Accelerometer-based human fall detection using sparrow search algorithm and back propagation neural network

Abstract

To reduce the injury caused by the fall and solve the problems of low efficiency and low accuracy of traditional fall prediction methods, an optimized BP neural network fall prediction model based on the Sparrow Search Algorithm (SSA) is established. Taking sliding window to extract discrete features, selecting the maximum value, minimum value, mean value, and variance as the output indicator, the three-axis acceleration and resultant acceleration as the input of influencing factors, the fall prediction model is established for error prediction after data preprocessing. Experimental results show that the improved BP neural network could avoid falling into the locally optimal solution, and the convergence speed is faster, the fall detection accuracy of 98.3%, 92.0% and 96.1% based on DLR, Smart Fall and URFall datasets, respectively. This study may provide technical supports for wearable fall detection that can adapt to different environmental requirements, portable and low power consumption.