Non-contact liquid component identification model training method, identification method, system and device

The invention provides a non-contact liquid component identification model training method, an identification method, a system and a device, and the method comprises the steps: transmitting a sweep frequency signal with the frequency from low to high to liquid in a set distance range through a millimeter wave radar in an identification process; dividing the obtained intermediate frequency signal into a plurality of frequency bands with different starting frequencies and ending frequencies, and respectively carrying out fast Fourier transform to obtain the reflection characteristics of the liquid target to the millimeter wave signals of the plurality of frequency bands in the set range. The multi-band reflection features are processed through the customized neural network, the reflection features related to the to-be-detected liquid are extracted from the multi-band reflection features, and the liquid components are classified and identified, so that the identification precision of the liquid components can be