Fan blade state monitoring method and system

Abstract

The invention relates to a fan blade state monitoring method. The method comprises the following steps of S1, constructing a training data set and a test data set of fan blade images; S2, constructinga convolutional neural network; S3, training a convolutional neural network by using the training data set to obtain a trained convolutional neural network; S4, the trained neural network being tested by using the test data set, if conditions are met, the step S5 being executed, and otherwise, the step S3 being executed; S5, compressing the trained convolutional neural network by using an alternating direction multiplier method to obtain a compressed convolutional neural network; and S6, acquiring a to-be-tested fan blade image, and inputting the to-be-tested fan blade image into the compressed convolutional neural network to obtain a fan blade state. Compared with the prior art, the method has important significance in safe and stable operation of the wind turbine generator and the fieldof equipment maintenance in the wind power field.

Description

technical field [0001] The invention relates to the field of fan blade fault diagnosis, in particular to a fan blade state monitoring method and system. Background technique [0002] Fan blades are one of the core components of wind turbines, accounting for about 15%-20% of the total cost of the fan. The condition of the fan blades will directly affect the performance and benefits of the fan. [0003] At present, all the research on fan blade condition monitoring is mainly focused on the processing of sensor signals. By installing different sensors on fan blades to obtain corresponding signals (including vibration signals, tension signals, acoustic emission signals, ultrasonic signals, etc.), and then Through the corresponding signal processing to achieve the purpose of fault diagnosis. In addition, in normal operating wind farms, most of the on-site direct visual inspections are carried out through telescopes and "Spiderman". [0004] First of all, the signal acquisition ...

AI Technical Summary

Problems solved by technology

[0004] First of all, the signal acquisition of the sensor is unstable, and it is easily affected by the surrounding environment. Since the material of the fan blade is a high-damping material, the propagation of the signal is easily affected. In order to ensure the accuracy of the diagnosis, various and multiple A sensor needs to be installed on the fan blade, which brings about the problem of sensor installation and blade manufacturing cost; secondly, during the operation time of the wind turbine for 20 to 30 years, a large amount of signal data obtained (sensor’s high sampling frequency) processing and storage is also a problem

There is a problem of work efficiency in manual visual inspection, which will cause the unit to shut down for a long time and affect the economic benefits of the wind farm

Images