Search

CN-122022157-A - Reliability rating method, system, medium and computer equipment for wind turbine generator

CN122022157ACN 122022157 ACN122022157 ACN 122022157ACN-122022157-A

Abstract

The invention discloses a reliability rating method, a system, a medium and computer equipment of a wind turbine, wherein the method comprises the steps of preparing and aligning data, modeling ARIMA/seasonal ARIMA and obtaining characteristic indexes, structuring ARIMA output into enhancement characteristics available for a tree model and fusing the enhancement characteristics with time coding, hysteresis and rolling statistics and static parameters of the wind turbine, and adopting CART or gradient to promote tree output reliability grade/risk score and provide interpretable rule paths, online updating and auditing to cope with concept drift and seasonal variation. The reliability rating method is stable, interpretable and auditable, the accuracy, recall rate and cross-season stability are remarkably improved, and low-cost operation can be realized through lightweight feature calculation and periodic update in an edge deployment environment.

Inventors

  • DU JIAN

Assignees

  • 华电郑州机械设计研究院有限公司

Dates

Publication Date
20260512
Application Date
20260128

Claims (10)

  1. 1. The reliability rating method for the wind turbine generator is characterized by comprising the following steps of: s1, data preparation and alignment, namely collecting SCADA data and static metadata of a fan, and completing time alignment, deletion filling and standardization processing; S2, modeling a time sequence, namely constructing an ARIMA/seasonal ARIMA model aiming at a unit or a group based on the preprocessed data, and obtaining an ARIMA characteristic index; S3, feature transformation and fusion, namely structuring ARIMA feature indexes into enhancement features available for the decision tree model, and fusing the enhancement features with time coding, hysteresis and rolling statistics and static parameters to generate an enhancement feature set; S4, grading and interpretation, namely training a CART model based on the enhanced feature set, outputting a reliability grade or risk score, and providing rule paths and feature contribution interpretation; and S5, online updating and auditing, namely monitoring concept drift and seasonal variation, periodically updating an ARIMA/seasonal ARIMA model, and generating an audit report and an evidence chain record.
  2. 2. The wind turbine reliability rating method according to claim 1, wherein in S2, the ARIMA characteristic index at least includes a predicted value (T), residual e (t), confidence interval width, information criterion AIC/BIC, seasonal intensity, and (p, d, q) parameter combinations, where p, d, and q are core parameters of the ARIMA/seasonal ARIMA model, respectively, representing autoregressive terms, differential times, and moving average terms, respectively.
  3. 3. The method for rating reliability of wind turbine generator according to claim 1, wherein in S3, the enhancement feature set, time and statistics enhancement includes at least Hour/Weekday/montan/Season/holiday code, mean/std/min/max/skew/kurt Lag and rolling statistics, and min/hr/day cross-scale aggregation.
  4. 4. The wind turbine reliability rating method according to claim 1, wherein in S4, the CART model is divided based on the genie unrepeace or the information gain, the path rule and the characteristic contribution are output, and the interval mapping and the stability score are provided, and meanwhile, the reliability rating is dynamically adjusted based on the combination of the residual distribution and the confidence interval anomaly rate of the ARIMA/seasonal ARIMA model, the cross-season stability score and the maintenance policy threshold.
  5. 5. The method for rating reliability of wind turbines according to claim 1, wherein said drift monitoring characteristics in S5 include at least a variable point count, a half-life, a trend slope, a rolling fluctuation index, and an anomaly density.
  6. 6. A wind turbine reliability rating system adapted for use in a method of rating reliability of a wind turbine as claimed in any of claims 1-5, comprising: the data acquisition and preprocessing module is used for acquiring SCADA data and static metadata and completing time alignment, deletion filling and standardization processing; The time sequence modeling module is linked to the data acquisition and preprocessing module and is used for constructing an ARIMA/seasonal ARIMA model and outputting ARIMA characteristic indexes; the feature engineering module is connected to the time sequence modeling module and is used for converting ARIMA feature indexes into enhancement features available for the decision tree model and fusing the enhancement features with time coding, hysteresis and rolling statistics and static parameters; the grading and interpretation module is connected to the characteristic engineering module and is used for training and reasoning the CART model, outputting reliability grade or risk grade and providing rule path and characteristic contribution interpretation; And the audit and consistency check module is connected to the evaluation and interpretation module and is used for generating an audit report and an evidence chain record and establishing a consistency check rule to carry out consistency check.
  7. 7. The wind turbine reliability rating system of claim 6, wherein the audit and consistency verification module records evidence source metadata and points to a reference link field of a comparison matrix file comparison _matrix.csv, and the established consistency verification rule performs synchronous verification in terms of entry existence, link validity, and threshold map integrity.
  8. 8. The wind turbine reliability rating system of claim 6, further comprising a visualization module for visually displaying evidence chain records and rule paths and visually outputting maintenance level suggestions and corresponding risk descriptions.
  9. 9. A computer readable storage medium, characterized in that a computer program is stored, which, when being executed by a processor, causes the processor to perform the steps of the wind turbine reliability rating method according to any of claims 1-6.
  10. 10. A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the wind turbine reliability rating method of any of claims 1-6.

Description

Reliability rating method, system, medium and computer equipment for wind turbine generator Technical Field The invention relates to the technical field of cross application of wind power generation, smart grid, predictive maintenance and machine learning, in particular to a wind turbine reliability rating method, a system, a medium and computer equipment based on CART decision tree and ARIMA time sequence prediction integration algorithm. Background The reliability evaluation and predictive maintenance of the wind turbine generator are required to simultaneously characterize a short-term dynamic trend and a long-term steady-state characteristic. Traditional methods based on static thresholds or single statistical indexes are susceptible to seasonal, working condition changes and conceptual drift, resulting in increased rating fluctuations and false positives. CART (Classification and Regression Tree, classification regression tree) has good interpretability and regular expression capability, but is easy to lose time structure information when used alone under the scene of strong time sequence dependence, seasonal term and residual variance. ARIMA (seasonal ARIMA-containing) (AutoRegressive Integrated Moving Average, autoregressive integral moving average) is good at modeling trends, seasonality, and residuals of time series, but direct output is difficult to translate into interpretable classification rules. Disclosure of Invention The method, the system, the medium and the computer equipment for grading the reliability of the wind turbine generator set, which are stable, interpretable and auditable, can at least solve one of the technical problems. In order to solve the technical problems, the invention adopts the following technical scheme: A reliability rating method of a wind turbine generator comprises the following steps: s1, data preparation and alignment, namely collecting SCADA data and static metadata of a fan, and completing time alignment, deletion filling and standardization processing; S2, modeling a time sequence, namely constructing an ARIMA/seasonal ARIMA model aiming at a unit or a group based on the preprocessed data, and obtaining an ARIMA characteristic index; S3, feature transformation and fusion, namely structuring ARIMA feature indexes into enhancement features available for the decision tree model, and fusing the enhancement features with time coding, hysteresis and rolling statistics and static parameters to generate an enhancement feature set; S4, grading and interpretation, namely training a CART model based on the enhanced feature set, outputting a reliability grade or risk score, and providing rule paths and feature contribution interpretation; and S5, online updating and auditing, namely monitoring concept drift and seasonal variation, periodically updating an ARIMA/seasonal ARIMA model, and generating an audit report and an evidence chain record. Further, in the step S2, the ARIMA characteristic index at least includes a predicted value(T), residual e (t), confidence interval width, information criterion AIC/BIC, seasonal intensity, and (p, d, q) parameter combinations, where p, d, and q are core parameters of the ARIMA/seasonal ARIMA model, respectively, representing autoregressive terms, differential times, and moving average terms, respectively. Further, in the step S3, the enhancement feature set, the time and statistics enhancement at least includes Hour/Weekday/montan/Season/holiday code, mean/std/min/max/skew/kurt Lag and rolling statistics, and min/hr/day cross-scale aggregation. Further, in S4, the CART model performs segmentation based on the base purity or the information gain, outputs the path rule and the feature contribution, provides the interval mapping and the stability score, and performs dynamic adjustment of the reliability level based on the combination of the residual distribution and the confidence interval anomaly rate of the ARIMA/seasonal ARIMA model, the cross-season stability score and the maintenance policy threshold. Further, in S5, the drift monitoring feature includes at least a variability count, a half-life, a trend slope, a rolling fluctuation index, and an abnormal density. A reliability rating system of a wind turbine generator is applicable to a reliability rating method of the wind turbine generator, and comprises the following steps: the data acquisition and preprocessing module is used for acquiring SCADA data and static metadata and completing time alignment, deletion filling and standardization processing; The time sequence modeling module is linked to the data acquisition and preprocessing module and is used for constructing an ARIMA/seasonal ARIMA model and outputting ARIMA characteristic indexes; the feature engineering module is connected to the time sequence modeling module and is used for converting ARIMA feature indexes into enhancement features available for the decision tree model and fusing the enhancement features with ti