| abstract |
The invention discloses a sleep-onset detection system in a light environment. The system includes a light color recognition unit, a sleep-onset recognition unit, a dimmable light group and a control unit; The duration of falling asleep is used as the input, and the user's eye opening, eye-closing duration, heart rate, body movement frequency, and body temperature change rate are used as the output to establish a dynamic recursive Elman neural network to characterize ambient lighting conditions. The mapping relationship between the user’s sleep efficiency and the user’s sleep efficiency; the neural network is trained offline based on samples collected in various light environments, and the trained network is used to online predict the relevant parameters of the user’s sleep efficiency under on-site lighting conditions. The invention introduces the time length from the turning point of falling asleep into the input of the neural network, so the physical parameters at a certain time point after falling asleep can be predicted, and a basis for searching and recommending a light environment with potential high falling asleep efficiency is provided. |