CN-121434789-B - Wind power tower barrel load measuring method

CN121434789BCN 121434789 BCN121434789 BCN 121434789BCN-121434789-B

Abstract

The invention provides a wind power tower load measurement method, which comprises the steps of dividing a preset time period into N time intervals, collecting training temperature measurement data of at least one temperature sensor and training center wavelength data of at least one load sensor in each time interval, training to obtain N load temperature compensation models based on the training temperature measurement data of the at least one temperature sensor and the training center wavelength data of the at least one load sensor collected in each time interval, collecting small wind stop temperature measurement data of the at least one temperature sensor and small wind stop center wavelength data of the at least one load sensor under small wind stop of a wind power tower, selecting a target load temperature compensation model from the N load temperature compensation models based on the small wind stop temperature measurement data, obtaining a center wavelength predicted value through the target load temperature compensation model, and calculating the wind power tower load based on the small wind stop center wavelength data and the center wavelength predicted value.

Inventors

WEI JUN
LEI JUNPING
LING JINGFANG
GU YILONG

Assignees

上海拜安传感技术有限公司

Dates

Publication Date: 20260512
Application Date: 20251226

Claims (9)

1. A wind power tower load measuring method is characterized in that: Dividing a preset time period into N time intervals, and collecting training temperature measurement data of at least one temperature sensor and training center wavelength data of at least one load sensor in each time interval, wherein the at least one temperature sensor and the at least one load sensor are matched and arranged on a wind power tower, and N is an integer greater than or equal to 1; training to obtain a plurality of load temperature compensation models based on time sequence modeling based on training temperature measurement data of at least one temperature sensor and training center wavelength data of at least one load sensor, which are acquired at each time interval, and obtaining a median time-varying curve of the training temperature measurement data of the temperature sensor in a plurality of sampling periods in each time interval; collecting small wind shutdown temperature measurement data of at least one temperature sensor and small wind shutdown center wavelength data of at least one load sensor under small wind shutdown of a wind power tower; Based on the small wind shut-down temperature measurement data, comparing the deviation degree of the small wind shut-down temperature measurement data at each sampling moment with the temperature median at the corresponding moment of each median time change curve one by one, taking the mean value of the deviation degree as similarity, selecting a target load temperature compensation model from the load temperature compensation models, and obtaining a central wavelength predicted value through the target load temperature compensation model; And calculating the load of the wind power tower barrel based on the small wind power shutdown central wavelength data and the central wavelength predicted value.
2. The load measurement method according to claim 1, wherein, The temperature sensors and the load sensors are matched in pairs, and each pair of temperature sensors and the load sensors are distributed at 90-degree intervals along the circumferential direction of the wind power tower.
3. The load measurement method according to claim 2, wherein, The training temperature measurement data of the at least one temperature sensor and the training center wavelength data of the at least one load sensor acquired based on each time interval are trained to obtain a plurality of load temperature compensation models, and the training temperature compensation models comprise: training temperature measurement data of a temperature sensor in a pair of temperature sensors and a load sensor matched in pairs is taken as model input, and training center wavelength data of the corresponding load sensor is taken as output to obtain the load temperature compensation model.
4. A load measuring method according to claim 3, wherein, The selecting a target load temperature compensation model from the plurality of load temperature compensation models based on the small wind shut-down temperature measurement data comprises: Obtaining a median time-dependent curve of training temperature measurement data of the temperature sensor in a plurality of sampling periods in each time interval for N time intervals; and comparing the similarity of the small wind shut-down temperature measurement data and the plurality of median time-varying curves, and taking a load temperature compensation model corresponding to the median time-varying curve with the highest similarity as a target load temperature compensation model.
5. The load measurement method according to claim 1, wherein, And inputting the small wind stop temperature measurement data of the temperature sensor acquired under the small wind stop of the wind power tower into the target load temperature compensation model to obtain the central wavelength predicted value.
6. The load measurement method according to claim 5, wherein, Based on the small wind shutdown center wavelength data of the load sensor collected under the small wind shutdown of the wind power tower and the center wavelength forecast value, the center wavelength variation of the load sensor is obtained; and obtaining a load measurement result based on the central wavelength variation.
7. The load measurement method according to claim 6, wherein, Calculating the load of the wind power tower by using the central wavelength variation of a pair of orthogonal load sensors, comprising: based on the load sensors at the 0 degree and 90 degree positions, a first pair of orthogonal load sensors calculate the load as follows: , Wherein, the And The coefficients corresponding to the load sensors at the 0 degree and 90 degree positions respectively, And The center wavelength variation amounts of the load sensor after temperature compensation at the 0 degree and 90 degree positions respectively, And The loads of the wind power tower barrel after temperature compensation at the positions of 0 degrees and 90 degrees are respectively, Is the first resultant load; and calculating the load based on the load sensors at the 180 degree and 270 degree positions and a second pair of orthogonal load sensors as follows: , , Wherein, the And The coefficients corresponding to the load sensors at the 180 degree and 270 degree positions respectively, And The center wavelength variation of the load sensor after temperature compensation at 180 degrees and 270 degrees respectively, And The loads of the wind power tower after temperature compensation at the positions of 180 degrees and 270 degrees respectively, Is the second combined load.
8. The load measurement method according to claim 7, wherein, And calculating the load of the wind power tower based on the measured data before temperature compensation, and comparing the load with the load after temperature compensation to evaluate the accuracy of the load measurement after temperature compensation.
9. The load measurement method according to claim 1, wherein, And fusing the small wind stop temperature measurement data of the at least one temperature sensor and the small wind stop center wavelength data of the at least one load sensor acquired under the small wind stop of the wind power tower to the training temperature measurement data of the temperature sensor and the training center wavelength data of the load sensor in a pair of temperature sensors and the load sensor matched in pairs, and retraining the load temperature compensation model.

Description

Wind power tower barrel load measuring method Technical Field The invention relates to the technical field of measurement, in particular to a wind power tower load measurement method. Background Wind power tower load monitoring is a key technology for guaranteeing safe and stable operation of a wind generating set, and temperature drift is one of main factors affecting measurement accuracy of a load sensor. The current wind power tower load monitoring faces the technical challenges of (1) the limitation of the traditional temperature compensation method that wind power equipment is often deployed in areas with large temperature difference and severe environments, such as offshore or high altitude areas. The measurement drift of the load sensor, which is affected by temperature, has time-varying and nonlinear characteristics, and linear compensation or polynomial fitting methods, such as fixed-coefficient temperature compensation algorithms, are mostly adopted in the related art, but these methods have difficulty in coping with nonlinear temperature characteristics of the sensor under complex working conditions. (2) The existing intelligent compensation technology has the defects that although some researches try to apply a machine learning method to carry out temperature compensation, the problem that the model generalization capability is weak and the long-term dependency is not captured enough generally exists. Especially for the application scene with strong time sequence characteristic of wind power load monitoring, the time dynamic characteristic of temperature influence is difficult to effectively learn by the conventional neural network. In view of the above, there is a need in the wind power industry for a temperature compensation method that can adapt to different environmental conditions and accurately capture the temperature-load coupling relationship. Disclosure of Invention In order to overcome at least one of the problems in the related art, the invention provides a wind power tower load measuring method. Comprising the following steps: Dividing a preset time period into N time intervals, and collecting training temperature measurement data of at least one temperature sensor and training center wavelength data of at least one load sensor in each time interval, wherein the at least one temperature sensor and the at least one load sensor are matched and arranged on a wind power tower, and N is an integer greater than or equal to 1; training to obtain a plurality of load temperature compensation models based on training temperature measurement data of the at least one temperature sensor and training center wavelength data of the at least one load sensor, wherein the training temperature measurement data are acquired at each time interval; collecting small wind shutdown temperature measurement data of at least one temperature sensor and small wind shutdown center wavelength data of at least one load sensor under small wind shutdown of a wind power tower; Selecting a target load temperature compensation model from the plurality of load temperature compensation models based on the small wind shutdown temperature measurement data, and obtaining a central wavelength predicted value through the target load temperature compensation model; And calculating the load of the wind power tower barrel based on the small wind power shutdown central wavelength data and the central wavelength predicted value. In alternative embodiments, the temperature sensors and the load sensors are paired together, and each pair of temperature sensors and load sensors is spaced 90 degrees apart along the circumference of the wind turbine tower. In some optional embodiments, the training of the training temperature measurement data of the at least one temperature sensor and the training center wavelength data of the at least one load sensor based on each time interval to obtain a plurality of load temperature compensation models includes: training temperature measurement data of a temperature sensor in a pair of temperature sensors and a load sensor matched in pairs is taken as model input, and training center wavelength data of the corresponding load sensor is taken as output to obtain the load temperature compensation model. In some alternative embodiments, the selecting a target load temperature compensation model from the plurality of load temperature compensation models based on the small wind shut-down temperature measurement data includes: Obtaining a median time-dependent curve of training temperature measurement data of the temperature sensor in a plurality of sampling periods in each time interval for N time intervals; and comparing the similarity of the small wind shut-down temperature measurement data and the plurality of median time-varying curves, and taking a load temperature compensation model corresponding to the median time-varying curve with the highest similarity as a target load temperature compensation mode