Research On Emotional Physiological Response And Identification Methods Of Robot Assisted Active Rehabilitation

The latest research results of sports rehabilitation medicine show that patients' active participation in sports rehabilitation training can improve the effect of rehabilitation training effectively.At present,the highly successful robot-assisted rehabilitation training mainly relies on biomechanics and bioelectricity to achieve interactive control.However,these two methods cannot guarantee that patients maintain a proactive state during the training process.Based on the above research status,this paper developed Robot-assisted active rehabilitation emotional physiological response experiments and identification methods,specific work is as follows :(1)This paper understanded the causes of stroke and the insufficiency of traditional rehabilitation training,and known the limitations of patients' emotions in real-time during the rehabilitation process.At the same time,analysis and research was co nducted on robot assisted rehabilitation at home and abroad.Based on the above studies,an emotional model based on rehabilitation training for stroke patients was established.(2)Robot assisted active rehabilitation emotion physiology measurement expe riment design.Firstly,by using the Barrett WAM robotic arm as the hardware basis,an emotional physiology experimental platform for robot-assisted active rehabilitation training for stroke patients is designed;secondly,based on the rehabilitation training experimental platform,the goals of “frustration”,“excitement”,and “boring” are respectively induced.Finally,the emotional physiological survey questionnaire was designed to select the healthy subjects and stroke patients to carry out the emotional physiological measurement experiments for the tactile interactive rehabilitation training,and the physiological parameters detection sensors were used to collect the subjects' ECG,skin electricity,respiration and facial electromyography(frowning)under different emotional states.Physiological response parameters such as muscle and zygomatic muscle).(3)Robot assisted active rehabilitation emotional response feature extraction.The physiological response characteristics of the subjects under the three targets of "frustration","excitement" and "weary" were extracted from the subjects' facial electromyography(scowl muscle and zygomatic muscle),electrocardiogram,pulse and skin electrophysiological response parameters.(4)Experimental research on robotic assisted active rehabilitation rmotional physiology response.By using the multi factor experimental design method,respectively from the significance,the three aspects,the difference and correlation analysis of physiological characteristic parameters in different training mission,performance data,questionnaire survey the significant differences and different training task of questionnaire and physical characteristic parameters,the questionnaire and the relevance of performance data.(5)Emotion recognition method for robot assisted active rehabilitation.For the three kinds of target emotions,a multi-class emotion recognition classifier of the binary tree method is constructed,and an adaptive fuzzy support vector machine algorithm is selected to perform emotion recognition,and a cross-validation method is used.To verify the performance data can significantly improve the recognition rate.Therefore,two kinds of samples are designed: samples composed of physiological parameters,physiological parameters and performance data.The experimental results show that the recognition rate of the second sample is 6.63% higher than that of the first sample.