In this paper, electrocardiography, electroencephalography, oximetry, picoelectricity and body temperature will be selected as the analyzed signals, and the corresponding multimodal acquisition scheme and the overall design of the acquisition system will be proposed. Aiming at the problem that physiological signals are susceptible to external interference, the wavelet transform method is applied within the acquisition system to achieve the purpose of noise reduction processing of the acquired physiological signals. Comparing with the MP150 system to collect resting state data, the signals collected by this system and the MP150 system are basically similar in waveform amplitude, phase and spectral characteristics, which verifies the validity of the physiological signals collected by this system. In order to facilitate early diagnosis of psychological crises expressed in the form of the social behavior of adolescents through the Internet, an improved PSO algorithm is proposed in this study to fine-tune the parameter setting of the BP neural network, which will result in a psychological state identification model based on the improved PSO-BP neural network. The results show that the recognition accuracy reaches 95.5% compared with the actual test set classification. In this study, a mental health early warning system is designed, which is based on a multimodal acquisition system and uses the mental state recognition model to analyze physiological data, as a way to detect the psychological state of the adolescents being tested, and then make timely and accurate warnings.
