Method and apparatus for inspecting wind turbine blade, and device and storage medium thereof

A server for inspecting a wind turbine blade, wherein the wind turbine blade is a blade in a wind power generation device, the wind power generation device further comprising a tower provided with a sound acquisition device that is mounted away from the wind turbine blade;

• and
• the server comprises:
• a sound acquirer, configured to acquire, via the sound acquisition device, a sound signal generated by an impingement of wind on the wind turbine blade, wherein the sound signal comprises a sound signal generated by sliding of air between blades in the case that the wind impinges on the wind turbine blade;
• a frequency spectrum generator, configured to:
• extract a signal envelope from a time domain signal diagram formed by the sound signal by calling a signal analysis algorithm;
• determine a position of a point, where a wave trough appears on the signal envelope, in a time domain as a segmentation point;
• and
• convert the sound signal into a frequency spectrogram, and obtain a segmented frequency spectrogram by segmenting the frequency spectrogram based on the segmentation point;
• and
• a damage recognizer, configured to obtain a damage recognition result of the wind turbine blade by performing image recognition on the segmented frequency spectrogram based on a damage recognition model, wherein the damage recognition model is obtained by training a neural network model.

A wind power generation device, comprising:

• a sound acquisition device;
• a memory communicably connected to the sound acquisition device;
• and
• a processor communicably connected to the memory;
• wherein
• the sound acquisition device is configured to acquire a sound signal in response to an impingement of wind on a wind turbine blade of the wind power generation device, and store the sound signal in the memory;
• the memory is configured to store an executable instruction and the sound signal;
• and
• the processor is configured to load and execute the executable instruction stored in the memory to perform a method for inspecting the wind turbine blade, wherein the wind turbine blade is a blade in the wind power generation device, the wind power generation device further comprising a tower provided with the sound acquisition device that is mounted away from the wind turbine blade, the method comprising:
• acquiring, via the sound acquisition device, a sound signal in response to an impingement of wind on the wind turbine blade, wherein the sound signal comprises a sound signal generated by sliding of air between blades in the case that the wind impinges on the wind turbine blade;
• extracting a signal envelope from a time domain signal diagram formed by the sound signal by calling a signal analysis algorithm;
• determining a position of a point, where a wave trough appears on the signal envelope, in a time domain as a segmentation point;
• and
• converting the sound signal into a frequency spectrogram, and obtaining a segmented frequency spectrogram by segmenting the frequency spectrogram based on the segmentation point;
• and
• obtaining a damage recognition result of the wind turbine blade by performing image recognition on the segmented frequency spectrogram based on a damage recognition model, wherein the damage recognition model is obtained by training a neural network model.

A non-transitory computer-readable storage medium storing at least one instruction, at least one program, a code set, or an instruction set therein, wherein the at least one instruction, the at least one program, the code set, or the instruction set, when loaded and executed by a processor, causes the processor to perform a method for inspecting a wind turbine blade, wherein the wind turbine blade is a blade in a wind power generation device, the wind power generation device further comprising a tower provided with a sound acquisition device that is mounted away from the wind turbine blade, the method comprising:

• acquiring, via the sound acquisition device, a sound signal in response to an impingement of wind on the wind turbine blade, wherein the sound signal comprises a sound signal generated by sliding of air between blades in the case that the wind impinges on the wind turbine blade;
• extracting a signal envelope from a time domain signal diagram formed by the sound signal by calling a signal analysis algorithm;
• determining a position of a point, where a wave trough appears on the signal envelope, in a time domain as a segmentation point;
• and
• converting the sound signal into a frequency spectrogram, and obtaining a segmented frequency spectrogram by segmenting the frequency spectrogram based on the segmentation point;
• and
• obtaining a damage recognition result of the wind turbine blade by performing image recognition on the segmented frequency spectrogram based on a damage recognition model, wherein the damage recognition model is obtained by training a neural network model.