Implementation Of Fall Detection Based On Plantar Pressure

In China,the problem of aging population is increasingly serious and the proportion of elderly people living alone in empty nests has increased significantly.As the elderly grow older,their body functions will decline and fall easily.Accidental falls have become an important issue affecting the physical and mental health of the elderly.Timely and effective rescue is based on the accurate detection of falls,which has extremely important significance in result.Currently,video,scene and wearable are thought to be the main solutions of the fall detection.Due to advantages of small size,light weight,high degree of integration and low power consumption,suiting users to various environments,wearable method is especially favored by the market.However,the current wearable fall detection device based on inertial sensors is not ideal for dumping or finally lying on the ground.Based on past research,this paper has designed a realization based on plantar pressure to detection of falls.The method of putting the wearable part into the shoe insole is designed to avoid the inherent disadvantages that inertial sensors are installed on the head,neck,waist,etc.The main work of the dissertation includes:Firstly,plantar pressure detection hardware system is designed.The hardware system consists of three modules: pressure acquisition module,data processing module,and wireless communication module.The pressure acquisition module is composed of eight pressure sensors placed in the cotton insole.The sensors are distributed in the designated eight areas according to the skeletal structure of the human foot and are used to measure the pressure value of the specified area.The data processing module takes the high-performance STM32F407 chip as the core.The signal conditioning circuit in this module converts the sensor resistance value into a stable voltage value within the chip’s measurable range.Data processing module realizes the real-time measurement of the plantar pressure data.The wireless communication module packages the data from the data processing module according to the protocol and sends it to the computer through Bluetooth.Secondly,fall detection acquisition analysis software is designed.The computer software has three functions: remote control of plantar pressure hardware system,Simultaneousily receiving and displaying plantar pressure data,determination of fall behavior.Using Bluetooth wireless communication,software can send four command reports as activation,hibernation,collection and stop then receive response reports.This software receives data reports synchronously,analyzes pressure data.Intuitive waveforms are used to display multi-channel pressure data in this system.The software has data processing functions such as time window interception,model training and fall detection as well.Then,fall detection algorithm based on distribution of the plantar pressure is designed.Through the plantar pressure collection system above,the process of the plantar pressure during body movement and the corresponding pressure value can be obtained,but the detection of the fall requires further pattern recognition algorithms.This paper proposes a feature extraction method based on gait features,and uses Support Vector Machine(SVM)two-category method to distinguish between fall behavior and four types of non-fall behavior.Finally,plantar pressure measurements and fall detection experiments are completed.In order to evaluate the detection performance of the system’s fall detection,four types of daily behavior’s plantar pressure and fall behavior’s foot pressure data were collected during the experiment.We Quantitativeily analyze the eigenvalues of the five types of behavior and analyze the similarities and differences between fall behaviors and daily behaviors.Experiments have shown that the system has good accuracy and reliability for human fall behavior and can be used for human fall detection.