Traditional methods for studying the whole-night sleep process often require subjects to wear multiple devices to capture electroencephalogram(EEG)or electrocardiogram(ECG)signals

which can interfere with natural sleep and are inconvenient for home sleep health monitoring.To address this issue

a non-intrusive sleep signal acquisition device based on thin-film piezoelectric sensors is designed to obtain the signals necessary for sleep staging.Then

parameters such as body movement intensity

respiratory variability(RV)

and heart rate variability(HRV)in both time and frequency domains are extracted

and preliminary feature parameters are selected.A back propagation(BP)neural network is utilized to achieve four-stage sleep staging.To improve the recognition accuracy of the four-stage sleep structure

the most significant feature parameters affecting sleep staging are further screened through experiments

and the grey wolf optimization(GWO)algorithm is applied to optimize the BP neural network.Experimental results show that the GWOBP neural network

based on five feature parameters

can achieve a sleep staging recognition accuracy of 76.82% in healthy subjects.The combination of thin-film piezoelectric sensors with sleep staging algorithms simplifies the monitoring process and enhances the convenience of home sleep monitoring.