes: receiving the stability and anomaly assessment parameters of each verification point; performing spatial mapping and weight allocation on the assessment parameters; constructing the state transition cost matrix between sub-regions within the verification interval; and based on the constructed state transition cost matrix, starting from the starting sub-region of the verification interval, comprehensively considering the continuity constraints of the stability assessment parameters and the jump threshold of the anomaly assessment parameters, searching for the verification path that minimizes the global assessment cost to obtain candidate verification paths. The candidate verification path set is evaluated from multiple dimensions. The global stability score and anomaly accumulation degree of each candidate path are calculated. The optimal verification path is selected by considering path length, computational complexity, and real-time requirements. The final verification path from the starting sub-region to the target sub-region is determined. Based on the final verification path, the evaluation parameters of all verification points on the path are fused temporally and spatially. The overall stability index and anomaly risk index of the path are calculated to form a global evaluation output. The global evaluation output is input into a preset transformation function. The transformation parameters are dynamically adjusted according to the deviation between the global evaluation output and the preset benchmark value. The global evaluation output is linearly mapped to a correction coefficient between 0 and 1. The weight configuration scheme in the dynamic weight integration mechanism is calibrated based on the correction coefficient to generate improved initial integration information; The improved initial integrated information is collaboratively analyzed, the differences are compared and deeply integrated with the surrounding operating conditions, and the correction completion information is output.

2. The intelligent analysis method for high-speed motor operating data according to claim 1, characterized in that, Preprocessing operations are performed on basic operational information and surrounding operating conditions to form initial integrated information, including: The system collects basic operating information of the motor and surrounding operating conditions from the data source, and performs format standardization processing on the collected data to ensure data format consistency, resulting in data with a unified format. Perform missing value imputation and outlier removal operations on data after it has been standardized to improve data integrity; The processed data is dimensionally aligned and time-series synchronized to construct a multidimensional data point set; The multidimensional data point set is normalized to eliminate the influence of dimensions and form initial integrated information.

3. The intelligent analysis method for high-speed motor operating data according to claim 2, characterized in that, Based on the initial integrated information, boundary contour calculations are performed on the multidimensional data point set to extract the convex boundary point set of the data distribution. Based on this convex boundary point set, the coverage area of ​​the data distribution is determined, and a minimum convex polygon verification interval encompassing all data points is established, including: Receive initial integrated information, perform dimensional structure analysis and distribution feature identification on the multidimensional data point set after normalization of the initial integrated information, and generate analytical results that characterize the spatial form of the data. Based on the analytical results, the incremental convex hull algorithm is used to perform boundary contour calculation on the multidimensional data point set. The convex boundary point set that completely represents the extensional shape of the data distribution is identified and extracted from the analytical results through point-by-point iteration. By using the set of convex boundary points, the spatial distance and angular relationship between each boundary point are calculated. Based on the spatial geometric characteristics of the set of convex boundary points, the overall coverage of the data distribution is determined, and a spatial boundary description framework that accurately describes the characteristics of the data boundary is constructed. Based on the spatial boundary description framework, a geometric optimization strategy is adopted to connect key vertices and optimize the shape of the set of convex boundary points in the framework, generating the minimum convex polygon that strictly contains all data points, and defining the minimum convex polygon as the verification interval.

4. The intelligent analysis method for high-speed motor operating data according to claim 3, characterized in that, The validation interval is adaptively segmented according to the data density gradient and the magnitude of operating condition changes, forming multiple segmented sub-regions with different feature densities. The geometric features, distribution density, and boundary curvature of each segmented sub-region are extracted as feature parameters, and these feature parameters are mapped and converted into stability evaluation parameters and anomaly evaluation parameters for the corresponding validation points, including: The system receives the verification interval, performs density distribution scanning and operating condition change analysis on the multidimensional data points within the verification interval, calculates the data density gradient distribution map and the operating condition change amplitude matrix, and obtains an adaptive parameter set to guide the stratification and segmentation. Based on the adaptive parameter set, an adaptive hierarchical cutting operation is performed on the verification interval. According to the local characteristics of the data density gradient change trend and the working condition change magnitude, the cutting threshold and cutting direction are dynamically adjusted to divide the verification interval into multiple cutting sub-regions with consistent internal data distribution characteristics and different feature density levels. For each of the formed sub-regions, the geometric morphological features, data distribution density features, and boundary morphological features of each sub-region are calculated to obtain a multi-dimensional set of feature parameters. The set of feature parameters is transformed into stability evaluation parameters for the corresponding verification points under the current working conditions through feature weight allocation and nonlinear mapping relationship. Based on stability evaluation parameters, and combined with historical anomaly pattern database and real-time operating condition comparison analysis, the feature parameters are further mapped and converted into anomaly evaluation parameters for corresponding verification points.

5. The intelligent analysis method for high-speed motor operating data according to claim 4, characterized in that, The weight configuration scheme in the dynamic weight integration mechanism is calibrated based on the correction coefficient to generate improved initial integration information, including: Receive correction coefficients, perform numerical analysis and calibration direction identification on the correction coefficients, and determine the magnitude and direction parameters of the weight adjustment; Based on the amplitude and direction parameters, an adaptive calibration operation is performed on the original weight configuration scheme in the dynamic weight integration mechanism. According to the magnitude and sign characteristics of the correction coefficient, the weight allocation ratio of each data dimension is dynamically adjusted to obtain the calibrated weight configuration scheme. By using a calibrated weight configuration scheme, the initial integrated information is reconstructed with weights, and the multidimensional data point set is re-integrated and calculated according to the calibrated weight ratios to eliminate the influence of biases in the original data and obtain improved initial integrated information with optimized weight distribution.

6. The intelligent analysis method for high-speed motor operating data according to claim 5, characterized in that, The improved initial integrated information is collaboratively analyzed, compared, and deeply integrated with surrounding operating conditions to output correction completion information, including: Receive the improved initial integration information and simultaneously acquire the real-time surrounding operating conditions corresponding to the initial integration information to construct a comprehensive dataset containing motor operating status and environmental conditions; Based on the comprehensive dataset, a spatiotemporal correlation analysis was performed on the motor operating characteristics and surrounding operating conditions in the improved initial integrated information to identify the coupling relationship between operating status and changes in operating conditions, and to obtain collaborative analysis results. Perform a difference comparison operation on the collaborative analysis results, calculate the deviation between the current operating status and the historical benchmark status under the same working conditions, quantify the degree of abnormal fluctuation of motor operating parameters, and generate a difference comparison report. Based on the difference comparison report, the improved initial integrated information and surrounding operating conditions are deeply integrated. The operating condition interference factors are eliminated by multi-source data fusion algorithm, the essential characteristics of motor operation are extracted, and the corrected operating status evaluation data is obtained. The corrected operating status assessment data is standardized and packaged, and timestamps, operating condition identifiers, and confidence level labels are added. The output includes correction completion information containing comprehensive assessment results of motor operating status and correction suggestions.

7. A high-speed motor operation data intelligent analysis system, wherein the system implements the method as described in any one of claims 1 to 6, characterized in that, include: The acquisition module is used to collect basic operating information of the motor and surrounding operating conditions. The integration module is used to perform preprocessing operations on basic operational information and surrounding working conditions to form initial integrated information. The calculation module is used to calculate the boundary contour of the multidimensional data point set based on the initial integrated information, extract the convex boundary point set of the data distribution, determine the coverage of the data distribution based on the convex boundary point set, and establish the minimum convex polygon verification interval containing all data points. The verification interval is adaptively segmented according to the data density gradient and the magnitude of the change in operating conditions to form multiple segmented sub-regions with different feature densities; the geometric features, distribution density and boundary curvature of each segmented sub-region are extracted as feature parameters, and the feature parameters are mapped and converted into stability evaluation parameters and anomaly evaluation parameters of the corresponding verification points. The optimization module performs path optimization analysis on the stability and anomaly assessment parameters of each verification point, determines the final verification path from the starting sub-region to the target sub-region of the verification interval, and converts the global assessment output corresponding to the final verification path into correction coefficients. This includes: receiving the stability and anomaly assessment parameters of each verification point; performing spatial mapping and weight allocation on the assessment parameters; constructing a state transition cost matrix between sub-regions within the verification interval; and based on the constructed state transition cost matrix, starting from the starting sub-region of the verification interval, comprehensively considering the continuity constraints of the stability assessment parameters and the jump threshold of the anomaly assessment parameters, searching for the verification path that minimizes the global assessment cost, and obtaining candidate... Select a set of verification paths; evaluate the candidate verification paths from multiple dimensions, calculate the global stability score and anomaly accumulation degree of each candidate path, and select the optimal verification path based on path length, computational complexity and real-time requirements to determine the final verification path from the starting sub-region to the target sub-region; based on the final verification path, perform temporal fusion and spatial aggregation on the evaluation parameters of all verification points on the path, calculate the overall stability index and anomaly risk index of the path, and form a global evaluation output; input the global evaluation output into a preset transformation function, and dynamically adjust the transformation parameters according to the degree of deviation between the global evaluation output and the preset benchmark value, linearly mapping the global evaluation output to a correction coefficient between 0 and 1; The calibration module is used to calibrate the weight configuration scheme in the dynamic weight integration mechanism based on the correction coefficient, and generate improved initial integration information. The processing module is used to perform collaborative analysis, difference comparison and deep integration of the improved initial integrated information with surrounding operating conditions and factors, and output correction completion information.

8. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images