Data-driven fan blade icing prediction method and device
A fan blade, data-driven technology, applied in forecasting, engines, wind turbines, etc., can solve the problems of complex mechanism modeling, poor generalization ability and actual forecasting effect, etc., and achieve the effect of improving the AUC value.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0078] This embodiment introduces a data-driven fan blade icing prediction method and device, including:
[0079] Obtain the SCADA data collected in advance by the fan, eliminate outliers, fill in missing values, and complete the preprocessing of the data set;
[0080] According to the preprocessed SCADA data set, a method of combining down-sampling based on data distribution similarity analysis and data adaptive comprehensive over-sampling is used to balance the distribution of wind turbine blade icing and non-icing data in the SCADA data set ;
[0081] According to the SCADA data set after the balanced distribution, the importance assessment of the high-dimensional feature data is carried out through the random forest algorithm, and the feature data is obtained after screening and reconstruction;
[0082] The feature data after the screening and reconstruction are trained using the hidden layer of the long short-term memory network, and the feature vector after the output t...
Embodiment 2
[0145] This embodiment provides a data-driven fan blade icing prediction device, including:
[0146] The preprocessing unit is used to obtain the SCADA data collected in advance by the fan, eliminate abnormal values, fill in missing values, and complete the preprocessing of the data set;
[0147] The processing unit is used to balance the icing and non-icing of wind turbine blades in the SCADA data set based on the combination of down-sampling based on data distribution similarity analysis and data adaptive comprehensive over-sampling according to the pre-processed SCADA data set. Distribution of icing data;
[0148] The evaluation unit is used to evaluate the importance of high-dimensional feature data through the random forest algorithm according to the SCADA data set after the balanced distribution, and obtain the feature data after screening and reconstruction;
[0149] The training unit is used to train the feature data after the screening and reconstruction using the hi...
Embodiment 3
[0154] This embodiment provides a data-driven fan blade icing prediction device, including a processor and a storage medium;
[0155] The storage medium is used to store instructions;
[0156] The processor is configured to operate according to the instructions to execute the steps of the method according to any one of Embodiment 1.
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


