Pattern Matching Between Bisonar Pulse Sequence Trains And Peripheral Dynamic Structural Deformations In Pratt's Roundleaf Bats

Hipposideridae and Rhinolophidae bats can move freely in very complex natural environments and efficiently complete sensing tasks such as target identification and prey capture.They both emit ultrasonic sonar signals from nostrils,receive echo signals by ears,and obtain sensing information about the surrounding environment by analyzing the echo signals.Its sonar systems have unique dynamic characteristics,which are mainly shown as follows:(1)Bats can dynamically adjust parameters such as intensity,direction,duration and sound frequency of ultrasonic sonar signals according to the difference of surrounding environment and sensing tasks,for example,bats emit more ultrasonic pulse signals and sound frequency is higher in areas with dense vegetation or prey capture phase;(2)For sonar system structure,i.e.folded structure(i.e."noseleaf")in the shape of baffle around nostril and pinna,dynamic deformation in different forms can occur under the control of muscle during the process of ultrasonic signal transmission and echo signal reception,for example,noseleaf can undergo closing-opening motion(i.e.closing-opening motion)and random motion;The pinna can move rigidly and flexibly.Numerical simulation and bionic research show that the movement of noseleaf and pinna can significantly change the sound field distribution and provide additional useful sensing information for bats.Biological research has found that there is a certain coupling relationship between the movement of noseleaf and pinna,that is,there is a coupling relationship between the structural dynamic characteristics of the ultrasonic transmitting system and the echo receiving system,while whether there is a coupling relationship between the two dynamic characteristics of the bat sonar system(i.e.,dynamic sonar signal and dynamic structural deformation)is still unknown.In this paper,the research object is Pratt's roundleaf bats.The high-speed camera and ultrasonic microphone are used to synchronously collect the bat's noseleaf/pinna motion data and its corresponding sonar pulse sequence data.Aiming at the two coupling motion modes of the sensing end with high frequency,i.e.noseleaf opening-pinna flexible motion and noseleaf random-pinna rigid motion,the sonar pulse sequences of the two coupling motion modes are classified by using machine learning algorithm to study whether there is coupling relationship between dynamic structural deformation and variable sonar signals.This will improve the theoretical model of dynamic characteristics of bat sonar system and deepen bats' efficient understanding of coded information.Based on the small attenuation of sonar signal in some media,the working range of engineering sensing system is widened.It provides technical guidance for the development of bionic sensing system and further optimizes the detection performance and structural size of engineering sonar.The following part focus on our main research contents and final results:1.Be familiar with the ultrasonic acquisition system of sensing end movement.The system mainly includes hardware and software.Hardware contains binocular high-speed camera array,ultrasonic microphone and auxiliary equipment,while software includes GigaView acquisition software,Matlab control and ultrasonic data visualization code matched with the camera.Through software and hardware interactive control and data flow direction,the factors that may affect the experimental accuracy in the experimental steps are controlled,which is conducive to reducing the experimental error.2.102 sets of images and ultrasonic data from 66 experiments of 14 male Pratt's roundleaf bats were obtained,then stereo vision,acoustic signal processing and other methods were used to respectively restore the three-dimensional coordinates marked on the noseleaf/pinna and simplify the combination of ultrasonic sequences into time points.According to the quantitative motion indexes determined between the three-dimensional coordinate points,such as the noselaf width,height and width of pinna,it can be seen that the bat's noseleaf has open,close and random motion patterns.Pinna movement has rigid and flexible motions.And there are two coupling motion modes of sensing end with high frequency,among which 53 groups are nasal leaf opening-pinna flexible motion and 49 groups are noseleaf random-pinna rigid motion.The difference of sonar pulse sequence in different noseleaf/pinna coupling motion modes is mainly manifested in the difference of pulse number and pulse excitation time,among which,the difference of pulse number is the most significant,only one pulse exists in all noseleaf opening-pinna flexible motion periods,and two pulses exist in all noseleaf random-pinna rigid motion periods.3.Using machine learning algorithm,spike train similarity space(SSIMS)and Spike Community Detection(SCD)are used to classify and cluster the ultrasonic sequence pattern differences corresponding to noseleaf opening-pinna flexible motion and noseleaf random-pinna rigid motion respectively.The experimental data analysis results show that SSIMS and SCD methods have established the matching of ultrasonic pulse sequence and sensing end motion to a certain extent.