Parkinson's Disease Monitoring And Auxiliary Rehabilitation Device Based On Pneumatic Muscle System

Parkinson's disease is a common neurological disease of the middle-aged and elderly.Its important clinical feature is the symptoms of resting tremor.This project is to collect and study resting tremor at the time of onset,and proposes a classification of Parkinson's disease.Diagnosis method,and develop a wearable device to assist patient rehabilitation.This article first uses the sensor MPU6050 to collect the acceleration signal of the six-axis gyroscope.De-noise the collected signal by Kalman filtering and sliding smooth filtering.After three-axis signal fusion and data normalization,the signal was analyzed in time and frequency domain,and 10 feature vectors including mean and square difference were extracted.Afterwards,a sliding window method is used to segment the time series signals,and the collected data is made into standardized samples that can be used for supervised learning and training.In the identification and classification algorithm of Parkinson's static tremor,firstly,through principal component analysis(PCA),the data is processed for dimensionality reduction under the premise of ensuring classification accuracy,so as to facilitate subsequent training and judgment.Then through support vector machine(SVM),Naive Bayes algorithm,K nearest neighbor algorithm and decision tree algorithm,complete the training and classification of the collected data.Grid search method and random search method are used to optimize and select parameters.After comparing different algorithms,the training results of different algorithms are merged by means of soft voting.The classification accuracy based on one-dimensional features is above 95%,and the classification accuracy of two-dimensional features can reach 99%..In the hardware design and implementation part,based on the braided Mc Kibben-type pneumatic muscle,according to the size and structural characteristics of the human body,two sets of arm-type and jacket-type rehabilitation equipment were designed and implemented.The device is connected to the development board,and the inflation and deflation control is performed through the solenoid valve.After testing,it can effectively reduce the tremor amplitude of 50%-70% of the patients during the static onset,which has a certain help for rehabilitation.In the user interface development part,the use of socket-based network programming and multi-threaded programming to achieve the transmission and storage of data to the host computer.In the user interface,functions such as real-time visual drawing of data and rapid classification of disease conditions are realized.And through the online and offline mode verification,the work of the whole set of equipment from sensing the onset of tremor to controlling the pneumatic muscle to suppress the static tremor is achieved.This article completes the monitoring and classification of the tremor information of Parkinson's patients,and realizes the reduction of the tremor amplitude of the patients through the design of wearable hardware,which has a certain help for the rehabilitation of Parkinson's patients.