Aoustis 8 Paris Detetion of Sallow Underground Buried Objet Using Air Vibration Probe Yuji Sato a, Tomoiro Okamura b, Koii Mizutani a and Naoto Wakatsuki a a Tsukuba Univ., Tsukuba Siene City, 35-8573 Ibaraki, Japan b Univ. of Tsukuba, 1-1-1 Tennodai, 35-8573 Tsukuba, Japan yuji@alab.esys.tsukuba.a.jp 1693
Aoustis 8 Paris An air vibration probe is a devie for measuring aousti impedane. It avails of a vibration of an air olumn and a delay line osillation. Advantage of te probe is simple and ontat-free detetion. We suggest a sallow underground detetion using te air vibration probe for its appliation. A two dimensional finite differene time domain (2-D FDTD) metod and an experiment ave been performed to evaluate te detetability of te probe. Sand is used for a medium of te ground and an empty polyetylene tereptalate (PET) bottle is used for buried objet. Results of te alulation and experiment are almost agreed qualitatively. Te 5t armoni overtone is used for detetion. Te resonane frequeny is derived by te 2-D FDTD metod and te osillation frequeny is used in experiment. Tey beome ig wen tere is te PET bottle buried in underground. Te dept of te objet is sallower or te learane between te probe and ground is wider, te frequeny beomes iger. Tus, te frequeny beomes ig over te objet wen te probe sans te ground inluding te objet. Tese results mean te probe an detet buried objet in sallow underground. However, te response of te probe is affeted by te irregularity of te ground surfae. 1 Introdution An air vibration probe is a devie for measurement of aousti impedane. Te probe onsists of a loudspeaker, a miropone, a voltage amplifier and a ylindrial tube. Tis onfiguration is similar to an impedane tube wi is general devie for measurement of aousti impedane [1, 2]. It needs to ut a speimen to insert it into te tube, wi is a problem beause te speimen is destruted. Te miropone needs to be moved te interior of te tube in te ase of using a standing wave [1]. Tis requires time and ost to meanially move a miropone. Te measurement using te transfer funtion uses a number of miropones and eavy signal proessing [2]. Te air vibration probe an measure te aousti impedane witout any destrution. A losed iruit is formed by te feedbak of te reeived signal at te miropone. So tat a delay line osillation is generated. A ange of te transfer funtion affets te osillation frequeny. Tus, te miropone does not ave to be moved. Te measurement is sorten in te time and simplified. We suggest tat an underground detetion for an appliation of te probe. Te metal detetor (MD) is used for sensing of landmines. Te ground pulse radar (GPR) is used to detet te nonmetal objet. A robot or a veile mounting te MD or te GPR is studied to detet landmines in safety, reently [3, 4]. Sugimoto studied underground detetion using SH wave and transduer array in aousti metod [5, 6, 7]. Suessful results are aieved in tese metods, owever te devies ave got larger and more ompliated as te detetion as got iger performane. Terefore, tere is availability for a simple devie like as te air vibration probe to know its detetability. In tis report, a buried empty polyetylene tereptalate (PET) bottle in sand is deteted wit te air vibration probe of ontat-free. Simulation and an experiment are arried out to evaluate te property of te probe. 2 Priniple of air vibration probe A priniple of te air vibration probe is explained by a vibration of an air olumn and a delay line osillation. A sound propagation in a tube is sown in Fig. 1. A voltage in te pipe V as funtion of l and, j l V e e j l. (1) Here, l is a distane from an input end to a measurement point on te transmission line, is an angler frequeny, j is an imaginary unit and is a pase veloity of sound. and are amplitudes of forward and bakward sound pressures expressed as following; e j L re j L Vi Z i e Z j L re j L, (2) r, (3) were, V i is a driving voltage, L is a lengt of te tube, Z is a aousti impedane of te tube, and Z i is a synteti impedane of an amplifier and a loudspeaker. A refletion oeffiient r is Zr Z r. (4) Z Z r A radiation impedane of te tube Z r is approximately Z r a a J1 S1 Z. (5) 1 j a a 2 2 2 Here, J 1 is te Bessel funtion of te first kind, S 1 is te Struve funtion and a is an inside diameter of te tube. Te voltage V is derived by solving Eqs. (1) to (5). Toug Z r is te radiation impedane wen te tube is in te free spae, if tere are some boundaries of te aousti impedane Loudspeaker L l Sound Tube Input: V i Transfer funtion: T Output: V Miropone a Medium 1 (Air) Medium 2 (Sand) Medium 3 (Objet) Fig. 1 Semati view of air vibration probe in sensing. 1694
Aoustis 8 Paris 926 u 65 8 d s 158 26 65 1 1 Soure Non-refleting boundary Perfet refletor (Stainless tube) Medium 1 (Air) Sound speed 344.8 m/s Medium 2 (Sand) Sound speed 125.7 m/s Medium 1 (Empty PET bottle) Fig. 2 Arrangement of 2-D FDTD metod. exist near te air vibration probe, r and V are anged. An osillation frequeny of te delay line iruit is determined by a transfer funtion T wi is derived from V and V i. Te osillation ondition is T 1, (6) T 2n, (7) were, n is natural number. A frequeny omplying wit Eqs. (6) and (7) is te osillation frequeny. Wen any boundaries of mediums of different aousti impedane exist, a sound pressure is refleted from te boundary. For example, wen tere is a learane between te probe and sand below te probe, te sound pressure is partly refleted at te boundary of te air and te sand. If some objets buried in te sand, te sound pressure tat enters in sand is also refleted at te boundary of te sand and te objet. T and te osillation ondition are anged beause of te refleted sound pressure. Te osillation frequeny is finally anged. 3 Simulation An effiieny of te air vibration probe is evaluated wit a two-dimensional finite differene time domain (2-D FDTD) metod qualitatively. An arrangement of alulated domain and parameters of mediums are sown in Fig. 2. A sound speed in a medium 1 is 344.8 m/s orresponding to te air. A sound speed in a medium 2 is 125.7 m/s orresponding to te sand. Parameters of buried objet are same to te medium 1 orresponding to an empty PET bottle. Te existene of PET surfae is ignored. A refletion oeffiient in ase of an inidene from medium 1 to te medium 2 is.12 in te measurement. A disretizing interval is 1 mm in spae and.5 s in time. Te Gaussian pulse is inputted at te soure. Te transfer funtion T is alulated from te impulse response. T sows some peaks wi mean te armoni overtones of te tube. Te frequeny of te 5t armoni overtone is treated ereafter. is a learane between te stainless tube wi is expressed wit a rigid wall and te sand. u is te upper side learane of te Frequeny (Hz) Frequeny deviation (Hz) 516 515 514 513 512 511 d = 5 mm d = 1 mm d = 2 mm Witout bottle 54 4 6 8 1 12 14 16 18 2 (mm) 2. 1.5 1..5 Fig. 3 Frequeny of 5t armoni overtone in various and d. d 4 4 16 1 1 1 5 (mm) 1 5 5 16 1 5 1 15 2 25 3 35 4 45 5 s (mm) Fig. 4 Frequeny deviation of 5t armoni overtone in orizontal san. stainless tube, wi is equal to 25 (mm). d is a dept of te buried PET bottle. s is te left side learane between te PET bottle and te exterior boundary of te alulated domain. Te frequeny of 5t armoni overtone in anging from 4 mm to 2 mm at interval 2 mm and d = 5, 1 and 2 (mm)is sown in Fig.3. Te PET bottle is under te tube at s = 5 mm. Te frequeny beomes ig wen is large and d is small in all ases. It is expeted to know te dept of te PET bottle in omparison beause te urve of te frequeny of ea d does not ross ea oter. Te deviation of te frequeny is large wen is small. It is about 3 Hz at = 4 mm. It sows te probe is more sensitive wen te probe approaes to te ground. Diffiulty inreases wen te probe gets away from te ground beause te deviation of te frequenies is small. Te frequeny in anging s from 5 to 5 (mm) at interval of 5 mm, = 4, 1 and 16, and d = 5 and 1 (mm) is sown in Fig. 4. Cange of s is equivalent to orizontal san of te probe. Te left edge of te tube is just above te rigt edge of te PET bottle at s = 5 mm. Te overlapping area inreases gradually wen s omes large. Te vertial axis of 1695
Aoustis 8 Paris x Holder Sound y x-y stage Loudspeaker Miropone Stainless tube Lengt L = 79 mm Inside diameter a = 26 mm PET bottle 6 d 6 GP-IB A-D Sand Tank Fig. 5 Semati view of experiment. PC te Fig. 4 sows te variation of te frequeny from te minimal frequeny in ea arrangement. It beomes large as s inreases. Tis indiates te frequeny beomes ig wen te PET bottle omes under te tube. Te detetion is easier in smaller and d beause te frequeny deviation beomes larger. Te maximal inrement is near 2 Hz wit = 4 mm and d = 5 mm. It is not enoug for uman to distinguis te deviation. Te rate of inrease is large in ase is small. It means te probe is sensitive in small as same as Fig. 3. 4 Experiment An experimental apparatus is sown in Fig. 5. Te diameter of te aperture of te loudspeaker (S.J ES-663) is 64 mm. Te miropone is an eletrial ondenser miropone, its sape is ylinder, and te diameter and eigt are bot 1 mm. Te inside diameter of te stainless tube is 26 mm and te lengt is 79 mm. Te loudspeaker and te tube are attaed by a board wose tikness is 1 mm. Te PET bottle is 6 mm on te side and te eigt is 17 mm. Te widt of te tank is 33 mm, te lengt is 53 mm and te eigt is 29 mm. A room temperature is 23.7 C in all experiments. Te tank is filled wit sand and te upper surfae of sand as been evened out so as to be parallel to x-y stage. Te air vibration probe is fastened on te x-y stage by a older. Te x-y stage is ontrolled by PC via a GP-IB interfae. Te osillation frequeny is logged after te osillating signal is onverted to te digital data by te A-D onverter (National Instrument 662-E) wose sampling frequeny is 1 khz. 4.1 Canging learane between te probe and te ground Te osillation frequeny is measured under various eigts of detetion and depts of te PET bottle buried in sand d. is anged from 4 mm to 2 mm at intervals of 2 mm. d is anged to 5, 1 and 2 (mm), and te ase witout te PET bottle. Results are sown in Fig. 6. Te frequeny as been ig wen te is large or d is small. It agrees te result of te Frequeny (Hz) 547 546 545 544 543 542 541 d = 5 mm d = 1 mm d = 2 mm Witout bottle 54 4 6 8 1 12 14 16 18 2 (mm) Fig. 6 Osillation frequeny in various and d. simulation sown in Fig. 3. Terefore, it is expeted te air vibration probe an detet te PET bottle buried witin 2 mm. However, te frequenies at d = 1 and 2 (mm) are almost te same. Tus, it is diffiult for te air vibration probe fabriated in tis experiment to distinguis a PET bottle buried deeper tan 1 mm. It is different from te result of simulation. Te attenuation is ignored in te simulation, wi is te reason of tese differenes. Te amplitude sould be large to inrease te detetable dept omplimenting te transfer loss. Te variation of te frequeny is about 4 Hz at d = 5 mm. It is about 2 Hz in alulation. Tis differene is aused by te differene of te dimension. Te alulation is performed in 2 dimension. Tus, te variation of te overlapped area of te diameter of te tube and te PET bottle is larger in te experiment tan one of te alulation. 3 dimensional alulation is one of future works. 4.2 Buried objet detetion by sanning Te air vibration probe is sanned along te ground in pratial using. Tus, it is important to evaluate te air vibration probe in te orizontal san. Te air vibration probe is moved above te buried PET bottle. d and are varied 5 and 1 (mm) and 4, 1 and 16 (mm) respetively. Te widt of san is 2 mm at interval of 5 mm. Results are sown in Fig. 7. Te vertial axis of te figure means a frequeny deviation from an average frequeny. Te orizontal axis y means te position of te rigt edge of te stainless tube. Two vertial broken lines troug graps indiate te side limit of te overlapped area of te stainless tube and te PET bottle. Te frequeny beomes ig wen te probe omes over te PET bottle. Te peak of frequeny appears wen te probe is sanned troug te PET bottle. It agrees wit te result of te simulation. Any evident peaks of te frequeny deviation are not observed between te vertial broken lines wen te PET bottle is not buried. It is possible to detet buried objet by using te air vibration probe in sallow underground in te experiment. Neverteless, some problems exist. Te frequeny deviation is too small. Te maximum deviation is about.3 Hz in = 4 and d = 5 (mm). Additionally, te probe is 1696
Aoustis 8 Paris.5 d = 5 mm d = 1 mm Witout bottle.3 (a) = 4 mm Frequeny deviation (Hz) -.4.15 -.25.15 (a) = 4 mm (b) = 1 mm () = 16 mm Frequeny deviation (Hz) -.3.1 (b) = 1 mm -.1 Stainless tube Large gain Hig frequeny Normal 2 PET bottle 25 5 75 1 125 15 175 2 y (mm) -.2 Stainless tube d PET bottle Fig. 8 Osillation frequeny in orizontal san using louder and iger tone wit (a) = 4 mm and (b) = 1 mm. affeted by te irregularity of te ground surfae. It is diffiult to determine weter te peak is aused by te buried objet or te irregularity of te ground surfae. Te frequeny of = 4 and d = 1 (mm) makes a peak between y = and 75 (mm) as sown in Fig. 7 (a). Tis is due to te sligt onavity of te ground. Te air vibration probe beomes more sensitive wen is smaller. However, it beomes more diffiult to use te probe above an irregular ground. 25 5 75 1 125 15 175 2 y (mm) Fig. 7 Osillation frequeny in orizontal san wit (a) = 4 mm, (b) = 1 mm and () = 16 mm. 4.3 Expansion of detetable dept Te detetable dept is an important property of te probe. Inreases of T is one of te available metods beause te larger sound pressure is radiated from te probe. It also 1697
Aoustis 8 Paris beomes larger wen te osillation frequeny beomes ig beause te radiation impedane is almost equal to te aousti impedane of te tube in su a ase. Te gain of te amplifier is inreased to make T large. An eletri filter is used to make te osillation frequeny ig. Te probe is sanned over te PET bottle buried in d = 2 mm wit anging = 4 and 1 (mm). Te amplitude of te delay line osillation is 1.4 Vp-p and te osillation frequeny is about 55 Hz wen te gain is enlarged. Te amplitude is 1.2 Vp-p and te frequeny is about 76 Hz wen te filter is used. Te measurement same as te former subsetion is done for omparison. Tese tree measurements are named as Large gain, Hig frequeny and Normal in order to prevent onfusions, ereafter. Results are sown in Fig. 8. Te detetion performanes of Large gain and Hig frequeny are better tan Normal in Fig. 8(a) beause te frequeny peaks of tese two metods appear over te PET bottle but Normal does not. Altoug te frequeny peaks of tree metods appear over te PET bottle in Fig. 8(b), Normal sows a blunt peak and Hig frequeny sows a very small ange. Large gain is te best metod among tree metods in tis experiment beause te peak is sarp and lear. It is neessary to evaluate about te sape or te iruit design of te probe for improvement. 5 Conlusion We suggest a sallow underground detetion using te air vibration probe. Te performane of te probe as been evaluated by te 2-D FDTD metod and te experiment. Te frequeny of te 5t armoni overtone as been alulated. Te osillation frequeny as been measured in te experiment. Results ave sown some arateristi properties as following in alulation or te experiment. (1) Te frequeny beomes iger wen te learane between te probe and te sand is large. If te empty PET bottle exists under te sand, wen te dept of te PET bottle is sallower, te frequeny beomes iger. (2) Wen te probe as sanned te sand inluding te PET bottle in orizontal position, te frequeny beomes ig over te PET bottle. (3) Te probe as good detetability wen te amplitude of te sound is large. However, tere are some problems wi are te detetable dept is extremely sallow and te probe is affeted by an irregularity. Enlargement of te gain of te amplifier is one of effetive metod to inrease te detetable dept. Referenes [1] ISO1534-1, Determination of sound absorption oeffiient and impedane in impedane tubes -- Part 1: Metod using standing wave ratio [2] ISO1534-2, Determination of sound absorption oeffiient and impedane in impedane tubes -- Part 2: Transfer-funtion metod [3] S. Matsunaga, K. Nonami, A Consideratino on Trajetory Following Control and Detetion Performane o Controlled Metal Detetor Mounted of Mine Detetion Robot, Nion Kikai Gakkai Ronbunsyu (C) 72, 169-176 (26) [in Japanese] [4] F. Oka, C. Jyomuta, N. Aomori, Y. Sakamoto, T. Kenmizaki, F. Kitagawa, N. Misumi, Integrated ultrawide bandwit vetor-radar-type-sensor system, IEICE. Te. Rep.61, 47-51 (25) [in Japanese] [5] T. Sugimoto, H. Saito, M. Okujima, Improvement of Underground Image: Underground Imaging Using Sear Waves, Jpn. J Appl. Pys. 36, 3197-3198 (1997) [6] T. Sugimoto, H. Saito, M. Okujima, Improvement of Underground Image (II): Underground Imaging Using Sear Waves, Jpn. J Appl. Pys. 37, 312-3121 (1998) [7] N. Yosizumi, T. Sugimoto, Improvement of Underground Image (III): Underground Imaging Using Sear Waves, Jpn. J Appl. Pys. 4, 3621-3622 (21) 1698