Multi-resonance flextensional hydrophone for open-circuit scuba diver detection

Compared to ships, open-circuit scuba divers are small underwater targets that are very difficult to detect for wearing self-contained breathing systems and they are small in size and always swim at low speed, thereby posing challenges to underwater safety alerts and life rescue.¹ 

Many researchers have studied the underwater acoustic signal of the open-circuit scuba diver. Stolkin et al. have found that underwater respiratory noise can reliably detect diver targets.^2,3 Lohrasbipeydeh et al. have found that the frequency range of the underwater acoustic signal of open-circuit scuba divers in the port marine environment covers 100 Hz to 10 kHz, and the inhaling signal is concentrated mainly in the frequency range from 2 to 12 kHz, which is mainly due to the vibration of the pressure relief valve, while the exhaling signal is mainly concentrated in the range of 200 Hz to 1 kHz, which mainly comes from the discharge of the bubble group.⁴ Erbe et al. have used hydrophones to measure the acoustic signal within 16 kHz of the open-circuit scuba diver and analyzed the acoustic signal characteristics of the diver’s inhalation and exhalation, which provide a basis for the identification of the human target underwater.^5,6 Korenbaum et al. have studied the underwater acoustic signal of the diver below 5 kHz and found that the exhaling signal of a diver is mainly distributed in the frequency range from 200 to 1000 Hz, which is derived from the tumbling of the bubbles exhaled by the diver, while the inhaling signal is mainly distributed in the frequency range from 2.5 to 4.9 kHz, which mainly comes from the high-pressure relief valve of the diver’s underwater breathing apparatus.^7–9 Through the experiment in the lake, Wang Ping has found that the acoustic signal of the bubbles exhaled by the open-circuit scuba diver is concentrated below 2 kHz and the energy is prominent. However, the vibration energy of the pressure relief valve caused by the diver’s inhalation is mainly concentrated above 2 kHz, where the diver’s identification characteristic is very prominent. The feasibility of passive sonar to reliably detect respiratory acoustic signals is further verified in the experiment.^10,11 In addition, the Cerberus 360 sonar produced by QinetiQ sets the passive detection frequency range of the diver from 2 to 8 kHz.¹² 

However, there are still few types of passive detection sonar systems for divers around the world, and the existing sonars generally have no large-scale application; the main reasons for that are as follows:

1. Most researchers only study the respiratory signal characteristics of divers suspending or swimming in the water, while few research their signal characteristics in the actual underwater swimming process. However, divers often swim underwater and it is necessary to study their acoustic signal characteristics in two different states of suspension and swimming.

2. When collecting underwater acoustic signals of divers, most researchers only focus on the signal within 20 kHz or even lower frequency, which are vulnerable to the interference of shallow sea background noise, but do not study the signal characteristics of higher frequency. The main reason may be limited by the sampling rate and other performance of the acquisition instrument.

3. A diver itself is a small target with a weak signal, and whose low-frequency signal is easy to be affected by the background noise of the marine environment, which is mainly distributed below 20 kHz and can reduce the signal-to-noise ratio of the diver signal, resulting in short distance and poor robustness of the passive detection.

4. In addition, many researchers use common standard hydrophones that are easily available in the market to detect divers, while most standard hydrophones have low sensitivity—generally lower than −200 dB (ref. 1 V/μPa)—and their wide bandwidth lacks pre-selectivity to the diver signal, which leads to the short detection distance of the diver target and the complexity of back-end signal processing.

To solve these problems in diver detection, in this paper, a high-sensitivity, multi-resonance, narrow-band hydrophone was specially designed and fabricated for open-circuit scuba diver detection. The acoustic signal of a diver in two states of suspending and swimming was analyzed through experiments by using standard hydrophones and acquisition equipment with a higher sampling rate, and the main frequencies of the underwater acoustic signal of a diver were determined. Two frequency ranges at high frequency and low frequency with better signal-to-noise ratio were selected as the basis for designing the special hydrophone for diver detection. Then, according to the high sensitivity characteristic of the flextensional hydrophone, a Cymbal hydrophone with multi-resonance frequencies was designed by using the finite element method (FEM), and a prototype was made. The Cymbal hydrophone is the fifth type of flextensional transducer that has the advantages of simple structure, low cost, high sensitivity, and is suitable for batch processing. Through the performance test of the prototype and underwater test for diver detection using the prototype, it was found that the high and low resonant frequencies of the hydrophone developed in this paper are well-matched with the two frequency ranges of the diver signal that have high signal-to-noise ratio, high receiving sensitivity, and narrow bandwidth at the corresponding resonant frequency, which is expected to improve the detection range of underwater divers and reduce the complexity of back-end signal processing.

To design a special hydrophone for diver detection, it is necessary to fully understand the characteristics of the underwater acoustic signal of the diver. Therefore, we collected the acoustic signal of a diver in two states of suspension and swimming in the pool using omnidirectional standard hydrophones and an acquisition instrument with a high sampling rate to analyze the signal.

We collected the acoustic signal of a professional diver in a constant temperature diving pool, as shown in Fig. 1. The length × width × depth of the pool is 10 × 5× 3 m³, the wall and bottom were waterproof bricks, and the earth was below the bottom of the pool. The diver who participated in the experiment was a 23-year-old man with more than 200 h of diving experience, 1.72 m tall, and 64.6 kg of weight. He wore a full set of open-circuit scuba diving equipment. The standard hydrophones used were two omnidirectional hydrophones with an effective working frequency range from 10 Hz to 100 kHz, with the receiving sensitivity of −208.0 dB (ref. 1 V/μPa), the fluctuation of the sensitivity curve was ≤±3 dB, the signal sampling rate of acquisition equipment was 90 kHz per channel with 24-bit accuracy, and the hydrophones and acquisition equipment had been professionally calibrated before the experiment.

The distance between the two standard hydrophones was 0.1 m. These two hydrophones were jointly arranged at the bottom and about 0.3 m away from the bottom of the pool, and located in the center of the sidewall and about 0.3 m away from the pool wall. The connecting line of the two hydrophones was parallel to the bottom and wall of the pool.

During the experiment, the pure environmental background noise in the pool, the underwater acoustic signal of the diver in the fixed depth suspension state (in the center of the pool, suspended at 1.5 m water depth, only maintaining basic respiration), and the swimming state (maintaining 1.5 m water depth, swimming back and forth normally along the center of the pool, and the swimming route is about 2.5 m away from the pool walls on both sides, as shown in Fig. 2) were collected. Each type of signal was continuously collected for 75 s.

The spectrograms (frame length 1024, Hamming window, window length 1024, 50% overlap) of the three types of signal within the Nyquist frequency were analyzed and shown from Figs. 3–5.

It can be analyzed that respiration was the main source of the diver’s underwater acoustic signal recognized by the ear—the diver’s exhaling signal mainly came from the tumbling of exhaled bubbles in the water, while the inhaling signal mainly came from the vibration and sound of respirator pressure relief valve. The periodic respiratory characteristics of a diver underwater were obvious, and the respiratory cycle was about 2–7 s. The respiratory cycle was significantly longer during suspension due to less oxygen demand than during swimming. Whether in a suspended state or swimming state, the diver’s acoustic signal was mainly concentrated within 8 kHz and around the 34.5 kHz frequency. The energy of the exhaling signal below 2.5 kHz was higher than that of the inhaling signal, and the energy of the inhaling signal within 3–45 kHz was higher than that of the exhaling signal. The inhalation sound was significantly easier to distinguish than the exhalation sound. Furthermore, the signal energy near the frequency of 34.5 kHz was significantly stronger in the swimming state than in the suspended state.

Therefore, to achieve reliable detection of the open-circuit scuba diver, the resonant frequency of the hydrophone should be designed around 3–8 and 34.5 kHz, as the underwater acoustic signal of the diver in these two frequency ranges has higher energy and better signal-to-noise ratio.

To meet the requirements of multi-resonance and high sensitivity of the diver detection hydrophone, we chose Cymbal as the initial structure of the diver detection hydrophone for design because Cymbal, as the fifth type of flextensional transducer, can be designed with multi-order resonant frequencies and high sensitivity, and is suitable for batch processing for its simple structure.¹³ 

We designed the multi-resonance frequency hydrophone by using the FEM method to obtain high design accuracy.^14,15 The purpose of the design is to make the first-order resonant frequency of the Cymbal fall within the frequency range of 3–8 kHz, close to the median value of 5.5 KHz, and shift to the low frequency, which is difficult to obtain. The high-order resonant frequency should fall near 34.5 kHz. The designed Cymbal hydrophone should have high receiving sensitivity at two resonance frequencies. To reduce the complexity of later signal processing and make the hydrophone have the characteristics of early selection for the diver signal, the bandwidth at the resonant frequency should be narrow, which is generally considered to be less than 10% of the corresponding resonant frequency.¹⁶ 

To reduce the amount of calculation, we used Cymbal’s 1/2-D axisymmetric model for the performance calculation, as shown in Figs. 6 and 7. The polarization direction of the piezoelectric ceramic sheet was the Y-axis (along the thickness direction of the ceramic sheet).^14,15 PZT-5A and the brass available on the market were selected for the constituent materials, and the material parameters were shown in Table I.

In addition, the piezoelectric materials polarized along the Y-axis also need the following performance parameters in the finite element calculation:

1. The relative dielectric constant matrix of the PZT-5A ceramic is

   $ξr=ξr11ξr33ξr11=916830916.$

   (1)
2. The piezoelectric stress constant matrix of PZT-5A ceramic is

   $e=e31e33e31e150e15=−5.415.8−5.412.3012.3C/m2.$

   (2)
3. The elastic stiffness coefficient matrix of the PZT-5A ceramic is

   $C=C11C13C12C13C33C13C12C13C11C44C66C44=12.17.527.547.5211.17.527.547.5212.12.112.262.11×1010N/m2.$

   (3)

The frequency corresponding to the maximum value of Cymbal’s conductance G in a certain frequency range was calculated by FEM, which was the resonant frequency of Cymbal within this frequency range.

In the calculations, the electric charge Q of the node on the positive electrode of the ceramic disk was measured, and the variation law of the hydrophone’s admittance with frequency f can be obtained using the following formula:¹⁷ 

1. The admittance Y of the hydrophone is

   $Y=∂Q/∂tU=jωQU=j2πfQU,$

   (4)

   where U is the excitation voltage loaded on the ceramic sheet, which is generally taken as 1 V during the calculation.
2. For the conductance component G and the susceptance component B of the admittance Y, the following formula is used:

   $Y=G+jB.$

   (5)
3. According to the reciprocity principle of a piezoelectric transducer, a node in the far-field acoustic axis is taken to determine its sound pressure p. Therefore, the current generated by the hydrophone after receiving the external sound pressure is evaluated as follows:

   $I=∂Q∂t=jωQ=j2πfQ,$

   (6)

   where r is the distance between the selected node and the equivalent sound receiving center of the hydrophone.

The transmitting current response level (SIL) is

The sound pressure p is complex, so its modulus is taken as |p|.

Following the spherical wave reciprocity principle, the Cymbal hydrophone satisfies the following equation:

Here, M_f is the free-field voltage sensitivity, S_I is the transmitting current response, and the ratio J_S is constant.

Therefore, the receiving voltage sensitivity level (FFVS) of the hydrophone is

It is measured in dB.

After many calculations, we determined 0.05 m as the diameter of the hydrophone. Then, the corresponding prototype was packaged and manufactured, as shown in Fig. 8. We tested the conductance and receiving sensitivity of the prototype in the anechoic water tank. The conductivity of the prototype was measured by using the HP4294A impedance analyzer, and the receiving sensitivity was measured by using the free-field comparison method. Within the two frequency ranges of 3.7–4.7 and 34–35 kHz, 20 frequency points were taken at equal frequency intervals to measure the receiving sensitivity, and the interval between each measured adjacent frequency point was 50 Hz.

We compared the finite element calculation results of the designed Cymbal with the performance test results of the prototype in the two frequency ranges. The FEM calculation values of Cymbal’s first-order and high-order resonant frequencies were compared with the actual test values. It can be found that the first-order resonant frequency of the designed Cymbal in the finite element calculation was 4253 Hz and the high-order resonant frequency was 34407 Hz. In the actual test results, the first-order resonant frequency was about 4100 Hz and the higher-order resonant frequency was about 34450 Hz. The FEM calculation results and the actual test results are shown in Figs. 9 and 10.

The FEM calculated values of the Cymbal’s receiving sensitivity level and bandwidth were also compared with the actual test values. It can be seen from the finite element calculation that the designed multi-resonance Cymbal had a receiving sensitivity level of −161.6 dB (ref. 1 V/μPa) at the first-order resonance frequency of 4253 Hz and a −3 dB bandwidth of about 400 Hz; the receiving sensitivity level at the high-order resonance frequency of 34407 Hz had reached −193.9 dB (ref. 1 V/μPa) and −3 dB bandwidth was about 880 Hz. In the actual measurement, the receiving sensitivity level of the prototype reached −165 dB (ref. 1 V/μPa) at the frequency of 4100 Hz, while the corresponding finite element calculation value was −161.6 dB (ref. 1 V/μPa), with a difference of 3.4 dB; when the measured frequency was near 34450 Hz, the receiving sensitivity level of the prototype reached −196.2 dB (ref. 1 V/μPa), and the corresponding finite element calculation value was −193.9 dB (ref. 1 V/μPa), with a difference of 2.3 dB. The specific calculation results and the actual test results are shown in Figs. 11 and 12, and the comparison between calculated values and test values can be seen in Table II.

The same experimental method as in Sec. II A was used to test the actual detection performance of the multi-resonance Cymbal we designed, and two multi-resonance Cymbal prototypes developed in this paper were used to collect signals from the same diver in the same pool. The sampling rate remained at 90 kHz/channel, the position of the hydrophones and the diver remained unchanged, but the duration of each acquisition was shortened to 15 s.

The spectrograms (frame length 1024, Hamming window, window length 1024, 50% overlap) of three types of signals by using the Cymbal prototypes within the Nyquist frequency were analyzed and shown in Figs. 13 and 14. At the first-order and high-order resonant frequencies, the multi-resonance Cymbal hydrophones developed had collected obvious periodic respiratory signals of the diver. Moreover, the signal-to-noise ratio near the high-order resonant frequency was significantly better than that of the first-order resonant frequency, and the respiratory cycle of the diver was more obvious near the high-order resonant frequency, which was important for the diver detection in external environments, such as shallow seas, where the background noise signals are mainly distributed below 20 kHz. In addition, it can be seen from the spectrogram that the signal energy of the diver in the swimming state was more significant than that in the suspension state near the frequency of 34.5 kHz.

As shown in Table II, the error between the calculated FEM value and the experimental test value of each index parameter of the multi-resonance Cymbal hydrophone developed was kept within 5%, indicating that the FEM used in this paper had high calculation accuracy. The Cymbal hydrophone developed had high and low resonance frequencies of 4100 and 34450 Hz, respectively, which were matched with the two main signal frequencies of the diver. The highest receiving sensitivity levels at the first-order and high-order resonant frequencies were −165 dB (ref. 1 V/μPa) and −196.2 dB (ref. 1 V/μPa), respectively, which were higher than those of the general standard hydrophones on the market. The −3 dB bandwidths at the two resonant frequencies were 400 and 840 Hz, which were less than 10% of the corresponding resonant frequency, indicating that the Cymbal hydrophone developed is a narrow-band hydrophone with good early selectivity for the underwater acoustic signal of the diver. Furthermore, we found that the underwater acoustic signal of the diver has more significant energy and a higher signal-to-noise ratio in the swimming state than in the suspension state near the frequency of 34.5 kHz in the prototype test, which provides a more reliable basis for underwater diver detection.

In this paper, we used standard hydrophones to collect and analyze the underwater acoustic signal of the open-circuit scuba diver and found that the underwater acoustic signal energy of the diver is mainly concentrated in the frequency range of 3–8 kHz and around the frequency of 34.5 kHz, which is mainly generated by the underwater respiration of the diver, while the signal-to-noise ratio of the frequency of 34.5 kHz is higher and more pronounced when the diver is swimming. According to the frequency distribution of the diver’s underwater acoustic signal, we used the FEM to design a Cymbal hydrophone with multi-resonance frequencies, making its first-order and high-order resonant frequencies match the two main frequency ranges of the diver’s signal, and a prototype hydrophone was prepared. The actual results of the prototype show that the FEM has high design accuracy. The Cymbal hydrophone has the advantages of multi-resonance frequency, high sensitivity, narrow bandwidth, and simple processing. Next, the prototypes were used to detect the open-circuit scuba diver in an experiment, the result of which verifies that the multi-resonance Cymbal hydrophone developed in this paper can detect the open-circuit scuba diver with high sensitivity in both high and low frequency ranges. This multi-resonance design of Cymbal hydrophone is expected to improve the reliability of passive detection of divers. However, all our experiments were carried out at a constant temperature indoors. If we need to comprehensively test the actual detection performance of the multi-resonance Cymbal hydrophone developed for open-circuit scuba divers, we need to conduct experiments in shallow sea environments, such as ports, and optimize the design in the future.

This paper was supported by the China National Natural Science Fund Program (Grant No. 11372350) and the Naval University of Engineering, PLA Scientific Research Development Fund Self-establishment Program (Grant No. 425317S091). H.C. would like to thank the members of the project team for their efforts in designing the hydrophone and the Wuhan Tianjin Technology Co., Ltd. for fabricating and experimenting with the prototype. We also express our sincere gratitude to Professor Deshi Wang of the Naval University of Engineering for his constructive guidance in the composition of this paper.

The authors have no conflicts to disclose.

Yuchen Sun: Conceptualization (equal); Data curation (equal); Methodology (equal); Writing – original draft (equal). Weiyi Chen: Funding acquisition (lead); Investigation (equal); Project administration (lead); Writing – review & editing (lead). Zongji Li: Conceptualization (equal); Formal analysis (equal); Methodology (equal); Software (equal). Huadong Chen: Conceptualization (equal); Formal analysis (equal); Supervision (equal); Writing – original draft (equal). Li Dong: Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Software (equal). Yongyong Zhu: Conceptualization (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Shizhe Wang: Data curation (equal); Software (equal); Validation (equal); Visualization (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.