Contactless Multi-dairy Goats Respiration Monitoring Using Acoustic Signal

The dairy goat industry plays an important role in our country's dairy industry.Driven by the national dairy industry revitalization strategy,our country's dairy goat industry is developing rapidly.Smart farming is the only way for the development of dairy goat industry.Monitoring the physiological status of dairy goats in the breeding scene through intelligent and large-scale technology can provide data support for the management and decision-making in the process of dairy goat breeding,and optimize the management mode of the breeding industry,thereby increasing the production of goat milk and the income of farmers.Respiration is one of the most basic and important physiological signs for dairy goats.Common infectious diseases often manifest as abnormal respiration patterns.The high incidence of diseases in the dairy goat breeding industry will lead to a large number of deaths in dairy goat population and cause greater economic losses to farmers.Therefore,it is of great significance for the development of our country's smart breeding industry to detect dairy goat diseased individuals in time,buy time for farmers to intervene in treatment,block the spread of diseases,and improve the survival rate of dairy goats.However,the existing contact respiration monitoring methods will interfere with the natural state of dairy goats and may easily cause equipment damage,while the contactless respiration monitoring methods have the problems of being sensitive to location,orientation and occlusion of the measured targets,and insufficient multitarget respiration monitoring capability.So,it is difficult to apply them well in practice.In view of the above problems,this study adopts the acoustic perception method,utilizes the rich indoor acoustic multipath reflection phenomenon,and implement the multi-target respiration monitoring of dairy goats in natural resting state based on system impulse response,which is robust to location,orientation and occlusion.The main contents and results of this study are as follows:(1)The design of the multi-channel synchronous acoustic transceiver platform.In view of the narrow frequency response range of the existing acoustic perception platform,which will interfere with the daily state of dairy goats and does not have the ability to transmit signals in all directions.To meet the needs of high synchronization of transmitting and receiving signal,wide coverage,convenient communication and control,we design and implement a multichannel synchronous platform for transmitting and receiving acoustic signal.The platform mainly includes USB sound card and acoustic signal multiplexer.The platform supports omnidirectional synchronous acoustic signal transmitting and receiving in the frequency band of 20 Hz to 40 k Hz,the sampling rate is 96 k Hz.Meanwhile,the platform has USB communication interface,which can be directly recognized as a general audio device by common operating systems.It provides platform support for multi-target dairy goats respiration monitoring in natural resting state.(2)Design of multi-target respiration monitoring method based on system impulse response.Aiming at the shortcomings that the existing contactless multi-target respiratory monitoring works are sensitive to location,orientation and occlusion of the measured targets.We first use acoustic signal and draw support for the rich indoor acoustic multipath effect to propose an omnidirectional multi-target thoraco-abdominal undulation signal extraction method based on system impulse response.Then,we use acoustic system impulse response to profile the signal passing through the thoracic and abdominal heave from the multipath signal received by the microphone,and demodulate the thoracic and abdominal undulation signal from it,so as to implement respiration monitoring which is robust to location,orientation and occlusion of measured targets.Meanwhile,we use the autocorrelation function to construct a high-precision frequency calculation method to implement respiratory monitoring in natural resting state,which can adapt to changes in the number of measured targets and multiple targets with similar respiration frequencies.It provides method support for multi-target dairy goats respiration monitoring in natural resting state.(3)Experimental verification of respiration monitoring system.Based on the multichannel synchronous acoustic transceiver platform and multi-target respiration monitoring method.We design and implement a multi-target respiration monitoring system for dairy goats.The ground truth respiration of the dairy goats in natural resting state is recorded by a camera.Comparing the respiration measured by the respiration monitoring system with the ground truth respiration recorded by the camera.We carry out the error analysis of the respiration monitoring results of the respiration monitoring system.In order to fully validate the multitarget respiratory monitoring capability of the respiratory monitoring system proposed in this paper,we conduct respiration monitoring trials on humans.Results show that the average respiration monitoring error for a single dairy goat is 0.32 breaths/min;the average respiration monitoring error for two dairy goats is 0.46 breaths/min;the average respiration monitoring error for three dairy goats is 0.48 breaths/min.It is also verified that the system can implement the respiration monitoring under the condition of similar respiration frequency and has the ability of automatic targets recognition.The respiration monitoring system will not interfere with the daily state of dairy goats,and can solve the problem of location,orientation,occlusion sensitive of measured targets.What's more,the system can solve the problem of insufficient respiratory monitoring capability for multiple targets.In general,this system can implement multi-dairy goats respiration monitoring in natural resting state,which is robust to location,orientation and occlusion.