Research On Audio Monitoring System Based Household For Aged People

As aging of China's population is accelerating,research on monitoring of aged people has just started rising,but still not enough for expected requirement.Currently most monitoring system design needs people to wear powered devices.Disadvantages are that devices are constrained by miniaturization and energy consumption,and such devices are relatively expensive.For household indoor environment,this thesis proposes a monitoring platform design based on micro-phone sensor network.It uses audio signal as the only monitor-ed signal,utilizes smart device terminal to collect audio signal,adopts wireless net-work to transmit data and directives,and introduces one acoustic geometrical location algorithm based on energy ratio as the core function of this monitoring system.At first,a Matlab simulation experiment confirmed rationality of the principle and feasibility of the algorithm.Then,a microphone recording test was performed on four heterogen-eous Android devices involving parameter amendment.Finally,a test platform was built and on which the demo experiment attained usable precision for monitoring.The thesis also explores household sound classification and alarm method on abnormal sound signal.Common household sound samples were selected from open dataset,on which procedures like time and frequency feature extraction,feature selection and so on was applied.Accuracy of 3 classifiers all achieved improvements around 10%in relative to baseline result.Based on classification thought,the thesis proposes abnormal sound alarm method aggregating multiple sound features which obtained fine testing results on the dataset.In this design localization and alarming can be accomplished with solely audio signals,and idle smart devices can be reused to reduce cost.There's no need for aged people to wear equipment anymore,thus easier to use than traditional methods.