A method of intelligent perception of indoor occupants based on artificial neural network

Abstract

The invention discloses an intelligent indoor personnel perceiving method based on the artificial neural network. By monitoring and analyzing the attenuation and fluctuation influence on radio-frequency signal RSSI by personnel, namely comparing and analyzing changes of the radio-frequency signal RSSI before and after the personnel enter the area to be tested, the perceiving aim is achieved. Two indexes are calculated for the RSSI collected in real time and represent the RSSI deviation degree of the static condition after the personnel enter the area to be tested and the condition when no personnel enters the area and the RSSI deviation degree of the motion condition after the personnel enter the area to be tested and the condition when no personnel enters the area respectively; then the static or moving scenes of the personnel inside and outside the area to be tested are simulated largely, and the two indexes are calculated and stored in real time and serve as samples; the neural network is trained through the sample data so as to build a nonlinear mapping relation between the index data and the fact whether the personnel enter the area or not; after the training is finished, whether the personnel enter the area to be tested or not is automatically judged through the neural network.

Description

Technical field [0001] The invention relates to an intelligent perception method for indoor personnel based on an artificial neural network, and belongs to the technical field of indoor accurate perception. Background technique [0002] Nowadays, people’s demand for accurate perception of targets continues to grow, especially in certain places with high security alert levels, such as museums, archives, computer rooms, bank vaults, prisons, and dangerous goods depositories. It is often necessary to monitor whether there are illegal personnel in real time. Approaching, entering and exiting; or applied in "smart home" and "smart building", dynamically adjust the lighting, ventilation and room temperature of the building according to the presence and movement of people to achieve the purpose of saving energy. In short, it is necessary to accurately sense whether there are people in a certain area. [0003] Traditional solutions for this include video surveillance, infrared alarm and a...

AI Technical Summary

Problems solved by technology

[0003] Traditional solutions for this include video surveillance, infrared alarm and anti-theft devices, etc., but they all have limitations: they are sensitive to changes in environmental factors, and light, temperature, smoke, and dust can all affect their perception accuracy; The installation location has strict requirements, and there must be no dead corners or blind areas; the equipment is exposed and easily damaged, etc.

On the basis of this principle, researchers have proposed specific algorithms to realize indoor occupancy perception. The existing algorithms mostly have the following two shortcomings: (1) Determine whether there are people in the area to be tested by threshold comparison, and the threshold is determined by a certain formula Calculated or artificially specified based on experimental observation results, the former cannot cope with complex and changeable specific application environments, because different environments cannot be described uniformly by a formula, and the latter requires manual participation, which is likely to cause deviations

(2) Only by capturing the change of RSSI caused by the movement of people, it is easy to cause perception failure when people enter the area to be tested and stay still

Images