基于多权值神经网络的老人跌倒智能识别研究

随着我国人口老龄化及对延年益寿的期望加剧,老年人的健康问题受到广泛的关注。针对这一社会问题,建立人体跌倒模型,并对三轴加速度传感器采集来的不同人体跌倒姿态的高维数据做主成分分析(PCA),降维处理使其特征投影到低维空间,再将降维后的特征向量借助多权值神经元网络算法识别人体跌倒姿态。最后,实际采样的人体跌倒姿态数据验证了该方法的有效性。此外,与支撑向量机(SVM)算法相比较,实验结果表明,多权值神经元网络比支撑向量机算法在人体跌倒应用中更加具有优越性。

Research on the Intelligent Recognition of Falling of Aged People Based on Multi-weights Neural Network

With our aging population and the increased expectations of longevity, the health problems of the elderly have widely attracted attention. In response to this social problem, the model of falls was created. The principal component analysis (PCA) was used to obtain the principal components of high-dimensional data from different body posture collected by three-axis accelerometer. Dimension-reduced processing makes features project onto low-dimensional space. Further, feature vectors from dimension reduction are utilized to train multi-weights neural network (MWNN) to identify the falling body posture. Finally, actual sampling data of the falling body posture have verified the validity of the method. In addition, simulation results also indicate that MWNN utilized in this paper is more excellent than support vector machine (SVM) in the application of falling of aged people.