Health grading method and system for a driveline, electronic device and storage medium

Abstract

The application provides a health degree grading evaluation method and system of a transmission chain, an electronic device and a storage medium. The method comprises the following steps: collecting temperature time series data and working condition data of measuring points on a transmission chain of a wind turbine; pre-processing the temperature time series data to obtain standardized time series samples; establishing a temperature time series reference behavior baseline of each measuring point according to the working condition data collected in different working condition intervals and the standardized time series samples under a preset normal operation condition; constructing a temperature time series confidence statistical boundary of each measuring point according to statistical distribution characteristics output by the reference behavior baseline; calculating a health degree index according to deviation characteristics of real-time standardized time series samples relative to the temperature time series confidence statistical boundary; and outputting a health grade and an abnormal part positioning result according to the health degree index. Through working condition layered modeling, confidence boundary construction and health scoring of multiple temperature measuring points of the transmission chain, a systematic and engineering feasible technical solution is provided for predictive operation and maintenance of the transmission chain of the wind turbine.

Description

Technical Field

[0001] This application relates to the field of wind power generation and equipment condition monitoring, and in particular to a method, system, electronic device and storage medium for assessing the health status of a transmission chain. Background Technology

[0002] The drivetrain of a wind turbine is the core mechanical system that converts wind energy into electrical energy. Enduring alternating loads, lubrication aging, and environmental temperature fluctuations over long periods, the drivetrain is one of the subsystems with the highest failure rate in wind turbines. Failure of drivetrain components not only leads to unplanned downtime but also results in substantial economic losses due to difficulties in installation and repair, and high spare parts costs. Therefore, online monitoring and early warning of the drivetrain's health status are core requirements for predictive maintenance of wind turbines.

[0003] Temperature is one of the most direct and sensitive indicators of the health status of the drivetrain. However, because drivetrain temperature is strongly affected by operating conditions such as wind speed, power, and ambient temperature, the false alarm rate of a single fixed threshold temperature alarm method is extremely high, making it difficult to distinguish between normal operating condition fluctuations and temperature anomalies caused by actual faults, thus limiting its practical value in actual operation and maintenance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application proposes a method, system, electronic device, and storage medium for assessing the health of a transmission chain, thereby resolving the problems in the prior art.

[0005] This application provides a method for grading and assessing the health of a transmission chain, the method comprising:

[0006] Simultaneously collect temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine's drive train;

[0007] The temperature time series data is preprocessed to obtain standardized time series samples;

[0008] Based on the operating condition data collected synchronously under different operating condition intervals and the standardized time series samples under preset normal operating conditions, operating condition hierarchical modeling is performed to establish the temperature time series benchmark behavior baseline for each measuring point.

[0009] Based on the statistical distribution characteristics of the baseline output of the benchmark behavior, the temperature time-series confidence statistical boundary of each measurement point is constructed;

[0010] The health index is calculated based on the deviation characteristics of the real-time standardized time-series samples relative to the confidence statistical boundary of the temperature time series.

[0011] The health level of the wind turbine drive chain and the location of abnormal parts are output based on the health index.

[0012] In a specific embodiment, the step of performing condition-level hierarchical modeling based on the synchronously collected operating condition data under different operating condition intervals and the standardized time-series samples under preset normal operating conditions, and establishing a temperature time-series benchmark behavior baseline for each measuring point, includes:

[0013] Using the active power in the operating data as the main hierarchical variable, the full power operating range of the wind turbine is divided into multiple power ranges at equal intervals.

[0014] Using the ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature.

[0015] For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under preset normal operating conditions collected synchronously with the determined working condition data are determined, and a temperature time series reference behavior baseline is established for each measuring point.

[0016] In a specific embodiment, for each of the measurement points in each working condition grid in the two-dimensional working condition hierarchical grid, if the number of standardized time series samples is lower than a preset minimum threshold, the working condition grid of the measurement point is smoothly supplemented based on the standardized time series samples in the adjacent working condition grids. Based on the working condition dataset and standardized time series samples in each working condition grid after smoothing and supplementation, a temperature time series benchmark behavior baseline for each measurement point is established.

[0017] In one specific embodiment, the method further includes:

[0018] Based on the standardized time series samples under the preset normal operating conditions within the latest time range, the mean value of each measurement point in each working condition grid in the two-dimensional working condition hierarchical grid is updated in a rolling manner, and the temperature time series benchmark behavior baseline of each measurement point is updated based on the updated data of each working condition grid.

[0019] In a specific embodiment, the step of constructing the temperature time-series confidence statistical boundary of each of the measuring points based on the statistical distribution characteristics of the baseline output of the benchmark behavior includes: constructing the temperature time-series confidence statistical boundary of each of the measuring points using one or more of the following methods: sliding time window statistics, quantile statistics, confidence interval estimation, or control chart, based on the statistical distribution characteristics of the baseline output of the benchmark behavior.

[0020] In one specific embodiment, the preprocessing includes one or more of the following: time alignment, missing value compensation, outlier removal, and normalization.

[0021] The measuring points include two or more of the following: the front bearing temperature measuring point of the gearbox, the rear bearing temperature measuring point of the gearbox, the main bearing temperature measuring point, the generator bearing temperature measuring point, and the lubricating oil temperature measuring point.

[0022] The operating data includes one or more of the following: wind speed, ambient temperature, active power, speed, torque, pitch angle, and unit start-up / shutdown status.

[0023] The deviation features include one or more of the following: deviation amplitude, deviation duration, and multi-point coupled deviation features.

[0024] This application also proposes a transmission chain health grading assessment system, the system comprising:

[0025] The data acquisition module is used to synchronously acquire temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine's drive train.

[0026] The data preprocessing module is used to preprocess the temperature time series data to obtain standardized time series samples;

[0027] The benchmark modeling module is used to perform layered modeling of operating conditions based on the operating condition data collected synchronously under different operating condition intervals and the standardized time series samples under preset normal operating conditions, and to establish the temperature time series benchmark behavior baseline for each measuring point.

[0028] The confidence boundary construction module is used to construct the temperature time-series confidence statistical boundary for each measurement point based on the statistical distribution characteristics of the baseline output of the reference behavior.

[0029] The health assessment module is used to calculate a health index based on the deviation characteristics of the real-time standardized time-series samples relative to the temperature time-series confidence statistical boundary.

[0030] The graded output module is used to output the health level of the wind turbine drive chain and the location results of abnormal parts based on the health index.

[0031] Furthermore, the benchmark modeling module is used for:

[0032] Using the active power in the operating data as the main hierarchical variable, the full power operating range of the wind turbine is divided into multiple power ranges at equal intervals.

[0033] Using the ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature.

[0034] For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under preset normal operating conditions collected synchronously with the determined working condition data are determined, and a temperature time series reference behavior baseline is established for each measuring point.

[0035] This application also proposes an electronic device, including:

[0036] One or more processors;

[0037] Memory, used to store one or more programs;

[0038] When the one or more programs are executed by the one or more processors, the one or more processors implement the above-described transmission chain health grading assessment method.

[0039] This application also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described transmission chain health grading assessment method.

[0040] This application proposes a method, system, electronic device, and storage medium for assessing the health of a drivetrain. The method includes: synchronously collecting temperature time-series data and operating condition data from multiple preset measuring points on the drivetrain of a wind turbine; preprocessing the temperature time-series data to obtain standardized time-series samples; performing operating condition stratification modeling based on the synchronously collected operating condition data under different operating condition intervals and the standardized time-series samples under preset normal operating conditions to establish a temperature time-series baseline behavior for each measuring point; constructing a temperature time-series confidence statistical boundary for each measuring point based on the statistical distribution characteristics output by the baseline behavior; calculating a health index based on the deviation characteristics of the real-time standardized time-series samples relative to the temperature time-series confidence statistical boundary; and outputting the health level and abnormal location results of the wind turbine drivetrain based on the health index. This solution achieves condition-adaptive and multi-component hierarchical quantitative assessment of the health status of the drivetrain by performing hierarchical modeling of multiple temperature measurement points in the drivetrain, constructing dynamic confidence boundaries, and scoring health at multiple measurement points. It effectively solves the problems of high false alarm rate and insufficient positioning accuracy of single measurement point analysis in the fixed threshold method, and provides a systematic and engineering-feasible technical solution for predictive operation and maintenance of wind turbine drivetrains. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart illustrating a method for grading and evaluating the health of a transmission chain according to an embodiment of this application.

[0043] Figure 2 This is a schematic diagram of the framework structure of a transmission chain health grading assessment system proposed in an embodiment of this application;

[0044] Figure 3 This is a schematic diagram of the frame structure of an electronic device proposed in an embodiment of this application. Detailed Implementation

[0045] Various embodiments of this disclosure will be described more fully below. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0046] The terminology used in the various embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the various embodiments of this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this disclosure pertain. Terms (such as those defined in a generally used dictionary) are to be interpreted as having the same meaning as in the context of the relevant technical field and are not to be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this disclosure.

[0047] Example 1

[0048] Embodiment 1 of this application discloses a method for grading and evaluating the health of a transmission chain, such as... Figure 1 As shown, the method includes the following steps:

[0049] Step 101: Synchronously collect temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine's drive chain;

[0050] First, data needs to be acquired as input for the entire system. Specifically, temperature time-series data of key measurement points in the transmission chain and corresponding unit operating condition data can be obtained in real time from the wind turbine SCADA (Supervisory Control And Data Acquisition) system, providing a sufficient data foundation for subsequent modeling and evaluation.

[0051] The temperature time-series data acquisition covers all key thermal state measurement points in the drivetrain. Specific measurement points include two or more of the following: gearbox front bearing temperature measurement point, gearbox rear bearing temperature measurement point, main bearing temperature measurement point, generator bearing temperature measurement point, and lubricating oil temperature measurement point. Among these, the front and rear bearing temperatures directly reflect the frictional thermal state of the high-speed and low-speed gear bearings, serving as the most sensitive early indicators of gear wear and bearing spalling. The main bearing temperature reflects the operating state of the main shaft support bearing; abnormal main bearing temperature often indicates grease failure or internal bearing damage. The lubricating oil temperature comprehensively reflects the overall thermal balance state of the gearbox; persistently high lubricating oil temperature is a systemic signal of decreased gearbox heat dissipation capacity or increased internal heat sources. The generator bearing temperature reflects the thermal state of the generator's non-drive end bearings and is closely related to the generator's alignment and bearing lubrication conditions. All of the above measurement points are standard data acquisition parameters for wind turbine SCADA systems, requiring no additional sensor investment and demonstrating high engineering feasibility.

[0052] Operating data includes one or more of the following: wind speed, ambient temperature, active power, speed, torque, pitch angle, and unit start-up / shutdown status.

[0053] Synchronous acquisition of operating condition data is a prerequisite for achieving adaptive operating condition analysis. This involves synchronously acquiring operating condition parameters such as active power, generator speed, wind speed, ambient temperature, pitch angle, and unit start-up / shutdown status. Active power and speed determine the mechanical load level of the drivetrain and are the most significant operating condition factors affecting the temperature baseline. Ambient temperature systematically influences the temperature at each measuring point through heat dissipation boundary conditions. Pitch angle reflects the current power regulation status. Unit start-up / shutdown status is used to identify abnormal operating phases such as shutdown and warm-up, so that these can be excluded from the effective data range for modeling and evaluation during preprocessing.

[0054] The temporal resolution of data acquisition can be set in layers according to the type of measurement point. For example, temperature time-series data is usually acquired with a 1-minute acquisition cycle, consistent with the standard acquisition frequency of SCADA systems; operating condition data is acquired synchronously with the same cycle to ensure accurate time correspondence between temperature samples and operating condition labels. The system timestamps the data at the acquisition end, providing a basis for subsequent time alignment processing.

[0055] In this way, by collaboratively collecting multiple key temperature measurement points and operating condition data of the drive train, a complete data input is provided for operating condition adaptive modeling and multi-measurement point coupled health assessment. Moreover, it can make full use of the existing SCADA infrastructure to achieve low-cost and high-coverage drive train thermal state perception.

[0056] Step 102: Preprocess the temperature time series data to obtain standardized time series samples;

[0057] Specifically, preprocessing includes one or more of the following: time alignment, missing value compensation, outlier removal, and standardization. Through preprocessing, the raw temperature time-series data can be systematically cleaned and normalized, eliminating the adverse effects of data noise, missing values, and outliers on the accuracy of subsequent modeling and evaluation. Specific preprocessing steps include the following:

[0058] Time alignment processing is used to resolve the inconsistency in timestamps caused by acquisition clock deviations or transmission delays in data from different measurement points. This time alignment process resamples and interpolates the data from each measurement point using a unified UTC (Coordinated Universal Time) time base, ensuring that the temperature vectors of multiple measurement points at the same time are strictly corresponding in the time dimension, laying the foundation for subsequent multi-measurement point coupling analysis.

[0059] Missing data compensation addresses temperature data loss caused by communication failures, sensor outages, or SCADA system maintenance. For example, for short-term data loss of no more than 10 minutes, linear interpolation is used for compensation; for longer periods of data loss exceeding 10 minutes, this period is marked as invalid and excluded from the effective calculation window for baseline modeling and health assessment, and is highlighted in gray in the assessment interface to alert maintenance personnel to the data acquisition status.

[0060] Outlier removal is used to identify and remove dirty data caused by sensor malfunctions, electromagnetic interference, or communication errors. This outlier removal process employs a robust statistical method based on the moving median. It calculates the median and absolute median difference (MAD) for each measurement point within a moving window (default window width 60 minutes). Data points deviating from the median by more than 5 times the MAD are marked as outliers and replaced with linear interpolation. Compared to the mean ± 3σ method, this method has stronger resistance to outlier contamination.

[0061] Standardization processes involve normalizing temperature time-series data under different operating conditions to eliminate the impact of temperature magnitude differences across different operating condition intervals on subsequent statistical modeling. Specifically, the mean and standard deviation of normal samples within each operating condition stratification interval are calculated. The temperature data is then standardized to zero mean and unit variance using a Z-score transformation, ensuring comparability of temperature deviations across different operating condition intervals and facilitating the setting of unified confidence boundary thresholds. In essence, the standardization process for temperature time-series data can be expressed as follows:

[0062]

[0063] in: and Representing the measuring points At any moment Standardized temperature values ​​and original temperature values; and Representing the measuring points The mean temperature and standard deviation of the normal sample under the corresponding operating conditions.

[0064] Data preprocessing effectively eliminated temporal inconsistencies, missing data, and abnormal noise in the original temperature data, significantly improving the data quality for the modeling and evaluation stages and providing a solid data foundation for the accurate construction of confidence boundaries and the reliable calculation of health scores.

[0065] Step 103: Based on the synchronously collected operating condition data under different operating condition intervals and the standardized time series samples under the preset normal operating conditions, perform operating condition layered modeling and establish the temperature time series benchmark behavior baseline for each measuring point.

[0066] Specifically, step 103 includes: using the active power in the operating data as the main hierarchical variable, and dividing the full power operating range of the wind turbine into multiple power ranges at equal intervals;

[0067] Using ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature.

[0068] For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under the preset normal operating conditions collected synchronously with the determined working condition data are determined to establish the temperature time series reference behavior baseline for each measuring point.

[0069] Specifically, based on the systematic stratification of unit operating conditions, a benchmark behavior model (i.e., temperature time series benchmark behavior baseline) reflecting the normal thermal state behavior law is established for each temperature measurement point, providing a statistical basis for the construction of confidence boundaries.

[0070] Operating condition stratification is the first step in baseline modeling and the core mechanism for achieving adaptive adjustment of temperature confidence boundary operating conditions. This module uses active power as the primary stratification variable, dividing the unit's full-power operating range (from cut-in power to rated power) into several equally spaced operating condition intervals (e.g., 10 intervals by default, each interval width being 10% of rated power). Ambient temperature is used as an auxiliary stratification variable, further subdividing each power interval into three ambient temperature segments: low temperature, medium temperature, and high temperature, forming a two-dimensional operating condition stratification grid of power × ambient temperature. For each operating condition grid, normal operating temperature samples under that condition are selected from historical data to independently establish a temperature time-series benchmark behavior baseline.

[0071] Screening normal operation samples is crucial to ensuring the purity of the baseline model. This module uses manually verified normal operation data (typically data from 3 to 6 months after commissioning) as the candidate training set, and automatically eliminates abnormal operation periods using the following rules: excluding data from the unit start-up and shutdown transition period (data within 30 minutes after startup and 30 minutes before shutdown); excluding data from historical warning event recording periods; and performing outlier pre-screening on candidate samples based on the temperature distribution within the operating condition cells, removing sample points that deviate from the median within the operating condition cell by more than 3 times the IQR (Interquartile Range), ensuring that the training sample set for each operating condition cell only contains representative temperature data under true healthy conditions.

[0072] In addition, for each working condition cell in the two-dimensional working condition hierarchical grid, if the number of standardized time series samples is lower than the preset minimum threshold, the working condition cell of the measuring point is smoothly supplemented based on the standardized time series samples in the adjacent working condition cells. Based on the working condition dataset and standardized time series samples in each working condition cell after smoothing and supplementation, the temperature time series benchmark behavior baseline of each measuring point is established.

[0073] The baseline for temperature time-series benchmark behavior is established based on the statistical characteristics of normal samples within each operating condition cell. For each measurement point within each operating condition cell, the module calculates the mean μ and standard deviation σ of the normal samples, forming the normal distribution benchmark parameters for that measurement point in that operating condition cell. When the number of normal samples in a certain operating condition cell is lower than the minimum sample size threshold (default 50), the module automatically borrows samples from adjacent operating condition cells for smoothing and supplementation, avoiding the instability problem of benchmark estimation in sparsely sampled operating condition cells.

[0074] Specifically, the average temperature under each operating condition can be expressed as:

[0075]

[0076] The standard deviation of temperature under each operating condition can be expressed as:

[0077]

[0078] in, Indicates the measuring point In the power range and ambient temperature range Average temperature under the corresponding operating condition; This represents the standard deviation of temperature under the corresponding operating condition. This indicates the number of normal samples in the corresponding operating condition cell; Indicates the first Temperature values ​​of a normally operating sample.

[0079] In addition, the method also includes: rolling updates of the mean value of each measuring point in each working condition grid in the two-dimensional working condition hierarchical grid based on the standardized time series samples under the current latest normal operating conditions within a certain time range, and updating the temperature time series benchmark behavior baseline of each measuring point based on the updated data of each working condition grid.

[0080] Furthermore, seasonal drift correction is a crucial mechanism for ensuring the year-round effectiveness of the baseline model. Seasonal variations in ambient temperature cause systematic shifts in temperature at various measurement points. Without correction, this leads to increased false alarm rates in summer and increased false alarm rates in winter. This scheme incorporates an ambient temperature dimension into the operating condition stratification, which to some extent absorbs the direct impact of ambient temperature. In addition, every quarter or other preset time periods, the baseline mean μ of each operating condition grid is lightly updated using normal operating samples from the most recent period (e.g., one month) to correct residual seasonal drift and ensure that the confidence boundary maintains a stable false alarm rate level throughout the year.

[0081] By using a two-dimensional hierarchical model of power × ambient temperature, a finely adaptive temperature baseline behavior is achieved to the operating conditions. This fundamentally solves the problem of systematic false alarms caused by changes in operating conditions in the fixed threshold method, and provides an accurate statistical basis for the dynamic construction of subsequent confidence boundaries.

[0082] Step 104: Construct the time-series confidence statistical boundary of temperature for each measuring point based on the statistical distribution characteristics of the baseline output of the benchmark behavior;

[0083] Specifically, step 104 includes: based on the statistical distribution characteristics of the baseline output of the benchmark behavior, constructing the temperature time series confidence statistical boundary for each measuring point using one or more of the following methods: sliding time window statistics, quantile statistics, confidence interval estimation, or control chart.

[0084] Specifically, the upper confidence boundary in the temperature time series confidence statistical boundary is constructed based on the statistical distribution of normal samples within the operating condition grid. The upper boundary of temperature is determined by the mean plus κ times the standard deviation, where κ is the confidence factor parameter, which is set to 3.0 by default, corresponding to approximately 99.7% coverage of normal samples under a normal distribution.

[0085] Specifically, the dynamic confidence upper bound of the temperature at the measuring point can be expressed as:

[0086]

[0087] Correspondingly, the lower boundary of the dynamic confidence level for temperature can be expressed as:

[0088]

[0089] in, Indicates the measuring point At any moment The upper bound of the temperature confidence level; Indicates the measuring point At any moment The lower boundary of the temperature confidence level; Indicates the confidence factor parameter; This indicates the ambient temperature correction factor; Indicates the current ambient temperature; Indicates the reference ambient temperature.

[0090] For operating condition grids with sufficient sample size, quantile methods can be further employed as upper bounds to better accommodate non-normally distributed temperature data. Lower bounds are used to detect abnormally low temperatures, which may correspond to abnormal conditions such as sensor malfunction or excessive heat dissipation.

[0091] The sliding time window dynamic adjustment mechanism enables the temperature time-series confidence boundary to be continuously updated according to changes in real-time operating conditions. Based on the unit's power and ambient temperature values ​​at the current moment, the module automatically indexes the corresponding operating condition cell, calls the reference parameters of that cell to calculate the upper and lower confidence boundaries for the current moment, achieving real-time tracking of boundary values ​​with operating conditions. When the unit's operating conditions rapidly switch between adjacent cells, the module uses linear interpolation of the reference parameters of adjacent cells to smooth boundary jumps, avoiding instantaneous false alarms caused by discontinuous changes in boundary values.

[0092] The introduction of an ambient temperature correction term further enhances the adaptability of the temperature time-series confidence boundary to changes in environmental conditions. Based on the stratification of operating conditions, a correction term that varies linearly with ambient temperature is superimposed on the upper temperature boundary. This correction term compensates for the residual effects of subtle changes in ambient temperature on the temperature boundary when the granularity of the operating condition grid is insufficient, further reducing the false alarm rate under extreme weather conditions.

[0093] This solution constructs a dynamic confidence boundary that adapts to operating conditions, raising the benchmark for judging temperature anomalies from a static fixed threshold to a statistical boundary that adjusts in real time according to operating conditions. This significantly reduces the false alarm rate caused by fluctuations in operating conditions while maintaining a high sensitivity to detect real temperature anomalies.

[0094] Step 105: Calculate the health index based on the deviation characteristics of real-time standardized time-series samples relative to the confidence statistical boundary of the temperature time series;

[0095] Specifically, step 105 calculates the drivetrain health index based on the deviation characteristics of each measuring point relative to the temperature time-series confidence boundary and the coupling relationship between multiple measuring points, thereby achieving a quantitative assessment of the drivetrain health status. Specific deviation characteristics include one or more of the following: deviation amplitude, deviation duration, and multi-measuring-point coupling deviation characteristics.

[0096] Specifically, the extraction of single-point out-of-bounds characteristics is the foundation of health status calculation. For each temperature measuring point, the following four types of deviation characteristics are calculated within each assessment time window: number of out-of-bounds occurrences, i.e., the total number of sampling times in which the temperature value exceeds the upper confidence boundary within the current assessment window, reflecting the frequency of out-of-bounds events; continuous out-of-bounds duration, i.e., the maximum duration of a single continuous exceedance of the upper confidence boundary, reflecting the persistence of abnormal temperature rise; maximum out-of-bounds amplitude, i.e., the maximum amount by which the temperature exceeds the upper confidence boundary within the assessment window, reflecting the severity of temperature deviation from the normal range; and out-of-bounds recovery time, i.e., the time required for the temperature to fall back from the out-of-bounds state to within the confidence boundary, reflecting the rate of thermal anomaly resolution.

[0097] Extracting multi-measuring-point coupling deviation features is used to capture systematic anomaly patterns that cannot be identified by single-measuring-point analysis. During normal operation of the drivetrain, there is a stable correlation between the temperatures of each measuring point. This module calculates the time series of temperature differences between key measuring point pairs and compares it with the temperature difference distribution under historical normal operating conditions. When the temperature difference significantly deviates from the normal range, it is determined to be a multi-measuring-point coupling anomaly. The coupling anomaly feature can effectively distinguish between single sensor drift (only one measuring point exceeds the limit while related measuring points are normal) and actual component thermal state anomalies (multiple related measuring points exceed the limit in the same direction and the temperature difference relationship is abnormal), significantly improving the accuracy of fault location.

[0098] The fusion calculation of the health score transforms single-point out-of-bounds features and multi-point coupled features into a unified health quantification index in the range of 0 to 1. The fusion calculation adopts a weighted linear combination method, which normalizes each feature component and then sums them according to preset weights to obtain the comprehensive health score H (H=1 indicates complete health, H=0 indicates severe abnormality).

[0099] Specifically, the overall health score can be expressed as:

[0100]

[0101] Among them, each weight parameter satisfies:

[0102]

[0103] in, This indicates an overall health score; Indicates the out-of-bounds amplitude characteristics; Indicates the duration of consecutive boundary crossings; Indicates the frequency characteristics of boundary crossings; This indicates anomaly characteristics caused by multi-point coupling; These represent the weight parameters for the corresponding features.

[0104] The setting of the weight parameters follows the following principles: the weights of the continuous out-of-bounds duration and the maximum out-of-bounds amplitude are higher than the number of out-of-bounds times, reflecting a higher sensitivity to persistent and severe abnormalities; the weight of the multi-point coupling abnormality feature is higher than the single-point out-of-bounds feature of the measurement point, reflecting the priority attention to the systematic change of the thermal state.

[0105] Thus, through the comprehensive fusion scoring of the single-point out-of-bounds feature and the multi-point coupling feature, a comprehensive perception of the abnormal thermal state of the transmission chain from point to surface is achieved. While improving the sensitivity of early abnormality detection, the false alarm rate caused by single-point drift of the sensor is effectively reduced, providing a reliable basis for quantifying the health degree for the hierarchical output module.

[0106] Step 106: Output the health level of the wind turbine drive chain and the positioning result of the abnormal part according to the health index.

[0107] According to the health index, step 106 classifies the health state of the drive chain into four levels, for example. Combining the position of the out-of-bounds measurement point, a component-level health assessment result is generated, and the corresponding operation and maintenance suggestions are output.

[0108] Exemplarily, four-level classification can be adopted. The division of the four-level health levels follows the progressive principle from normal to severe, and each level corresponds to a clear operation and maintenance response strategy. The healthy level (H>0.85) indicates that the temperature time series of each measurement point of the drive chain is within the confidence boundary, without obvious abnormal deviation, and the unit can operate according to the normal plan; the attention level (0.70<H≤0.85) indicates that individual measurement points have slight out-of-bounds or short-term out-of-bounds, and the overall temperature trend has not shown continuous drift. It is recommended that the operation and maintenance personnel include it in daily attention and increase the data viewing frequency; the warning level (0.50<H≤0.70) indicates that one or more measurement points have continuous out-of-bounds or multi-points have coupling deviation, and the temperature abnormality has a certain persistence. It is recommended to arrange targeted on-site inspections in the near future (usually within 1 to 2 weeks) and prepare relevant spare parts in advance; the severe abnormality level (H≤0.50) indicates that multiple measurement points have significant continuous out-of-bounds and the health degree score continues to decline, and there is a relatively serious abnormal thermal state in the drive chain. It is recommended to arrange shutdown maintenance as soon as possible to avoid further expansion of component damage.

[0109] Component-level abnormal positioning automatically infers the fault location of the drive chain based on the distribution pattern of the out-of-bounds measurement points. The system internally sets up a drive chain measurement point-component mapping rule library: if only the temperature of the front bearing of the gearbox is out-of-bounds and other measurement points are normal, the system infers that the high-speed bearing of the gearbox is abnormal; if the lubricating oil temperature and the gearbox bearing temperature are out-of-bounds at the same time and the temperature difference increases, it is inferred that the heat dissipation capacity of the lubrication system decreases or the oil product deteriorates; if the main bearing temperature is continuously out-of-bounds and the rest of the measurement points are normal, it is inferred that the lubricating grease of the main bearing fails or there is internal damage; if the generator bearing temperature is out-of-bounds, it is inferred that the generator has misalignment or poor bearing lubrication. The component-level positioning result is output to the monitoring interface in the form of text description to assist the operation and maintenance personnel to quickly determine the key points of on-site inspection.

[0110] The health level de-jitter mechanism is used to suppress frequent switching of health levels caused by brief operational disturbances. The system requires that the health score continuously meet the corresponding level conditions within N consecutive evaluation windows (N=3 for attention level, N=2 for warning level, and N=1 for severe anomaly level) before performing a level switch and triggering the corresponding warning push, thus avoiding unnecessary alarms caused by occasional fluctuations in a single evaluation window.

[0111] A multi-level (e.g., four-level) classification system presents the health status of the drivetrain to maintenance personnel in a hierarchical and operable manner. Combined with component-level anomaly localization, it significantly improves the pertinence and timeliness of maintenance responses, providing a clear decision-making basis for the formulation of predictive maintenance plans.

[0112] Example 2

[0113] Embodiment 2 of this application also discloses a health grading assessment system for a transmission chain, such as... Figure 2 As shown, the system includes:

[0114] Data acquisition module 201 is used to synchronously acquire temperature time-series data and operating condition data of multiple preset measuring points on the transmission chain of the wind turbine.

[0115] Data preprocessing module 202 is used to preprocess temperature time series data to obtain standardized time series samples;

[0116] The benchmark modeling module 203 is used to perform layered modeling of operating conditions based on the operating condition data collected synchronously under different operating condition intervals and the standardized time series samples under preset normal operating conditions, and to establish the temperature time series benchmark behavior baseline for each measuring point.

[0117] The confidence boundary construction module 204 is used to construct the temperature time series confidence statistical boundary for each measuring point based on the statistical distribution characteristics of the baseline output of the benchmark behavior.

[0118] The health assessment module 205 is used to calculate health indicators based on the deviation characteristics of real-time standardized time-series samples relative to the confidence statistical boundary of temperature time series.

[0119] The graded output module 206 is used to output the health level of the wind turbine drive chain and the location results of abnormal parts according to the health index.

[0120] Furthermore, the baseline modeling module 203 is used for:

[0121] Using the active power in the operating data as the main hierarchical variable, the full power operating range of the wind turbine is divided into multiple power ranges at equal intervals.

[0122] Using ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature.

[0123] For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under the preset normal operating conditions collected synchronously with the determined working condition data are determined to establish the temperature time series reference behavior baseline for each measuring point.

[0124] Furthermore, the benchmark modeling module 203 is specifically used for: for each working condition cell in the two-dimensional working condition hierarchical grid, if the number of standardized time series samples is lower than the preset minimum threshold, then based on the standardized time series samples in the adjacent working condition cells, the working condition cells of the measuring point are smoothly supplemented, and based on the working condition dataset and standardized time series samples in each working condition cell after smooth supplementation, the temperature time series benchmark behavior baseline of each measuring point is established.

[0125] Furthermore, the baseline modeling module 203 is also used for:

[0126] Based on the standardized time series samples under the current preset normal operating conditions within a certain time range, the mean value of each measuring point in each working condition grid in the two-dimensional working condition layered grid is updated in a rolling manner, and the temperature time series benchmark behavior baseline of each measuring point is updated based on the updated data of each working condition grid.

[0127] The confidence boundary construction module 204 is used to: construct the temperature time series confidence statistical boundary for each measuring point based on the statistical distribution characteristics of the baseline output of the benchmark behavior, using one or more of the following methods: sliding time window statistics, quantile statistics, confidence interval estimation, or control chart.

[0128] Specifically, preprocessing includes one or more of the following: time alignment, missing value compensation, outlier removal, and standardization.

[0129] The measuring points include two or more of the following: front bearing temperature measuring point of gearbox, rear bearing temperature measuring point of gearbox, main bearing temperature measuring point, generator bearing temperature measuring point, and lubricating oil temperature measuring point.

[0130] Operating data includes one or more of the following: wind speed, ambient temperature, active power, speed, torque, pitch angle, and unit start-up / shutdown status.

[0131] Deviation characteristics include one or more of the following: deviation amplitude, deviation duration, and multi-point coupling deviation characteristics.

[0132] Example 3

[0133] Embodiment 3 of this application also discloses an electronic device, such as Figure 3 The following are included:

[0134] One or more processors 302;

[0135] Memory 303 is used to store one or more programs 304;

[0136] When one or more programs are executed by one or more processors 302, the one or more processors implement the transmission chain health grading assessment method of any one of Embodiment 1.

[0137] like Figure 3 As shown, the wind turbine drivetrain health grading assessment system 300 based on multi-point temperature time-series confidence statistical boundaries may include electronic equipment 301 and data acquisition equipment 305. Electronic equipment 301 may be an industrial control computer, edge computing server, industrial tablet, or cloud server, etc. Electronic equipment 301 may include, but is not limited to, a processor 302 and a memory 303. Those skilled in the art will understand that the illustration is merely an example of the assessment system 300 and does not constitute a limitation on the system. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, electronic equipment 301 and data acquisition equipment 305 may communicate via industrial Ethernet or the OPC-UA protocol.

[0138] Processor 302 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0139] The memory 303 can be an internal storage unit of the electronic device 301, such as a hard disk or memory; or it can be an external storage device of the electronic device 301, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, or a flash card. Furthermore, the memory 303 can include both internal and external storage units of the electronic device 301. The memory 303 is used to store the computer program 304 and data required for system operation, such as operating condition stratification baseline parameters, confidence boundary statistical parameters, health score history records, abnormal event logs, and evaluation reports.

[0140] The data acquisition device 305 is used to acquire real-time temperature time-series data and operating condition data from multiple measuring points in the wind turbine drivetrain. The data acquisition device 305 reads real-time sampled values ​​of each temperature measuring point, as well as operating parameters such as power, speed, and ambient temperature, from the turbine's SCADA system via a standard Modbus TCP / IP or OPC-UA interface. After completing timestamp marking and data format encapsulation, the data is transmitted to the electronic device 301 via the communication interface for use by the data preprocessing module and subsequent analysis modules.

[0141] Example 4

[0142] Embodiment 4 of this application also discloses a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the transmission chain health grading assessment method of any one of Embodiment 1.

[0143] This application proposes a method, system, electronic device, and storage medium for assessing the health status of a wind turbine drivetrain. The method includes: synchronously collecting temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine drivetrain; preprocessing the temperature time-series data to obtain standardized time-series samples; performing hierarchical modeling of operating conditions based on synchronously collected operating condition data under different operating condition intervals and standardized time-series samples under preset normal operating conditions to establish a baseline behavior for the temperature time-series at each measuring point; constructing a confidence statistical boundary for the temperature time-series at each measuring point based on the statistical distribution characteristics output from the baseline behavior; calculating a health index based on the deviation characteristics of the real-time standardized time-series samples relative to the confidence statistical boundary for the temperature time-series; and outputting the health level and abnormal location results of the wind turbine drivetrain based on the health index. This solution, through hierarchical modeling of operating conditions, dynamic confidence boundary construction, and health scoring of multiple temperature measuring points in the drivetrain, achieves adaptive operating condition assessment and multi-component hierarchical quantitative evaluation of the drivetrain's health status. It effectively solves the problems of high false alarm rate and insufficient positioning accuracy of single-measuring-point analysis in fixed-threshold methods, providing a systematic and engineering-feasible technical solution for predictive maintenance of wind turbine drivetrains.

[0144] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application.

[0145] Those skilled in the art will understand that the modules in the apparatus of the implementation scenario can be distributed within the apparatus of the implementation scenario as described, or they can be located in one or more apparatuses different from this implementation scenario, with corresponding changes. The modules of the above-described implementation scenario can be combined into one module, or they can be further divided into multiple sub-modules.

[0146] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of the implementation scenario.

[0147] The above disclosures are only a few specific implementation scenarios of this application. However, this application is not limited to these. Any variations that can be conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for grading and assessing the health of a transmission chain, characterized in that, The method includes: Simultaneously collect temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine's drive train; The temperature time series data is preprocessed to obtain standardized time series samples; Based on the operating condition data collected synchronously under different operating condition intervals and the standardized time series samples under preset normal operating conditions, operating condition hierarchical modeling is performed to establish the temperature time series benchmark behavior baseline for each measuring point. Based on the statistical distribution characteristics of the baseline output of the benchmark behavior, the temperature time-series confidence statistical boundary of each measurement point is constructed; The health index is calculated based on the deviation characteristics of the real-time standardized time-series samples relative to the confidence statistical boundary of the temperature time series. The health level of the wind turbine drive chain and the location of abnormal parts are output based on the health index.

2. The method as described in claim 1, characterized in that, The process involves performing layered modeling of operating conditions based on synchronously collected operating condition data from different operating condition intervals and standardized time-series samples under preset normal operating conditions, establishing a temperature time-series baseline for each measuring point, including: Using the active power in the operating data as the main hierarchical variable, the full power operating range of the wind turbine is divided into multiple power ranges at equal intervals. Using the ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature. For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under preset normal operating conditions collected synchronously with the determined working condition data are determined, and a temperature time series reference behavior baseline is established for each measuring point.

3. The method as described in claim 2, characterized in that, For each working condition cell in the two-dimensional working condition hierarchical grid, if the number of standardized time series samples is lower than a preset minimum threshold, the working condition cell of the measuring point is smoothly supplemented based on the standardized time series samples in the adjacent working condition cells. Based on the working condition dataset and standardized time series samples in each working condition cell after smoothing and supplementation, a temperature time series reference behavior baseline for each measuring point is established.

4. The method as described in claim 2 or 3, characterized in that, Also includes: Based on the standardized time series samples under the preset normal operating conditions within the latest time range, the mean value of each measurement point in each working condition grid in the two-dimensional working condition hierarchical grid is updated in a rolling manner, and the temperature time series benchmark behavior baseline of each measurement point is updated based on the updated data of each working condition grid.

5. The method as described in claim 1, characterized in that, The step of constructing the temperature time-series confidence statistical boundary for each measuring point based on the statistical distribution characteristics of the baseline output of the benchmark behavior includes: constructing the temperature time-series confidence statistical boundary for each measuring point using one or more of the following methods: sliding time window statistics, quantile statistics, confidence interval estimation, or control chart, based on the statistical distribution characteristics of the baseline output of the benchmark behavior.

6. The method as described in claim 1, characterized in that, The preprocessing includes one or more of the following: time alignment, missing value compensation, outlier removal, and standardization. The measuring points include two or more of the following: the front bearing temperature measuring point of the gearbox, the rear bearing temperature measuring point of the gearbox, the main bearing temperature measuring point, the generator bearing temperature measuring point, and the lubricating oil temperature measuring point. The operating data includes one or more of the following: wind speed, ambient temperature, active power, speed, torque, pitch angle, and unit start-up / shutdown status. The deviation features include one or more of the following: deviation amplitude, deviation duration, and multi-point coupled deviation features.

7. A health grading and assessment system for a transmission chain, characterized in that, The system includes: The data acquisition module is used to synchronously acquire temperature time-series data and operating condition data from multiple preset measuring points on the wind turbine's drive train. The data preprocessing module is used to preprocess the temperature time series data to obtain standardized time series samples; The benchmark modeling module is used to perform layered modeling of operating conditions based on the operating condition data collected synchronously under different operating condition intervals and the standardized time series samples under preset normal operating conditions, and to establish the temperature time series benchmark behavior baseline for each measuring point. The confidence boundary construction module is used to construct the temperature time-series confidence statistical boundary for each measurement point based on the statistical distribution characteristics of the baseline output of the reference behavior. The health assessment module is used to calculate a health index based on the deviation characteristics of the real-time standardized time-series samples relative to the temperature time-series confidence statistical boundary. The graded output module is used to output the health level of the wind turbine drive chain and the location results of abnormal parts based on the health index.

8. The system as described in claim 7, characterized in that, The benchmark modeling module is used for: Using the active power in the operating data as the main hierarchical variable, the full power operating range of the wind turbine is divided into multiple power ranges at equal intervals. Using the ambient temperature in the operating condition data as an auxiliary hierarchical variable, for each power range, multiple temperature ranges are divided according to the ambient temperature from high to low or from low to high, so as to form a two-dimensional operating condition hierarchical grid of power and ambient temperature. For each working condition cell in the two-dimensional working condition layered grid, the working condition data for the working condition cell and the standardized time series samples under preset normal operating conditions collected synchronously with the determined working condition data are determined, and a temperature time series reference behavior baseline is established for each measuring point.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the health grading assessment method for the drive chain according to any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for grading and evaluating the health of the transmission chain according to any one of claims 1-6.

Images