Multi-mode fusion positioning method and system for diving bell emergency communication

Abstract

The invention discloses a multi-mode fusion positioning method and system for diving bell emergency communication, and relates to the field of signal localization, and the method comprises the steps: collecting underwater mixed sound signals, and carrying out the spectrum analysis of the underwater mixed sound signals, and obtaining environment background noise characteristic parameters; inputting the underwater mixed sound signal into a preset bistable stochastic resonance system model, and adjusting a barrier height parameter of the bistable stochastic resonance system model according to the environmental background noise characteristic parameter to obtain an effective signal waveform after signal-to-noise ratio enhancement; performing signal separation on the effective signal waveform to obtain a continuous sound wave carrier signal and a positioning pulse signal; doppler frequency shift analysis is carried out on the continuous sound wave carrier signal to obtain a radial speed, and time arrival difference calculation is carried out on the positioning pulse signal to obtain a slant distance; and carrying out data fusion on the radial speed and the slope distance to obtain fused data, and obtaining a motion track and a current three-dimensional coordinate of the diving bell according to the fused data.

Description

Technical Field

[0001] This application relates to the field of signal positioning, and in particular to a multimodal fusion positioning method and system for emergency communication of diving bells. Background Technology

[0002] In deep-sea saturation diving operations, the diving bell serves as the diver's life support compartment, physically connected to the surface support vessel via an umbilical cable. The umbilical cable not only bears the weight of the diving bell but also integrates various lifeline systems, including air supply lines, power cables, and fiber optic communication cables. However, in complex sea conditions or in the event of an accidental collision, the umbilical cable faces an extreme risk of breakage. If the umbilical cable breaks, the diving bell will lose its wired communication capability with the mother ship. Simultaneously, the broken high-pressure air pipe will instantly release a large amount of compressed gas underwater, forming a dense curtain of turbulent bubbles. These bubbles will create a layer of strong scattering medium around the diving bell, severely interfering with the propagation path of underwater acoustic signals.

[0003] In existing technologies, diving bells are typically equipped with two independent emergency communication systems: one is a voice communication system based on an underwater acoustic communicator, used to transmit divers' distress calls and physiological parameters; the other is an emergency positioning beacon, which transmits pulse signals at a fixed frequency for passive positioning by the mother ship. The underwater acoustic communicator uses continuous wave modulation, providing good voice transmission quality, but its signal processing algorithm is primarily optimized for communication quality, lacking precise ranging and velocity measurement capabilities, and thus unable to provide accurate location information for rescue. While the emergency positioning beacon can provide distance information, its single pulse signal is easily overwhelmed by multipath signals generated by bubble scattering in high-reverberation environments, leading to multiple spurious peaks in time delay measurements and causing significant jumps in positioning results. More critically, when the diving bell moves rapidly due to ocean currents, existing positioning systems cannot obtain its velocity vector in real time, resulting in inaccurate trajectory prediction and delays in obtaining the optimal rescue opportunity.

[0004] Furthermore, in the high-reverberation, high-noise environment created after an umbilical cable break, the bubble cloud generated by the continuous leakage from the ruptured trachea creates an acoustic blind zone, causing a sharp attenuation of the signal strength of traditional single-frequency positioning beacons, potentially reducing the signal-to-noise ratio to undetectable levels. Simultaneously, the nonlinear scattering characteristics of the bubbles generate broadband noise in the frequency domain, further deteriorating signal reception conditions.

[0005] In summary, under the high reverberation and strong noise environment caused by the breakage of the diving bell's umbilical cable, the existing underwater acoustic communication system and positioning beacon system operate independently. This results in the inability to accurately locate the three-dimensional coordinates of the diving bell or obtain its speed in real time under bubble interference conditions, which seriously restricts the efficiency and success rate of emergency rescue. Summary of the Invention

[0006] This application provides a multimodal fusion positioning method and system for emergency communication of diving bells, which is used to achieve accurate positioning and accurate acquisition of movement speed in the high reverberation and strong noise environment caused by the breakage of the diving bell umbilical cable, thereby improving the efficiency and success rate of emergency rescue.

[0007] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: Firstly, a multimodal fusion positioning method for emergency communication of a diving bell is provided, applied to a saturation diving support vessel. The saturation diving support vessel is connected to the diving bell via an umbilical cable. The diving bell is equipped with an underwater acoustic communication device and an emergency positioning device, and the saturation diving support vessel is equipped with an acoustic receiver. The method includes: In response to the detection of an umbilical cable breakage signal, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic wave carrier signal at a first center frequency, while the emergency positioning device is controlled to transmit an initial positioning pulse signal at a second center frequency at a preset period. The transmission time of the initial positioning pulse signal is aligned with the preset phase zero-crossing point of the initial continuous acoustic wave carrier signal, forming a dual-channel acoustic signal with a phase-locked relationship. Underwater mixed acoustic signals are acquired by an acoustic receiver, and the spectrum of the underwater mixed acoustic signals is analyzed to obtain the characteristic parameters of the environmental background noise. The underwater mixed acoustic signal is input into a preset bistable stochastic resonance system model, and the barrier height parameter of the bistable stochastic resonance system model is adjusted according to the environmental background noise characteristic parameters to obtain the effective signal waveform with enhanced signal-to-noise ratio. The effective signal waveform is separated to obtain a continuous acoustic carrier signal and a positioning pulse signal; Doppler frequency shift analysis was performed on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel, and the time arrival difference was calculated on the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel. The radial velocity and slant distance data are fused to obtain fused data, and the fused data is then solved to obtain the diving bell's motion trajectory and current three-dimensional coordinates.

[0008] In one possible implementation of the first aspect, the diving bell is further equipped with a strobe light, and the saturation diving support vessel is further equipped with underwater observation equipment. After calculating the diving bell's trajectory and current three-dimensional coordinates, the method further includes: Determine if the slope distance is less than a preset distance threshold; When the slant distance is less than a preset distance threshold, the strobe light is controlled to read the depth information collected by the pressure sensor inside the diving bell, and the depth information is mapped to the pulse emission interval time of the strobe light, so that the strobe light flashes according to the pulse emission interval time; The underwater observation equipment captures the light signal emitted by the strobe light, and calculates the time difference between adjacent light pulses based on the light signal. The current precise depth of the diving bell is obtained by reverse decoding based on the time difference. The depth coordinate component in the current 3D coordinates is replaced with the current accurate depth to obtain the corrected 3D coordinates.

[0009] In another possible implementation of the first aspect, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic carrier signal at a first center frequency, while the emergency positioning device is controlled to transmit an initial positioning pulse signal at a second center frequency at a preset period, including: According to the preset time-frequency slice interleaving protocol, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic wave carrier signal at the first center frequency. The time-frequency slice interleaving protocol is used to define the timing and phase relationship between the initial continuous acoustic wave carrier signal and the initial positioning pulse signal. According to the time-frequency slice interleaving protocol, the emergency positioning device is controlled to transmit positioning pulse signals at specific phase zero-crossing points of the continuous acoustic carrier signal.

[0010] In another possible implementation of the first aspect, the bistable stochastic resonance system model includes a barrier height parameter. The underwater mixed acoustic signal is input into a preset bistable stochastic resonance system model, and the barrier height parameter of the bistable stochastic resonance system model is adjusted according to the environmental background noise characteristic parameters to obtain an effective signal waveform with enhanced signal-to-noise ratio, including: Calculate the environmental noise energy spectral density based on the characteristic parameters of environmental background noise; The target barrier height parameters are determined based on the environmental noise energy spectral density and the underwater mixed acoustic signal. The barrier height parameter of the bistable stochastic resonance system model is adjusted to the target barrier height parameter; Using underwater mixed acoustic signals as input signals and ambient background noise as driving force, the signals are processed through a bistable stochastic resonance system model to obtain an effective signal waveform with enhanced signal-to-noise ratio.

[0011] In another possible implementation of the first aspect, the target barrier height parameter is determined based on the ambient noise energy spectral density and the underwater mixed acoustic signal, including: Extract the effective signal components from the underwater mixed acoustic signal and calculate the signal amplitude of the effective signal components; The noise intensity is obtained by calculating the integral value of the environmental noise energy spectral density within a preset frequency band. The signal-to-noise ratio level is determined based on the ratio of signal amplitude to noise intensity. Based on the preset mapping table, find the target barrier height parameter corresponding to the signal-to-noise ratio level.

[0012] In another possible implementation of the first aspect, signal separation is performed on the effective signal waveform to obtain a continuous acoustic carrier signal and a positioning pulse signal, including: The effective signal waveform is bandpass filtered to extract the first frequency band signal corresponding to the first center frequency, resulting in the separated continuous acoustic carrier signal; and The second frequency band signal corresponding to the second center frequency is extracted to obtain the separated positioning pulse signal.

[0013] In another possible implementation of the first aspect, Doppler frequency shift analysis is performed on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel, and the time difference of arrival is calculated on the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel, including: A short-time Fourier transform is performed on the continuous acoustic wave carrier signal to obtain the time-spectrum diagram; Identify frequency peaks in the time-spectrum graph and calculate the frequency offset of the frequency peaks relative to the first center frequency. The radial velocity of the diving bell relative to the saturated diving support vessel is calculated based on the frequency offset and Doppler effect formula. Perform pulse detection on the positioning pulse signal to identify the pulse arrival time; Calculate the signal propagation delay based on the pulse arrival time and the preset transmission time; Calculate the slant distance between the diving bell and the saturation diving support vessel based on the signal propagation delay and underwater sound speed.

[0014] In another possible implementation of the first aspect, radial velocity and slant range are fused to obtain fused data, and the fused data is then processed to obtain the diving bell's trajectory and current three-dimensional coordinates, including: An extended Kalman filter is constructed, and the three-dimensional position coordinates and three-dimensional velocity vector of the diving bell are used as state variables to establish a state-space model; Extend the Kalman filter by using radial velocity as the velocity observation input and slant range as the distance observation input. By using an extended Kalman filter, the prior state estimate for the current time step is predicted based on the state estimate and state transition matrix from the previous time step. By using the extended Kalman filter, the Kalman gain is calculated based on the current observation and prior state estimate, and the posterior state estimate is updated to the current time. The position coordinate components are extracted from the posterior state estimate and used as the current three-dimensional coordinates of the diving bell. By connecting the current three-dimensional coordinates of multiple consecutive moments in chronological order, the motion trajectory curve of the diving bell is constructed.

[0015] Secondly, this application provides a multimodal fusion positioning system for emergency communication of diving bells, comprising: The diving bell is equipped with an underwater acoustic communication device, an emergency positioning device, and a strobe light; A saturation diving support vessel, connected to a diving bell, equipped with an acoustic receiver and underwater observation equipment.

[0016] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned multimodal fusion positioning method for emergency communication of diving bells.

[0017] The above technical solutions effectively solve the problem of accurate positioning of a diving bell in a high-reverberation, high-noise environment after the umbilical cable breaks. The phase-locked dual-channel acoustic signal design establishes a temporal correlation between the communication and positioning signals, providing a reliable foundation for joint signal processing. Utilizing environmental noise energy to amplify weak signals allows for effective extraction of useful information even under low signal-to-noise ratio conditions caused by bubble scattering, significantly improving the signal-to-noise ratio compared to traditional methods. Doppler frequency shift analysis and time-delay ranging enable simultaneous measurement of the diving bell's position and velocity, overcoming the limitations of single positioning methods in dynamic environments, significantly improving positioning accuracy and trajectory prediction reliability, effectively reducing positioning errors, and enhancing velocity measurement accuracy. In summary, this provides accurate target location and motion status information for deep-sea emergency rescue, significantly shortening rescue response time, increasing rescue success rates, and effectively protecting the lives of divers.

[0018] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a multimodal fusion positioning method for emergency communication of a diving bell, provided as an embodiment of this application; Figure 2 This is a schematic diagram of an optical signal detection process provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a multimodal fusion positioning system for emergency communication of a diving bell, provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0022] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0023] Figure 1 The illustration shows a schematic flowchart of a multimodal fusion positioning method for emergency communication of a diving bell according to an embodiment of this application. Figure 1 As shown in the figure, this application provides a multimodal fusion positioning method for emergency communication of a diving bell, which is applied to a saturation diving support vessel. The saturation diving support vessel is connected to the diving bell via an umbilical cable. The diving bell is equipped with an underwater acoustic communication device and an emergency positioning device. The saturation diving support vessel is equipped with an acoustic receiver. The method may include the following steps.

[0024] S110. In response to the detection of an umbilical cable breakage signal, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic wave carrier signal at a first center frequency, and the emergency positioning device is controlled to transmit an initial positioning pulse signal at a second center frequency at a preset period. The transmission time of the initial positioning pulse signal is aligned with the preset phase zero-crossing point of the initial continuous acoustic wave carrier signal, forming a dual-channel acoustic signal with a phase-locked relationship. S120. Acquire underwater mixed acoustic signals through an acoustic receiver, perform spectrum analysis on the underwater mixed acoustic signals, and obtain environmental background noise characteristic parameters; S130. Input the underwater mixed acoustic signal into the preset bistable stochastic resonance system model, and adjust the barrier height parameter of the bistable stochastic resonance system model according to the environmental background noise characteristic parameters to obtain the effective signal waveform after signal-to-noise ratio enhancement. S140. Separate the effective signal waveform to obtain a continuous acoustic carrier signal and a positioning pulse signal; S150. Perform Doppler frequency shift analysis on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel, and calculate the time arrival difference of the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel. S160. The radial velocity and slant distance are fused to obtain the fused data. The fused data is then solved to obtain the motion trajectory and current three-dimensional coordinates of the diving bell.

[0025] In this embodiment, the bistable stochastic resonance system model includes barrier height parameters, damping coefficients, and driving force parameters.

[0026] When an umbilical cable breakage occurs, the emergency control unit inside the diving bell immediately detects the abnormal state of power line interruption or fiber optic communication failure. Umbilical cable breakage can be detected by both voltage and optical power monitoring modules; when both simultaneously detect signal interruption exceeding a preset threshold time, it is determined to be an umbilical cable breakage event. Once the breakage signal is confirmed, the emergency control unit initiates a dual-channel acoustic signal transmission procedure according to a preset time-frequency slice interleaving protocol.

[0027] Specifically, the underwater acoustic communication device first transmits an initial continuous acoustic carrier signal at a first center frequency, typically chosen within the 8-12 kHz range. This frequency band exhibits good propagation characteristics in the deep-sea environment and can maintain a relatively stable signal strength even under bubble interference conditions. The continuous acoustic carrier signal employs frequency-shift keying modulation, with the carrier phase controlled based on a precise clock source. Simultaneously, the emergency positioning device is activated, transmitting an initial positioning pulse signal at a second center frequency at a preset period. This frequency is typically chosen within the 15-20 kHz range, maintaining a sufficient frequency interval with the communication device to avoid mutual interference.

[0028] In this embodiment, the positioning pulse signal adopts a linear frequency modulated pulse design, with a pulse width typically ranging from 10 to 20 milliseconds and a repetition period set to 1 to 2 seconds. The emergency positioning device incorporates a high-precision phase detection circuit to monitor the phase state of the continuous carrier signal transmitted by the underwater acoustic communicator in real time. When the carrier signal is detected to have passed through a preset phase zero-crossing point (e.g., at a phase of 0 degrees or 180 degrees), the positioning device immediately triggers the transmission of the pulse signal, thereby effectively ensuring that the two signals have a strict synchronization relationship in the time domain, providing a reliable time reference for subsequent signal processing and data fusion.

[0029] When dual-channel acoustic signals propagate underwater, although they are affected by multipath effects and bubble scattering, the receiver can use this prior information for joint processing due to the phase-locked relationship at the transmitter, which significantly improves the accuracy of signal detection and parameter estimation.

[0030] Acoustic receivers on saturation diving support vessels employ a multi-element hydrophone array structure, typically consisting of 4 to 8 hydrophone units arranged in a specific geometric configuration, enabling omnidirectional sound field acquisition. When the umbilical cable breaks, the underwater environment rapidly fills with a dense cloud of bubbles generated by the leakage of high-pressure gas. These bubbles produce strong scattering and reverberation effects, while mechanical vibrations and airflow noise at the break point also generate broadband noise interference. The acoustic receiver continuously acquires mixed underwater acoustic signals at a high sampling rate (typically 96 kHz or higher), which includes dual-channel acoustic signals emitted by the diving bell, ambient background noise, bubble scattering noise, and the vessel's own noise.

[0031] The acquired raw signal is first pre-amplified and anti-aliasing filtered before entering the digital signal processing unit. Spectrum analysis employs the Fast Fourier Transform (FFT) algorithm to convert the time-domain signal to the frequency domain for analysis. Specifically, the acquired underwater mixed acoustic signal is segmented into frames, with each frame typically set to 1024 or 2048 sampling points. A 50% overlap rate is used between frames to ensure spectral continuity. Windowing is applied to each frame to reduce spectral leakage, and then a FFT is performed to obtain the spectrum of that frame.

[0032] Statistical analysis of the spectrum across multiple consecutive frames allows for the identification of stable and variable components. The extraction of environmental background noise characteristic parameters includes noise power spectral density, noise bandwidth, dominant noise frequency components, and time-varying noise characteristics. Power spectral density is calculated by averaging the squares of the spectral amplitude, reflecting the distribution of noise energy in the frequency domain. Noise bandwidth is determined by identifying the frequency range where the power spectral density exceeds a threshold. Dominant noise frequency components are identified by finding local peaks in the power spectral density; these peaks typically correspond to the operating frequencies of ship machinery or characteristic frequencies of marine environmental noise. The time-varying noise characteristics are characterized by calculating the rate of change of the power spectral density at different times, reflecting the dynamic changes in the noise environment. These environmental background noise characteristic parameters provide crucial adaptive adjustment criteria for subsequent stochastic resonance processing, enabling the signal enhancement algorithm to be optimized based on the actual noise environment.

[0033] A bistable stochastic resonance system is a nonlinear dynamical system whose core principle is to utilize noise energy of appropriate intensity to help weak signals overcome a potential barrier, thereby amplifying and enhancing the signal. This system can be described by the Langevin equation, where the potential function adopts the classic double-well form. The parameters a and b together determine the height of the potential barrier and the depth of the potential well. The barrier height parameter directly affects the system's response characteristics to the input signal. When the barrier height matches the input noise intensity, the system will produce the optimal stochastic resonance effect.

[0034] In practice, the environmental noise energy spectral density is first calculated based on the characteristic parameters of the background noise. This calculation is achieved by integrating the power spectral density over the frequency band of interest. Then, the effective signal components in the underwater mixed acoustic signal are extracted. A bandpass filter can be used to initially separate the signal energy near the first and second center frequencies, and the signal amplitudes within these frequency bands are calculated. Next, the ratio of signal amplitude to noise intensity is calculated to obtain the current signal-to-noise ratio (SNR) level. Based on a pre-established mapping table, the target barrier height parameter corresponding to the current SNR level is found. This mapping table, established through extensive simulation experiments and actual test data, records the optimal barrier height parameter configuration under different SNR conditions.

[0035] After adjusting the barrier height parameter of the bistable stochastic resonance system model to the target value, the underwater mixed acoustic signal is fed into the system as the input signal, and the ambient background noise participates in the dynamic evolution of the system as the driving force. The Langevin equation is solved, with the time step typically set to one-tenth of the sampling period to ensure numerical stability. The physical mechanism of stochastic resonance processing lies in the fact that when a weak signal is superimposed on noise of appropriate intensity, the random fluctuations of the noise periodically help the signal cross the barrier, causing the system to transition between two steady states. The frequency of this transition is synchronized with the frequency of the input signal, thereby amplifying the signal.

[0036] After random resonance processing, the signal-to-noise ratio of the output signal is significantly improved compared to the input signal. The resulting effective signal waveform retains the frequency and phase information of the original signal, but the amplitude is significantly enhanced, creating favorable conditions for subsequent signal separation and parameter extraction.

[0037] The effective signal waveform after random resonance enhancement contains two acoustic signal components with different frequencies, which need to be extracted independently using signal separation techniques. Signal separation is achieved using digital filtering, with two sets of bandpass filters designed to correspond to the first and second center frequencies, respectively. The center frequency of the first set of bandpass filters is set to the first center frequency, and the passband width is determined based on the modulation bandwidth of the continuous acoustic wave carrier signal. The filter has a flat amplitude-frequency response in the passband and a steep attenuation characteristic in the stopband. The stopband attenuation is typically required to reach above 60dB to effectively suppress interference from other frequency components.

[0038] After the effective signal waveform passes through this filter, the positioning pulse signal component at the second center frequency is effectively suppressed, and the output signal mainly contains the continuous acoustic carrier signal and its modulation information. The design principle of the second set of bandpass filters is similar. The center frequency is set to the second center frequency, and the passband width is determined based on the bandwidth of the positioning pulse signal. Since the pulse signal uses linear frequency modulation, its instantaneous bandwidth is relatively wide; therefore, the passband width of the filter needs to be increased accordingly, typically set to 20% to 30% of the center frequency. After the effective signal waveform passes through this filter, the continuous carrier signal component at the first center frequency is suppressed, and the output signal mainly contains the positioning pulse signal.

[0039] In this embodiment, to further improve the separation effect, time-domain waveform detection and verification are performed after filtering. For continuous acoustic carrier signals, the continuity of their envelope and the smoothness of their phase are detected. If abnormal amplitude jumps or phase abrupt changes are detected, it indicates that there may be residual pulse interference, requiring secondary filtering. For positioning pulse signals, their pulse shape and repetition period are detected, and the peak value of the pulse signal is further enhanced and noise is suppressed through matched filtering technology. The impulse response of the matched filter is designed as the time-reversed conjugate of the transmitted pulse waveform. When the received pulse signal passes through the matched filter, the output will generate a sharp peak value at the pulse arrival time. The amplitude of this peak value is proportional to the input signal-to-noise ratio, while the noise component is smoothed out due to mismatch with the matched filter.

[0040] After signal separation processing, two independent signals were obtained: the separated continuous acoustic carrier signal retains the frequency, phase and modulation information of the original carrier, which can be used for subsequent Doppler frequency shift analysis; the separated positioning pulse signal retains the arrival time and waveform characteristics of the pulse, which can be used for subsequent time delay measurement.

[0041] Doppler frequency shift analysis was performed on the separated continuous acoustic carrier signal, and the signal was expanded in the time-frequency domain using the short-time Fourier transform (SFT) method. The choice of the SFT window length requires a trade-off between time resolution and frequency resolution, typically set to a length encompassing 10 to 20 carrier cycles. This ensures sufficient frequency resolution for accurate frequency shift measurement while maintaining sufficient time resolution to track dynamic frequency changes. A Gaussian window function was used, exhibiting optimal energy concentration characteristics in the time-frequency domain.

[0042] A short-time Fourier transform (SFT) of the continuous acoustic carrier signal yields a time-spectrum graph. The horizontal axis represents time, the vertical axis represents frequency, and the color or brightness indicates the energy intensity of the frequency component at that moment. The frequency trajectory of the carrier signal can be clearly observed in the time-spectrum graph. When the diving bell moves relative to the saturated diving support vessel, the received carrier frequency deviates from the transmitted frequency due to the Doppler effect. By identifying the frequency peak, i.e., the frequency component with the strongest energy, within each time window, the instantaneous frequency can be obtained. The frequency offset of the instantaneous frequency relative to the first center frequency is calculated; this offset is the Doppler frequency shift. According to the physical principle of the Doppler effect, the relationship between the frequency offset and the radial velocity is: ,in For Doppler frequency shift, For the transmission frequency, denoted as radial velocity, and c as underwater sound speed.

[0043] The speed of sound underwater varies with water temperature, salinity, and depth. In practical applications, it needs to be corrected according to the environmental parameters of the current sea area. Accurate sound speed values ​​are usually obtained using empirical formulas or table lookup methods. The radial velocity of the diving bell relative to the saturation diving support vessel can be calculated using the above formula; a positive value indicates moving away, and a negative value indicates approaching.

[0044] For processing the positioning pulse signal, pulse detection is first performed to identify the pulse arrival time. Pulse detection uses an energy detection method combined with threshold decision to calculate the instantaneous energy of the signal. When the instantaneous energy exceeds a preset threshold, it is determined that the pulse has arrived. To improve detection accuracy, rising edge detection technology is used. After the energy exceeds the threshold, the rising process of the energy envelope is continued to be tracked, and the moment when the energy envelope reaches its peak value is taken as the precise pulse arrival time. Since the transmission time has been accurately recorded in step S110 through a phase-locked mechanism, the signal propagation delay, i.e., the difference between the pulse arrival time and the transmission time, can be directly calculated. According to the propagation characteristics of sound waves underwater, the propagation delay is proportional to the propagation distance. By multiplying the propagation delay by the underwater sound speed, the slant distance between the diving bell and the saturation diving support vessel can be obtained. It should be noted that in a multipath propagation environment, the receiver may detect multiple pulse peaks. The first arriving peak corresponds to the direct path, and subsequent peaks correspond to multipath signals reflected from the sea surface or seabed. By selecting the first peak exceeding the threshold as the valid signal, ranging errors caused by multipath interference can be effectively avoided.

[0045] After obtaining the radial velocity and slant range, a data fusion algorithm is needed to organically combine them to calculate the complete motion state of the diving bell. Data fusion is implemented using an extended Kalman filter framework. Specifically, a state-space model is first constructed, which stores the three-dimensional position coordinates of the diving bell. and three-dimensional velocity vector This forms a six-dimensional state vector. The state transition equation describes the evolution of the state vector over time. Without external forces, the change in position coordinates equals the velocity multiplied by the time step, while the velocity remains constant. The observation equation describes the relationship between the observed quantities and the state variables. The radial velocity observation is equal to the projection of the velocity vector onto the direction from the diving bell to the saturated diving support vessel, and the slant distance observation is equal to the magnitude of the position vector.

[0046] Since the observation equation is nonlinear, it needs to be linearized at the current state estimate, and the Jacobian matrix of the observation matrix needs to be calculated. The recursive process of the extended Kalman filter includes two steps: prediction and update. In the prediction step, based on the posterior state estimate and state transition matrix of the previous time step, the prior state estimate at the current time step is calculated through the state transition equation, and the prior error covariance matrix is ​​updated based on the process noise covariance matrix. In the update step, the predicted observation is first calculated based on the observation equation and the prior state estimate, and then the residual between the actual observation and the predicted observation, i.e., the innovation, is calculated. The Kalman gain is calculated based on the innovation and the prior error covariance matrix. The Kalman gain reflects the degree of confidence in the observation; when the observation noise is low, the Kalman gain is high, and the state estimate relies more on the observation; when the observation noise is high, the Kalman gain is low, and the state estimate relies more on the predicted value. Finally, the prior state estimate is updated based on the Kalman gain and the innovation to obtain the posterior state estimate at the current time step, and the posterior error covariance matrix is ​​updated again. The position coordinate components are extracted from the posterior state estimate, which are the current three-dimensional coordinates of the diving bell.

[0047] By recursively calculating over multiple consecutive time points, a series of three-dimensional coordinate points can be obtained. Connecting these coordinate points in chronological order constitutes the trajectory curve of the diving bell. This trajectory curve not only reflects the historical movement path of the diving bell but can also be used to predict future movement trends, providing important reference for rescue decisions. The advantage of the extended Kalman filter lies in its ability to comprehensively utilize multi-source observation information and minimize the variance of state estimation through optimal weighting. Even when the observation data contains noise and uncertainty, it can still obtain stable and reliable estimation results.

[0048] This embodiment effectively solves the problem of accurate positioning of a diving bell in a high-reverberation, high-noise environment after the umbilical cable breaks. The phase-locked dual-channel acoustic signal design establishes a temporal correlation between the communication and positioning signals, providing a reliable foundation for joint signal processing. By utilizing environmental noise energy to amplify weak signals, useful information can still be effectively extracted even under low signal-to-noise ratio conditions caused by bubble scattering, resulting in a significantly improved signal-to-noise ratio compared to traditional methods. Doppler frequency shift analysis and time-delay ranging enable simultaneous measurement of the diving bell's position and velocity, overcoming the limitations of single positioning methods in dynamic environments. This significantly improves positioning accuracy and trajectory prediction reliability, effectively reducing positioning errors and increasing velocity measurement accuracy. In summary, it provides accurate target location and motion status information for deep-sea emergency rescue, significantly shortening rescue response time, increasing rescue success rate, and effectively protecting the lives of divers.

[0049] In one embodiment of this invention, the diving bell is further equipped with a strobe light, and the saturation diving support vessel is further equipped with underwater observation equipment. After calculating the diving bell's trajectory and current three-dimensional coordinates, the method further includes the following steps: S210. Determine whether the slant distance is less than the preset distance threshold. S220. When the slant distance is less than the preset distance threshold, control the strobe light to read the depth information collected by the pressure sensor inside the diving bell, and map the depth information to the pulse emission interval time of the strobe light, so that the strobe light flashes according to the pulse emission interval time. S230. The underwater observation equipment captures the light signal emitted by the strobe light, calculates the time difference between adjacent light pulses based on the light signal, and decodes the time difference to obtain the current precise depth of the diving bell. S240. Replace the depth coordinate component in the current 3D coordinates with the current accurate depth to obtain the corrected 3D coordinates.

[0050] In this embodiment, the strobe light is controlled to read the depth information collected by the pressure sensor inside the diving bell, and the depth information is mapped to the pulse emission interval of the strobe light so that the strobe light flashes according to the pulse emission interval. This includes the following steps: The strobe light is controlled to read the pressure values ​​collected by the pressure sensor inside the diving bell. Based on the preset pressure-depth conversion formula, the pressure value is converted into depth information; According to the preset depth encoding mapping table, the depth information is mapped to the corresponding pulse emission interval time. The depth encoding mapping table defines the non-linear mapping relationship between the depth value and the pulse emission interval time. The strobe light is controlled to flash periodically according to the pulse emission interval.

[0051] After obtaining the diving bell's trajectory and current three-dimensional coordinates through an extended Kalman filter, the slant range needs to be determined to decide whether to activate the optical-assisted positioning system. The slant range determination mechanism is based on the propagation characteristics of light signals underwater. Due to the absorption and scattering of light by water, the effective propagation distance of light signals is much shorter than that of sound signals. In clear seawater, the attenuation coefficient of visible light is approximately 0.05 to 0.1 per meter, meaning that light intensity decreases to 60% to 37% of its original value every 10 meters of propagation. In turbid water or in the presence of air bubbles, the attenuation is even more severe.

[0052] Determining the preset distance threshold requires comprehensive consideration of the strobe light's luminous intensity, the sensitivity of the underwater observation equipment, and the current water transparency. In typical deep-sea saturation diving environments, the preset distance threshold is usually set to 50 to 100 meters. When an umbilical cable breaks, generating a large number of bubbles, the bubble cloud significantly reduces water transparency, in which case the preset distance threshold needs to be dynamically adjusted to 30 to 50 meters.

[0053] The slant distance determination employs a real-time comparison mechanism, comparing the calculated slant distance value with a preset distance threshold. If the slant distance is greater than the preset distance threshold, it indicates that the diving bell is too far from the saturated diving support vessel, and the optical observation conditions are not met. In this case, the system continues to rely on acoustic positioning methods for tracking, without activating the optical auxiliary system to conserve the diving bell's internal power resources. If the slant distance is less than or equal to the preset distance threshold, it indicates that the diving bell has entered the effective range of optical observation, at which point the optical assisted positioning process is triggered.

[0054] In this embodiment, to avoid frequent switching near the distance threshold, a hysteresis mechanism is introduced into the judgment logic. Two thresholds are set: a start threshold and a stop threshold. The start threshold is slightly lower than the stop threshold. The optical system is only activated when the slant distance decreases from greater than the stop threshold to less than the start threshold, and only deactivated when the slant distance increases from less than the start threshold to greater than the stop threshold. This hysteresis design effectively avoids system jitter at the threshold boundaries and improves system stability. The slant distance judgment also incorporates trend analysis of the motion trajectory. If the diving bell is detected moving towards the saturation diving support vessel, even if the current slant distance is slightly greater than the threshold, the optical system can be activated in advance for warm-up and preparation, ensuring that observation can begin immediately when the diving bell enters the effective range.

[0055] When the slant distance is determined to be less than a preset distance threshold, the emergency control unit inside the diving bell will activate the strobe light system and initiate the depth information encoding process. The strobe light typically uses a high-brightness LED light source, with the emission wavelength selected in the blue-green light band, which has the strongest penetration ability in seawater and a relatively small attenuation coefficient.

[0056] The strobe light control circuit first reads the pressure value collected by the pressure sensor inside the diving bell via the data bus. The pressure sensor's output signal is then converted from analog to digital to obtain a digitized pressure value. Next, the pressure value is converted into depth information according to a preset pressure-depth conversion formula. This formula is based on the principles of hydrostatics, and the relationship between depth and pressure is... Where D is depth and P is current pressure. The atmospheric pressure at sea surface. Let be the density of seawater, and g be the acceleration due to gravity. Seawater density varies with temperature and salinity, and in actual calculations, it needs to be corrected for environmental parameters of the operating area. The converted depth information is in meters.

[0057] Then, the depth information is mapped to the corresponding pulse emission interval time according to the preset depth encoding mapping table. The depth encoding mapping table defines a non-linear mapping relationship between depth values ​​and pulse emission interval time. The design of the mapping relationship needs to consider the time resolution capability of the human eye or optical sensor and the dynamic range of the encoding. Typically, the depth range of 0 to 300 meters is mapped to a pulse interval range of 0.5 to 3.0 seconds.

[0058] Specifically, to improve the robustness of the encoding, the mapping relationship adopts a piecewise linear or logarithmic function design. Smaller interval steps are used in shallow water areas to improve resolution, while larger interval steps are used in deep water areas to expand the encoding range. For example, the depth range of 0 to 50 meters is mapped to an interval of 0.5 to 1.0 seconds, with an increment of 0.1 seconds every 10 meters; the depth range of 50 to 150 meters is mapped to an interval of 1.0 to 2.0 seconds, with an increment of 0.2 seconds every 20 meters; and the depth range of 150 to 300 meters is mapped to an interval of 2.0 to 3.0 seconds, with an increment of 0.2 seconds every 30 meters. The mapping table is stored in the memory of the control unit in the form of a lookup table, and accurate mapping of arbitrary depth values ​​can be achieved through interpolation algorithms.

[0059] After obtaining the pulse emission interval, the control unit generates a corresponding timing signal to drive the strobe light to flash periodically according to the interval. The duration of each emission is fixed at 50 to 100 milliseconds to ensure that enough light energy is captured by the underwater observation equipment, while avoiding excessively long emission time that could cause the encoded information to become blurred.

[0060] Underwater observation equipment on saturation diving support vessels typically includes high-sensitivity underwater cameras or photodetectors, mounted on the underwater portion of the hull or deployed in the water via cables. The underwater cameras employ low-light image sensors, equipped with large-aperture lenses and narrow-band filters. The center wavelength of the narrow-band filter matches the emission wavelength of the strobe light, effectively suppressing ambient light and other interfering wavelengths, thus improving the signal-to-noise ratio. The camera's frame rate is set to 30 frames per second or higher to ensure accurate capture of every flicker event of the strobe light.

[0061] Once the diving bell enters the observation range, the underwater observation equipment begins continuously acquiring image sequences. Image processing algorithms analyze each frame in real time to detect the presence of light signals. Light signal detection employs a brightness thresholding method combined with a region growing algorithm, such as... Figure 2 As shown, the image is first converted to grayscale, then a brightness threshold is set, and pixels with grayscale values ​​exceeding the threshold are marked as candidate light source pixels. Since strobe lights have specific spatial distribution characteristics, they typically appear as one or more high-brightness connected regions. A region growing algorithm can cluster pixels belonging to the same light source together to form light spot regions.

[0062] The algorithm calculates the centroid coordinates and total brightness of the light spot area. When the total brightness exceeds a preset detection threshold, it is considered a valid light signal. To eliminate interference from other light sources in the environment, such as reflections from ship lights or bioluminescence, the detection algorithm also analyzes the temporal characteristics of the light signal. Only light sources exhibiting a periodic flickering pattern are identified as strobe light signals.

[0063] The timing of each optical signal occurrence is recorded in a continuous image sequence. By calculating the time difference between two adjacent optical signal occurrences, the pulse emission interval is measured. To improve measurement accuracy, a multi-period averaging method is employed, continuously measuring multiple pulse intervals and calculating the average value, which effectively suppresses the influence of random errors and measurement noise.

[0064] After obtaining the accurate pulse emission interval, reverse decoding is performed according to a preset depth encoding mapping table to find the mapping table entry closest to the measured interval, thus obtaining the corresponding depth value. If the measured interval falls between two mapping table entries, a linear interpolation method is used to calculate the precise depth value. The depth information obtained from reverse decoding is the current precise depth of the diving bell. This depth value comes directly from the pressure sensor measurement inside the diving bell and is not affected by acoustic propagation paths and multipath effects, thus possessing higher accuracy and reliability.

[0065] After obtaining the diving bell's precise current depth, this depth information needs to be integrated into the three-dimensional coordinates obtained from acoustic positioning to achieve accurate coordinate correction. Acoustic positioning methods calculate three-dimensional coordinates through slant range measurement and angle estimation. The accuracy of the depth coordinate component is affected by various factors, including uncertainties in the sound velocity profile, multipath propagation effects, and the geometric configuration of the receiver array. Positioning errors in the vertical direction are typically greater than those in the horizontal direction. In contrast, optical depth measurement based on pressure sensors directly reflects the diving bell's true depth, is unaffected by the acoustic propagation path, and therefore has higher reliability.

[0066] The coordinate correction process employs a selective replacement strategy, retaining the horizontal coordinate components (x and y coordinates) obtained from acoustic positioning while replacing only the depth coordinate component (z coordinate) with the current precise depth obtained from optical measurements. This strategy is reasonable because acoustic positioning offers good accuracy in the horizontal direction, especially when the saturation diving support vessel is equipped with a multi-element hydrophone array, enabling high-precision horizontal positioning. However, the accuracy of acoustic methods is limited in the depth direction; therefore, introducing optical depth measurement can improve the overall positioning accuracy.

[0067] Specifically, the x and y coordinates are extracted from the obtained current 3D coordinates and combined with the obtained current precise depth to form the corrected 3D coordinates. The corrected 3D coordinates combine the horizontal accuracy advantage of acoustic positioning with the depth accuracy advantage of optical measurement, achieving complementary advantages of multimodal information.

[0068] In this embodiment, to ensure the smoothness and continuity of coordinate correction and avoid coordinate jumps caused by measurement noise, Kalman filtering smoothing is introduced during the correction process. The corrected 3D coordinates are input as new observations into an extended Kalman filter and fused with the acoustic positioning observations. The filter automatically assigns weights based on the noise characteristics of each observation and outputs the optimal state estimate. Since the noise variance of optical depth measurement is much smaller than that of acoustic depth measurement, the filter assigns greater weight to optical depth, making the final depth estimation mainly dependent on optical measurement, while the horizontal position estimation still mainly depends on acoustic positioning.

[0069] The corrected 3D coordinates are updated in the diving bell's state vector for subsequent trajectory prediction and rescue decisions. Over time, the corrected coordinates from multiple consecutive moments form a more accurate trajectory that accurately reflects the diving bell's motion in 3D space, providing rescuers with reliable target location information.

[0070] This embodiment introduces an optical-assisted positioning system, achieving precise correction of depth coordinates as the diving bell approaches the saturated diving support vessel, effectively compensating for the insufficient accuracy of pure acoustic positioning methods in the depth direction. The distance-adaptive optical system activation mechanism ensures that the optical equipment is activated only within the effective propagation range of the light signal, avoiding unnecessary power consumption and improving the system's practicality. The optical encoding and transmission scheme for depth information utilizes pulse interval time as the information carrier, achieving reliable transmission of depth data without complex communication protocols; the encoding method is simple, efficient, and highly resistant to interference. Depth measurement based on pressure sensors directly reflects the true depth of the diving bell, with high measurement accuracy. The multimodal coordinate fusion strategy leverages the respective advantages of acoustic positioning in the horizontal direction and optical measurement in the depth direction, improving the accuracy of the corrected three-dimensional coordinates and reducing depth positioning errors. Precise three-dimensional coordinate information allows rescue personnel to accurately determine the spatial position of the diving bell, especially crucial in the final approach phase for avoiding collisions and achieving safe docking. In summary, this system achieves an organic combination of acoustic and optical positioning modes, constructing a hierarchical positioning system that combines long-range acoustic tracking with short-range optical precision positioning. This provides full-process, high-precision target positioning capabilities for deep-sea emergency rescue, improving the safety and success rate of rescue operations.

[0071] In one embodiment of this invention, controlling the underwater acoustic communication device to transmit an initial continuous acoustic carrier signal at a first center frequency, and simultaneously controlling the emergency positioning device to transmit an initial positioning pulse signal at a second center frequency at a preset period, includes the following steps: S310. According to the preset time-frequency slice interleaving protocol, control the underwater acoustic communication device to transmit an initial continuous acoustic wave carrier signal with a first center frequency, wherein the time-frequency slice interleaving protocol is used to define the timing relationship and phase relationship between the initial continuous acoustic wave carrier signal and the initial positioning pulse signal. S320. According to the time-frequency slice interleaving protocol, control the emergency positioning device to transmit positioning pulse signals at specific phase zero-crossing moments of the continuous acoustic wave carrier signal.

[0072] In this embodiment, the time-frequency slicing interleaving protocol includes the center frequency, transmission power, and duration of the continuous acoustic carrier signal; it also includes the center frequency, pulse width, and transmission period of the positioning pulse signal, as well as the phase alignment parameter of the positioning pulse signal relative to the continuous acoustic carrier signal. The phase alignment parameter is used to determine the position of the transmission time of the positioning pulse signal within the period of the continuous acoustic carrier signal.

[0073] The time-frequency slice interleaving protocol is a complete set of parameters pre-configured in the diving bell emergency control system. This protocol defines in detail all the key technical parameters of the dual-channel acoustic signals and their cooperative working mechanism. The protocol first specifies the center frequency of the continuous acoustic carrier signal, usually selected in the range of 8 to 12 kHz. This frequency band has good propagation characteristics in the deep-sea environment, can maintain a relatively stable signal strength under bubble interference conditions, and avoids the main energy concentration frequency band of marine environmental noise.

[0074] The transmit power parameter is set based on the expected communication distance and underwater acoustic channel conditions, ensuring sufficient signal coverage without causing nonlinear distortion or excessive consumption of the diving bell's emergency battery power due to excessive power. The duration parameter defines the transmission time window for the continuous carrier signal. After the umbilical cable breaks, the continuous carrier signal needs to be transmitted continuously to maintain the communication link with the mother ship. The duration is usually set to match the endurance of the emergency battery.

[0075] The protocol also specifies the technical parameters of the positioning pulse signal in detail. The center frequency is selected in the range of 15 to 20 kHz, maintaining a frequency interval of 5 to 8 kHz with the communication frequency. This frequency separation design ensures that the two signals can be clearly distinguished in the frequency domain, avoiding mutual interference and spectral aliasing. The pulse width is set to 10 to 20 milliseconds, using a linear frequency modulation (LFM) pulse design. Within this time window, the frequency linearly scans from the start frequency to the end frequency. The LFM design can improve the pulse's time-bandwidth product, enhancing its resistance to multipath interference and ranging accuracy.

[0076] The transmission period parameter defines the time interval between two adjacent pulse transmissions, which is usually set to 1 to 2 seconds. This period needs to be balanced between update rate and power consumption. A shorter period can provide a higher position update rate, but will increase power consumption; a longer period can save energy, but will reduce the real-time performance of tracking.

[0077] In this embodiment, the phase alignment parameter precisely defines the position of the positioning pulse signal's transmission time within the period of the continuous acoustic carrier signal. It is typically set to the positive or negative zero-crossing point of the carrier signal, i.e., the moment when the phase is 0 degrees or 180 degrees. This phase-locked design establishes a strict timing correlation between the two signals, providing a reliable time reference for joint signal processing at the receiving end.

[0078] Specifically, the emergency control unit first analyzes the parameters in the time-frequency slice interleaving protocol, configures the signal generator of the underwater acoustic communication device, sets the carrier frequency, modulation method, and transmission power, and then initiates the transmission of continuous carrier signals. The signal generator can produce a highly stable, low-phase-noise sinusoidal carrier signal, which is amplified by a power amplifier and then drives the underwater acoustic transducer to radiate into the water.

[0079] The emergency positioning device integrates a high-precision phase detection circuit and a timing trigger circuit to achieve precise phase synchronization with a continuous acoustic carrier signal. The phase detection circuit acquires the real-time waveform of the carrier signal from the output of the underwater acoustic communicator via coupling. This waveform is then converted from analog to digital and fed into the digital signal processing unit. The digital signal processing unit uses a zero-crossing detection algorithm to monitor the phase state of the carrier signal in real time. Zero-crossing detection identifies the moment the signal crosses zero by judging the sign change of consecutive sampling points. When the signal changes from a negative value to a positive value, it is determined to be a positive zero-crossing; when it changes from a positive value to a negative value, it is determined to be a negative zero-crossing.

[0080] To improve the accuracy of zero-crossing detection, interpolation techniques are introduced into the algorithm. By performing linear or cubic spline interpolation on the sampled values ​​before and after the zero-crossing point, the estimation accuracy of the zero-crossing moment can be improved. Based on the phase alignment parameters in the time-frequency slice interleaving protocol, the phase detection circuit selects a specific type of zero-crossing point as the trigger reference. For example, if the parameter is set to 0-degree phase alignment, a positive zero-crossing point is selected; if it is set to 180-degree phase alignment, a negative zero-crossing point is selected. When a target zero-crossing point is detected, the phase detection circuit immediately sends a trigger signal to the timing trigger circuit.

[0081] After receiving the trigger signal, the timing trigger circuit performs a period determination. Since each cycle of the continuous carrier signal generates a zero-crossing point, and the transmission period of the positioning pulse signal is much longer than the carrier period, not every zero-crossing point requires triggering pulse transmission. The timing trigger circuit internally maintains a counter to record the number of carrier cycles elapsed since the last pulse transmission. When the counter reaches a preset period ratio, the next zero-crossing point triggers the actual pulse transmission.

[0082] During pulse transmission, the emergency positioning device's signal generator produces a linear frequency modulated pulse waveform according to protocol parameters. The start time of this waveform is precisely aligned with the zero-crossing point of the carrier signal. After power amplification, the pulse waveform drives the positioning transducer to radiate into the water, forming a dual-channel acoustic signal with a phase-locked relationship together with the continuous carrier signal. The advantage of this phase-locked mechanism is that the receiver can use the phase information of the continuous carrier signal as a time reference to coherently process the positioning pulse signal, improving the estimation accuracy of the pulse arrival time and thus enhancing the accuracy of ranging and positioning.

[0083] This embodiment achieves coordinated transmission and precise synchronization of underwater acoustic communication signals and positioning pulse signals through a time-frequency slicing interleaving protocol, laying a solid foundation for multimodal fusion positioning. The protocol-based parameter configuration allows for flexible adjustment of the technical parameters of the dual-channel acoustic signals according to different operating environments and task requirements, improving the system's adaptability and scalability. Frequency separation design effectively avoids mutual interference between the two signals, ensuring that communication quality and positioning accuracy are unaffected, maintaining good signal separation even in high reverberation environments. A phase-locking mechanism establishes a time synchronization relationship between the two signals. This strict timing correlation provides reliable prior information for joint signal processing at the receiving end, enabling accurate extraction of signal parameters even under low signal-to-noise ratio conditions. High-precision phase detection and triggering circuits ensure the accuracy of pulse transmission timing, with minimal time synchronization error. The coordinated transmission mechanism also improves the spectrum utilization efficiency of the emergency communication system. Through the combination of time-division multiplexing and frequency-division multiplexing, both voice communication and precise positioning functions are achieved within limited spectrum resources, avoiding the spectrum waste caused by requiring separate frequency bands for each function in traditional solutions. In summary, this provides a highly reliable and precise dual-modal acoustic signal source for emergency rescue using diving bells, improving communication and positioning performance in complex marine environments.

[0084] In one embodiment of this invention, the bistable stochastic resonance system model includes a barrier height parameter. The underwater mixed acoustic signal is input into the preset bistable stochastic resonance system model, and the barrier height parameter of the bistable stochastic resonance system model is adjusted according to the environmental background noise characteristic parameters to obtain an effective signal waveform with enhanced signal-to-noise ratio. This includes the following steps: S410. Calculate the environmental noise energy spectral density based on the characteristic parameters of environmental background noise. S420. Determine the target barrier height parameters based on the environmental noise energy spectral density and underwater mixed acoustic signals; S430. Adjust the barrier height parameter of the bistable stochastic resonance system model to the target barrier height parameter; S440: Using the underwater mixed acoustic signal as the input signal and the ambient background noise as the driving force, the signal is processed through a bistable random resonance system model to obtain an effective signal waveform with enhanced signal-to-noise ratio.

[0085] Calculating the environmental noise energy spectral density is a prerequisite for achieving adaptive stochastic resonance processing. It reflects the frequency domain distribution characteristics of noise energy, providing a quantitative basis for subsequent barrier height parameter optimization. In practice, power spectral density data is first extracted from the obtained environmental background noise characteristic parameters. This data is obtained by performing a fast Fourier transform on the underwater mixed acoustic signal and taking the square of the amplitude. The power spectral density curve contains energy distribution information for various frequency components from low to high frequencies, but it includes both pure noise and useful signal components, requiring separation processing.

[0086] Noise component identification employs spectral smoothing and baseline estimation techniques. By applying a sliding window averaging to the power spectral density curve, the influence of signal peaks can be suppressed, extracting a relatively flat noise floor. The width of the sliding window is typically set to include 10 to 20 frequency points, effectively smoothing random fluctuations without excessively obscuring spectral features. After identifying the noise floor, calculating the ambient noise energy spectral density requires integration within the frequency band of interest, typically defined as covering the first and second center frequencies and their surrounding extended region, for example, 5 to 25 kHz.

[0087] The integral calculation employs the trapezoidal rule or Simpson's rule for numerical integration, summing the noise power spectral density values ​​at all frequency points within the frequency band and multiplying by the frequency resolution to obtain the total noise energy within that band. The unit of environmental noise energy spectral density is joules per hertz (J / Hertz), which physically represents the noise energy per unit frequency interval; a higher value indicates a more severe noise environment. In cases where a broken umbilical cable generates numerous bubbles, bubble scattering and turbulent noise will increase the environmental noise energy spectral density.

[0088] To improve computational accuracy, the environmental noise energy spectral density is calculated using a multi-frame averaging method. Averaging the results across multiple consecutive time windows effectively suppresses the influence of transient noise, yielding a more stable and reliable estimate. The calculated environmental noise energy spectral density, as an important environmental characteristic parameter, directly reflects the noise level of the current underwater acoustic channel, providing quantitative input information for the adaptive adjustment of parameters in the stochastic resonance system.

[0089] The target barrier height parameter needs to be precisely matched based on the relative intensity of the input signal and noise. In practice, the effective signal component in the underwater mixed acoustic signal needs to be extracted first. This step is achieved through preliminary bandpass filtering. Two sets of bandpass filters are designed, corresponding to the first center frequency and the second center frequency, respectively. The passband width of the filters is set to ±20% of the center frequency, which can retain the main energy of the signal while suppressing out-of-band noise.

[0090] After passing the underwater mixed acoustic signal through these two sets of filters, coarse extraction results for the communication signal component and the positioning signal component are obtained, respectively. The root mean square (RMS) amplitude of each signal is calculated. The RMS amplitude reflects the effective strength of the signal. The calculation method involves squaring each sample point of the signal waveform, then averaging the values, and finally taking the square root. The larger of the two RMS amplitudes is taken as the signal amplitude of the effective signal component, representing the energy level of the strongest signal at that moment.

[0091] The integral value of the environmental noise energy spectral density within a preset frequency band is calculated. This integral value, obtained in the preceding steps, is directly used as a quantitative indicator of noise intensity. For ease of subsequent processing, the noise intensity is converted into an equivalent noise amplitude. The conversion method involves taking the square root of the noise energy to give it the same dimensions as the signal amplitude. Then, the ratio of the signal amplitude to the noise intensity is calculated; this ratio is the current signal-to-noise ratio, expressed in decibels (dB) in logarithmic form.

[0092] Based on the signal-to-noise ratio (SNR) range, it is divided into several levels, for example, five levels: extremely low SNR, low SNR, medium SNR, high SNR, and extremely high SNR. Each SNR level corresponds to different noise environment characteristics, requiring different barrier height parameters to achieve optimal resonance. A pre-defined mapping table stores the correspondence between SNR levels and target barrier height parameters. This mapping relationship was established through extensive simulation experiments and water tank tests, and the barrier height parameters were optimized for different SNR conditions.

[0093] The general rule is that a lower barrier height is needed under low signal-to-noise ratio (SNR) conditions to fully utilize noise energy and help the signal overcome the barrier, while a higher barrier height can be used under high SNR conditions to avoid nonlinear distortion caused by over-amplification. Based on the current SNR level, the corresponding target barrier height parameter is looked up in the mapping table. If the SNR value falls exactly on the boundary between the two levels, a linear interpolation method is used to calculate the precise parameter value, ensuring the continuity and smoothness of parameter adjustment.

[0094] Adjusting the barrier height parameter of a bistable stochastic resonance system model to a target value is a parameter configuration process that involves reconstructing the system's potential function. The potential function of a bistable stochastic resonance system is in the form of a fourth-order polynomial, and its mathematical expression is: The parameters a and b together determine the shape of the potential function. The potential barrier height is defined as the difference between the local maximum and local minimum of the potential function. By differentiating the potential function and setting the derivative to zero, the location of the extreme points can be obtained. The relationship between the potential barrier height and the parameters a and b is as follows: .

[0095] In practice, parameter b is preset to a fixed value, and the barrier height is controlled by adjusting parameter a. Based on the determined target barrier height parameter, the corresponding parameter a value is calculated using the aforementioned relationship. The calculation formula is as follows: After obtaining the new parameter 'a', update the coefficient configuration of the system potential function and reconstruct the potential function curve.

[0096] In this embodiment, the adjustment of the potential function also needs to consider the system's stability. The values ​​of parameters a and b must ensure that the potential function exhibits obvious bistable characteristics, i.e., the existence of two symmetrical potential wells and an intermediate potential barrier. Improper parameter configuration may lead to the potential function degenerating into a monostable or unstable state. To verify the rationality of the parameter configuration, the potential function is numerically verified after adjustment. The numerical distribution of the potential function in a typical interval is calculated, confirming the existence of two local minimum points and one local maximum point, and that the barrier height meets the target setting. The parameter adjustment process is conducted in real-time online. When the environmental noise characteristics change, the system can quickly respond and update the barrier height parameter, achieving adaptive optimization.

[0097] Processing the underwater mixed acoustic signal input to the regulated bistable stochastic resonance system is the core step in achieving signal enhancement. The dynamic behavior of the bistable stochastic resonance system is described by the Langevin equation, which includes deterministic force terms, stochastic force terms, and damping force terms, mathematically expressed as follows: Where x is the system state variable, is the damping coefficient, s(t) is the input signal, and n(t) is the noise driving force.

[0098] Specifically, the underwater mixed acoustic signal is used as the input signal s(t), and the ambient background noise is used as the driving force n(t), both jointly driving the dynamic evolution of the system. The ambient background noise is extracted by subtracting the effective signal component from the underwater mixed acoustic signal, or by directly using the identified noise component. The Langevin equation is solved numerically using a numerical method, with a time step set to one-tenth of the sampling period to ensure accurate capture of the system's dynamic response. Within each time step, the derivatives of the state variables are calculated based on the current system state and the input signal, and then the state variables are updated for the next time step.

[0099] The physical mechanism of stochastic resonance lies in the fact that when the input signal is too weak to enable the system to overcome the potential barrier on its own, the superimposed random noise fluctuations periodically provide additional energy, helping the system transition between two steady states. The frequency of this transition is synchronized with the frequency of the input signal, thus achieving nonlinear amplification of the signal. The system output state variable x(t) contains the amplified signal information, but it is also mixed with some noise components, requiring post-processing to extract a clean signal waveform.

[0100] Post-processing employs a low-pass filtering method, with the filter's cutoff frequency set slightly higher than the highest frequency component of the signal. This preserves the entire signal spectrum while suppressing high-frequency noise. After filtering, an effective signal waveform with enhanced signal-to-noise ratio is obtained. This waveform has a higher signal-to-noise ratio than the input signal, making the weak signal that was originally submerged in noise clearly discernible, thus creating favorable conditions for subsequent signal separation and parameter extraction.

[0101] This embodiment achieves effective signal enhancement in high-reverberation, high-noise environments through adaptive bistable stochastic resonance processing, utilizing environmental noise energy as a beneficial resource rather than a simple source of interference. Precise calculation of the environmental noise energy spectral density provides a quantitative basis for system parameter optimization, enabling the stochastic resonance processing to adaptively adjust according to the actual noise environment. A barrier height parameter mapping mechanism based on signal-to-noise ratio (SNR) levels enables intelligent and automated parameter configuration, maintaining optimal processing performance under different noise conditions without manual intervention. The nonlinear characteristics of the bistable system, under appropriate parameter configuration, can produce a signal amplification effect, especially under low SNR conditions, where the amplification gain is more significant, effectively compensating for the performance deficiencies of traditional linear processing methods in high-noise environments. The adaptive parameter adjustment mechanism allows the system to track changes in the noise environment in real time. When the distribution and intensity of the bubble cloud dynamically change, the system automatically updates the barrier height parameter to maintain the optimal resonance state, ensuring the stability and continuity of the signal enhancement effect. The improved SNR of the effective signal waveform after stochastic resonance processing enhances the accuracy of subsequent processing steps such as signal separation, Doppler analysis, and time delay measurement, reducing positioning errors. In summary, this provides strong support for the reliable operation of the diving bell emergency communication system in extremely harsh acoustic environments, enhances the system's environmental adaptability and anti-interference performance, and buys valuable time for deep-sea emergency rescue.

[0102] In one embodiment of this invention, the target barrier height parameter is determined based on the ambient noise energy spectral density and the underwater mixed acoustic signal, including the following steps: S510. Extract the effective signal component from the underwater mixed acoustic signal and calculate the signal amplitude of the effective signal component; S520. Calculate the integral value of the environmental noise energy spectral density within the preset frequency band to obtain the noise intensity; S530. Determine the signal-to-noise ratio level based on the ratio of signal amplitude to noise intensity; S540. According to the preset mapping table, find the target barrier height parameter corresponding to the signal-to-noise ratio level.

[0103] Extracting the effective signal components requires separating the dual-channel acoustic signal emitted by the diving bell from the mixed underwater acoustic signal containing multiple components. In practice, two sets of digital bandpass filters are first designed to extract the signal at the first and second center frequencies, respectively. The center frequency of the first set of bandpass filters is set to the transmission frequency of the underwater acoustic communication device, typically in the range of 8 to 12 kHz. The passband width is determined based on the modulation bandwidth of the continuous carrier signal, generally set to ±15% to ±25% of the center frequency. For example, if the center frequency is 10 kHz, the passband range is 8.5 to 11.5 kHz.

[0104] The filter order is chosen from 6th to 8th to obtain a sufficiently steep transition band characteristic, and the stopband attenuation is required to reach more than 50dB to ensure that out-of-band noise and interference components are fully suppressed. The design principle of the second set of bandpass filters is similar. The center frequency is set to the transmission frequency of the emergency positioning device, usually in the range of 15 to 20kHz. Since the positioning pulse signal adopts a linear frequency modulation design, its instantaneous frequency will change during the pulse duration. Therefore, the passband width needs to be set wider, usually ±25% to ±35% of the center frequency, to fully preserve the spectral components of the pulse signal.

[0105] The underwater mixed acoustic signal is passed through these two sets of filters to obtain two initially separated signal components, which correspond to the coarse extraction results of the continuous carrier signal and the positioning pulse signal, respectively. Time-domain analysis is performed on each signal to calculate its instantaneous amplitude envelope. Envelope extraction is achieved using the Hilbert transform method, which converts the real signal into a complex signal by constructing an analytical representation of the signal; the magnitude of the complex signal is the instantaneous amplitude envelope.

[0106] Statistical analysis is performed on the envelope curve, and its root mean square (RMS) value is calculated as the effective amplitude of the signal. The formula for calculating the RMS value is to square each sampling point of the envelope curve, sum the results, divide by the total number of sampling points, and finally take the square root. Because continuous carrier signals and positioning pulse signals have different time-domain characteristics—the envelope of a continuous carrier signal is relatively stable, while the envelope of a pulse signal exhibits periodic peaks—the RMS values ​​for both need to be calculated within different time windows. For continuous carrier signals, a time window containing at least 100 carrier cycles is selected to ensure the stability of the statistical results; for pulse signals, a time window containing at least 3 to 5 complete pulses is selected to capture both the peak characteristics of the pulse and smooth out inter-period fluctuations.

[0107] The root mean square (RMS) values ​​of the two signals are compared, and the one with the larger value is selected as the dominant signal. Its RMS value is the signal amplitude of the effective signal component. This selection strategy is based on the consideration that, in practical applications, the strength of the two signals may differ due to factors such as propagation path and transducer efficiency. Selecting the stronger signal as a reference can more accurately reflect the system's signal receiving capability and provide a reliable signal strength benchmark for subsequent signal-to-noise ratio (SNR) calculations.

[0108] Quantitative calculation of noise intensity is a key indicator for assessing the severity of the acoustic environment. This parameter is obtained by frequency domain integration of the ambient noise energy spectral density. In practice, a preset frequency band is first determined. This band needs to cover the entire spectral region of the dual-channel acoustic signal, as well as adjacent frequency bands that may cause interference. It is usually set from the first center frequency minus twice its passband width to the second center frequency plus twice its passband width. For example, if the first center frequency is 10kHz and the passband width is 2kHz, and the second center frequency is 18kHz and the passband width is 4kHz, then the preset frequency band range is 6 to 26kHz.

[0109] This wideband setting ensures that the integration calculation can include all noise components that may affect signal reception, including noise in the same frequency band that overlaps with the signal frequency, as well as adjacent frequency band noise that may enter the signal channel through nonlinear effects or filter sidelobe leakage. Data points within a preset frequency band are extracted from the obtained ambient noise energy spectral density curve, which is a discrete spectrum obtained through fast Fourier transform, with each frequency point corresponding to an energy spectral density value.

[0110] The integration calculation is implemented using numerical integration methods, the most common being the trapezoidal rule. This method divides the integration interval into several smaller intervals, approximating the area under the curve with the area of ​​a trapezoid within each smaller interval, and then summing all the trapezoidal areas to obtain the total integral value. Specifically, the calculation iterates through all frequency points within the preset frequency band. For two adjacent frequency points, the average of their energy spectral density values ​​is calculated, and then multiplied by the frequency interval between the two points to obtain the energy contribution of that smaller interval. The energy contributions of all smaller intervals are then summed to obtain the total noise energy within the preset frequency band.

[0111] The unit of the integral value is joules, and its physical meaning is the total energy of noise within that frequency band. This value directly reflects the intensity level of the noise environment. To facilitate comparison with the signal amplitude, the noise energy needs to be converted into an equivalent noise amplitude. The conversion method is to take the square root of the noise energy and divide it by the square root of the bandwidth to obtain the normalized noise intensity. This intensity value has the same dimensions as the signal amplitude and can be directly compared. In an environment where a dense bubble cloud is generated due to an umbilical cable break, the scattering and turbulence noise from the bubbles will increase the noise intensity. This high-intensity noise is the main reason why traditional signal processing methods fail and is the fundamental motivation for employing nonlinear processing techniques such as stochastic resonance.

[0112] The process of determining the signal-to-noise ratio (SNR) level maps continuously changing SNR values ​​to discrete level categories, facilitating subsequent table lookup operations and parameter selection. In practice, the ratio of signal amplitude to noise intensity is first calculated; this ratio is the linear SNR, reflecting the strength advantage of the useful signal relative to the noise. For ease of representation and processing, the linear SNR is converted to a logarithmic form in decibels. The conversion formula is: SNR (dB) equals 20 times the commonly used logarithmic signal amplitude divided by the noise intensity.

[0113] The advantage of logarithmic representation lies in its ability to compress large-scale numerical variations into smaller intervals, while also conforming to the logarithmic characteristics of the human perceptual system, facilitating intuitive understanding and engineering applications. After obtaining the signal-to-noise ratio (SNR) value in decibels, it is categorized into corresponding SNR levels according to a pre-defined classification standard. This classification standard is typically based on the difficulty of signal processing and the differences in required processing strategies. A typical five-level classification scheme is: extremely low SNR, low SNR, medium SNR, high SNR, and extremely high SNR.

[0114] Each level corresponds to different acoustic environment characteristics and signal detectability, requiring different stochastic resonance system parameters to achieve optimal processing results. Level determination employs a threshold comparison method, comparing the calculated signal-to-noise ratio (SNR) value sequentially with the boundary thresholds of each level to determine its corresponding level category.

[0115] In this embodiment, to avoid frequent jumps at level boundaries, a hysteresis interval can be introduced. Different thresholds are used for level increases and decreases. For example, increasing from a low level to a high level requires a signal-to-noise ratio (SNR) exceeding the upper boundary threshold by 1 dB, while decreasing from a high level to a low level requires an SNR falling below the lower boundary threshold by 1 dB. This hysteresis mechanism improves the stability of level determination and avoids frequent parameter adjustments caused by small fluctuations in SNR. The determined SNR level serves as key classification information, directly guiding subsequent parameter lookup and system configuration processes.

[0116] The target barrier height parameters can be found by establishing a correspondence between signal-to-noise ratio levels and optimal system parameters through a pre-defined mapping table. This mapping table is a parameter database built through extensive theoretical analysis, numerical simulation, and experimental testing, recording the barrier height parameter configurations that produce the best stochastic resonance effect under different signal-to-noise ratio conditions.

[0117] The table construction process first determines the reasonable range of values ​​for the barrier height parameter through theoretical analysis. Based on stochastic resonance theory, the barrier height needs to match the input noise intensity. Generally, the higher the noise intensity, the lower the optimal barrier height, so as to fully utilize noise energy to help the signal overcome the barrier. Then, numerical simulations are used to traverse the barrier height parameter space under different signal-to-noise ratio (SNR) conditions. For each set of parameter configurations, the output SNR gain is calculated, and the parameter with the largest gain is selected as the optimal configuration under that SNR condition. Finally, the effectiveness of the simulation results is verified, and the parameters are fine-tuned and optimized based on measured data to ensure that the expected results can be achieved in practical application environments.

[0118] The mapping table is indexed by signal-to-noise ratio (SNR) levels, with each level corresponding to one or more barrier height parameter values. In practice, based on the determined SNR level, the corresponding entry in the mapping table is searched, and the target barrier height parameter for that level is directly read. If the table stores multiple parameter values ​​for each level, for example, different parameters are configured for different signal types or frequency ranges, a secondary selection is needed based on the current signal characteristics to choose the most suitable parameter value.

[0119] When the signal-to-noise ratio (SNR) value falls exactly on the boundary between two levels, a linear interpolation method can be used to obtain more accurate parameter configuration. This involves weighted averaging of the parameter values ​​corresponding to the two levels based on the relative positions of the SNR values ​​within the two level intervals, yielding precise interpolated parameters. The interpolation formula is: the target parameter equals the lower-level parameter plus the SNR relative position multiplied by the parameter difference, where the SNR relative position equals the current SNR minus the upper bound of the lower level divided by the level interval width. The target barrier height parameter obtained through table lookup and interpolation considers both the current SNR level and ensures the continuity and smoothness of parameter adjustment, providing a precise parameter basis for the optimal configuration of the stochastic resonance system.

[0120] This embodiment achieves intelligent adaptive configuration of a bistable stochastic resonance system through a systematic signal feature extraction and parameter optimization process, improving the system's signal processing performance in complex noise environments. Accurate extraction of effective signal components provides a reliable foundation for signal strength assessment, while multi-band bandpass filtering and envelope analysis techniques ensure the accuracy of signal amplitude calculation. Frequency domain integration of noise intensity comprehensively reflects the noise level of the acoustic environment, and the wideband integration strategy avoids missing important noise components, providing complete noise characteristic information for signal-to-noise ratio (SNR) assessment. The SNR-based classification method discretizes the continuous parameter space, simplifying parameter selection complexity, while a hysteresis mechanism ensures the stability of the level determination, avoiding parameter jitter. The establishment of the mapping table integrates the results of theoretical analysis, numerical simulation, and experimental verification, ensuring the scientific validity and effectiveness of parameter configuration, achieving near-optimal processing results under different SNR conditions. The introduction of linear interpolation further improves the precision of parameter configuration, enabling the system to smoothly adapt to continuous changes in SNR and avoiding performance fluctuations that may be caused by step parameter adjustments. The overall solution achieves a fully automated process from signal analysis to parameter optimization, allowing for real-time adjustments to system configuration based on environmental changes without manual intervention. Under extremely low signal-to-noise ratio (SNR) conditions, the optimized stochastic resonance processing can improve the output SNR, making previously undetectable signals identifiable, thus creating conditions for subsequent location calculations and enhancing the reliability and robustness of the emergency rescue system in harsh environments.

[0121] In one embodiment of this invention, signal separation is performed on the effective signal waveform to obtain a continuous acoustic carrier signal and a positioning pulse signal, including the following steps: S610. Perform bandpass filtering on the effective signal waveform to extract the first frequency band signal corresponding to the first center frequency, thereby obtaining the separated continuous acoustic carrier signal; and S620. Extract the second frequency band signal corresponding to the second center frequency to obtain the separated positioning pulse signal.

[0122] The effective signal waveform after bistable random resonance processing contains two acoustic signal components with different frequencies, requiring frequency domain separation technology to independently extract the continuous acoustic carrier signal. In specific implementation, a first bandpass filter is designed specifically for signal extraction at the first center frequency. The center frequency of this filter is precisely set to the transmission frequency of the underwater acoustic communication device, typically within the range of 8 to 12 kHz. The passband width of the filter is determined based on the modulation characteristics of the continuous carrier signal. If frequency shift keying modulation is used, the passband width needs to cover the frequency offset range after modulation, generally set to ±20% of the center frequency. For example, when the center frequency is 10 kHz, the passband range is set to 8 to 12 kHz.

[0123] The filter design employs an elliptic filter, which, due to its steep attenuation characteristics, enables a rapid transition from the passband to the stopband within a narrow frequency range, effectively suppressing the positioning pulse signal component at the second center frequency. The filter order is selected from 8th to 10th, with passband ripple controlled within 0.5dB and stopband attenuation required to reach over 60dB, ensuring that the energy of the second frequency band signal is sufficiently suppressed below the noise level.

[0124] After the effective signal waveform passes through the bandpass filter, only the signal components within the first frequency band are retained in the frequency domain, while the energy of the second center frequency and its adjacent frequency bands is significantly attenuated. To further improve the separation quality, time-domain waveform detection is performed after frequency-domain filtering to analyze the envelope continuity and phase smoothness of the output signal. The continuous carrier signal should exhibit stable sinusoidal oscillation characteristics, with a relatively flat envelope and a linear phase change with time. If abnormal amplitude abrupt changes or phase jumps are detected, it indicates the possible presence of residual pulse interference or insufficient filter performance. In this case, secondary filtering is initiated, using a higher-order filter or cascaded filter structure to further suppress interference components. The separated continuous acoustic carrier signal retains the frequency, phase, and modulation information of the original carrier, providing a high-quality input signal for subsequent Doppler frequency shift analysis.

[0125] The extraction of the positioning pulse signal employs a similar frequency domain filtering method, but the filter parameters need to be specifically designed based on the spectral characteristics of the pulse signal. The center frequency of the second bandpass filter is set to the transmission frequency of the emergency positioning device, typically within the range of 15 to 20 kHz. Since the positioning pulse uses a linear frequency modulation design, its instantaneous frequency scans linearly from the starting frequency to the ending frequency within the pulse duration, with a frequency variation range of up to ±30% of the center frequency. Therefore, the passband width of the filter needs to be correspondingly expanded to fully preserve the spectral components of the pulse, typically set to ±35% to ±40% of the center frequency.

[0126] The filter design also employs a high-order elliptic filter, with an order of 8 to 10, and a stopband attenuation requirement of over 60 dB to ensure effective suppression of the continuous carrier signal component at the first center frequency. After the effective signal waveform passes through this filter, the output signal mainly contains the positioning pulse signal component, and the energy of the continuous carrier is suppressed to a negligible level.

[0127] In this embodiment, to further enhance the pulse signal and suppress residual noise, matched filtering is applied after bandpass filtering. The impulse response of the matched filter is designed as the time-reversed conjugate of the transmitted pulse waveform. When the received pulse signal passes through the matched filter, the output will produce a sharp correlation peak at the pulse arrival time. The peak amplitude is proportional to the input signal-to-noise ratio (SNR), while the noise component is smoothed out because it is uncorrelated with the matched filter. Matched filtering can provide additional SNR gain. After the combined processing of bandpass filtering and matched filtering, the separated positioning pulse signal has a clear pulse shape and accurate arrival time marking, providing a reliable signal basis for subsequent time delay measurement and ranging calculation.

[0128] This embodiment achieves high-quality separation of continuous carrier signals and positioning pulse signals through dual-channel frequency domain filtering, effectively solving the signal extraction challenge under conditions of multi-mode signal coexistence. Dedicated filters designed for different signal characteristics ensure the separation effect of each channel. Precise configuration of the passband width preserves the complete spectral components of the signal while maximally suppressing interference from other frequency bands. The steep transition band characteristic of the high-order elliptic filter enables clear separation of the two signals in the frequency domain, reducing mutual interference to well below the signal detection threshold. Time-domain waveform detection and secondary filtering mechanisms further improve the separation quality, ensuring the purity and stability of the output signal. The introduction of matched filtering technology enhances the peak characteristics of the pulse signal, improves the time resolution, and increases the measurement accuracy of the pulse arrival time. The two separated signals each retain the key characteristic parameters of the original signal: the frequency and phase information of the continuous carrier signal is complete, and the waveform and time delay information of the positioning pulse signal are accurate, providing high-quality data input for subsequent Doppler analysis and ranging calculations. In summary, effective separation of multimodal signals was achieved, meeting the performance requirements of real-time positioning systems, laying a solid foundation for the accurate implementation of multimodal fusion positioning algorithms, and improving the positioning accuracy and reliability of the system.

[0129] In one embodiment of this invention, Doppler frequency shift analysis is performed on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel, and the time difference of arrival is calculated on the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel. This includes the following steps: S710. Perform a short-time Fourier transform on the continuous acoustic wave carrier signal to obtain the time-spectrum diagram; S720. Identify the frequency peak in the time spectrum and calculate the frequency offset of the frequency peak relative to the first center frequency. S730. Calculate the radial velocity of the diving bell relative to the saturated diving support vessel based on the frequency offset and Doppler effect formula. S740: Perform pulse detection on the positioning pulse signal and identify the pulse arrival time; S750: Calculate the signal propagation delay based on the pulse arrival time and the preset transmission time; S760. Calculate the slant distance between the diving bell and the saturation diving support vessel based on the signal propagation delay and underwater sound speed.

[0130] The Short-Time Fourier Transform (SFT) is a time-frequency analysis method that can simultaneously reveal the characteristic distribution of a signal in both time and frequency dimensions, making it particularly suitable for analyzing non-stationary signals whose frequency varies over time. In practice, performing SFT on a separated continuous acoustic carrier signal first requires determining the type and length of the analysis window. Gaussian or Hanning windows are typically chosen. Gaussian windows offer optimal energy concentration in the time-frequency domain, achieving a theoretically optimal balance between time and frequency resolution; Hanning windows provide good sidelobe suppression, reducing spectral leakage.

[0131] The choice of window length requires a trade-off between time resolution and frequency resolution. Shorter windows offer better time resolution but poorer frequency resolution, while longer windows offer better frequency resolution but poorer time resolution. For continuous carrier signals with a center frequency of 8 to 12 kHz, the window length is typically set to include 10 to 20 carrier cycles. For example, with a center frequency of 10 kHz and a carrier cycle of 0.1 milliseconds, a window length of 1 to 2 milliseconds corresponds to 100 to 200 sampling points (assuming a sampling rate of 96 kHz). This window length ensures a frequency resolution of 50 to 100 Hz, sufficient to detect frequency changes caused by Doppler shift, while also ensuring a time resolution at the millisecond level, enabling the tracking of dynamic changes in the diving bell's motion.

[0132] The calculation process of Short-Time Fourier Transform (SFT) involves sliding a window function along the time axis and performing a Fast Fourier Transform on the windowed signal segment at each time point to obtain the spectrum at that moment. The window sliding step size is typically set to 25% to 50% of the window length, meaning that adjacent windows overlap by 50% to 75%. This overlap design improves the smoothness and continuity of the time-spectrum and avoids information loss due to window edge effects.

[0133] For each signal segment within a time window, the signal is first multiplied by a window function for weighting, and then subjected to a Fast Fourier Transform to obtain the complex spectrum at that moment. The amplitude of the complex spectrum is then taken to obtain the power spectrum at that moment. The power spectra of all time positions are arranged in chronological order to form a two-dimensional time-spectrum graph. The horizontal axis of the graph represents time, and the vertical axis represents frequency. The color or brightness of each point represents the energy intensity of the frequency component at that moment.

[0134] The frequency trajectory of the carrier signal can be clearly observed in the time-spectrum diagram. As the diving bell moves relative to the saturated diving support vessel, the received carrier frequency deviates from the transmitted frequency due to the Doppler effect, and the frequency trajectory shows a trend of change over time. If the diving bell is approaching the mother ship, the frequency will be higher than the transmitted frequency; if the diving bell is moving away from the mother ship, the frequency will be lower than the transmitted frequency. The time-spectrum diagram provides an intuitive visualization and accurate data foundation for subsequent frequency peak identification and frequency shift calculation.

[0135] Frequency peak identification is a crucial step in extracting the instantaneous frequency of the carrier signal from the time-spectrum graph. This process requires finding the most energetic frequency component in the spectrum of each time slice. In practice, each time column of the time-spectrum graph is processed, corresponding to the power spectrum distribution at a specific moment. First, the global maximum value is found in the power spectrum; the frequency corresponding to this maximum value is the dominant carrier frequency at that moment.

[0136] To improve the accuracy and robustness of peak identification, several optimization techniques are employed. The first is bandwidth limitation, restricting the search range to ±10% of the first center frequency. For example, with a center frequency of 10kHz, the search range is 9 to 11kHz. This limitation is based on the assumption that the diving bell's speed will not cause excessive Doppler shift. Bandwidth limitation effectively eliminates the influence of interference signals from other frequency bands, improving the reliability of peak identification.

[0137] The second step is peak verification. After finding the maximum value, we check whether the peak is significantly higher than the surrounding spectral background. We calculate the ratio of the peak to the average value of its neighboring frequency points. This ratio is required to be greater than a preset threshold. Only peaks that meet the significance condition are recognized as valid carrier frequencies.

[0138] Next is peak refinement. Since the spectrum obtained by the Fast Fourier Transform is discrete, the frequency resolution is limited by the window length. To obtain a more accurate peak frequency, parabolic interpolation or the centroid method is used for sub-resolution estimation. The parabolic interpolation method fits a parabola to the amplitude values ​​of the peak point and its two adjacent frequency points. The vertex of the parabola is the accurate peak frequency. This method can improve the accuracy of frequency estimation.

[0139] The peak identification process described above is repeated for each time slice of the time-spectrum graph, resulting in a series of instantaneous frequency values ​​that vary over time. These frequency values ​​constitute the frequency trajectory curve of the carrier signal. Then, the deviation of each instantaneous frequency relative to the first center frequency, i.e., the frequency offset, is calculated using the formula: frequency offset equals instantaneous frequency minus the first center frequency. The sign of the frequency offset indicates the direction of motion of the diving bell; a positive value indicates an upward shift in frequency, corresponding to the diving bell approaching the mother ship; a negative value indicates a downward shift in frequency, corresponding to the diving bell moving away from the mother ship. The magnitude of the frequency offset reflects the magnitude of the radial velocity, providing direct measurement data for subsequent velocity calculations.

[0140] The Doppler effect describes the frequency shift caused by the relative motion between a wave source and an observer. In underwater acoustic communication, when a diving bell moves relative to a saturation support vessel, the frequency of the received sound wave will shift. In practice, according to the physical principles of the Doppler effect, there is a definite mathematical relationship between the frequency shift and the radial velocity, described by the Doppler equation: ,in For receiving frequency, Let be the transmission frequency, and c be the speed of sound in water. This is the radial velocity (with approximation as the positive direction).

[0141] Rewrite the formula in the form of frequency offset: Further solve for the radial velocity: In actual calculations, the obtained frequency offset is substituted into the above formula, and an accurate underwater sound speed value needs to be obtained. Underwater sound speed is affected by environmental factors such as water temperature, salinity, and depth (pressure). Under standard seawater conditions (temperature 15℃, salinity 35‰, depth 0 meters), the sound speed is approximately 1500 m / s, but in actual deep-sea environments, the sound speed may vary within the range of 1450 to 1550 m / s.

[0142] To obtain accurate sound speed values, saturation diving support vessels are typically equipped with sound velocity profilers to measure the sound speed distribution at different depths in real time. Based on the estimated depth of the diving bell, the sound velocity profile is queried to obtain the corresponding sound speed value at that depth.

[0143] After calculating the sound velocity value based on the environmental parameters of the operating sea area, it is substituted into the radial velocity formula for calculation. Since the time-spectrum diagram provides frequency offsets for multiple consecutive moments, a series of instantaneous radial velocity values ​​can be calculated. These velocity values ​​reflect the temporal evolution of the diving bell's motion. To improve the stability and accuracy of the velocity estimation, the instantaneous velocity sequence is filtered and smoothed. Moving average filtering or Kalman filtering methods are used to suppress the influence of measurement noise, resulting in a smooth radial velocity curve.

[0144] Pulse detection requires reliably detecting the presence of a pulse signal and accurately pinpointing its timing and location within a noisy background. In practice, the separated positioning pulse signal is first subjected to energy detection to calculate its instantaneous energy or amplitude. The instantaneous energy is obtained by squaring the sampled signal values, and the instantaneous amplitude is obtained by extracting the signal envelope using a Hilbert transform. The advantage of energy detection lies in its ability to accumulate signal energy, improving the signal-to-noise ratio, and making it particularly suitable for detecting short-duration pulse signals.

[0145] The calculated instantaneous energy sequence is compared with a preset detection threshold. When the instantaneous energy exceeds the threshold, it is determined to be a possible pulse arrival. The detection threshold is set using a constant false alarm rate criterion, dynamically adjusted according to the statistical characteristics of the background noise to keep the false alarm probability at an acceptable level. The background noise power is obtained by statistically estimating the energy sequence during periods without signal, and the detection threshold is set as a multiple of the noise power.

[0146] Once the energy exceeds the threshold, a precise pulse arrival time estimation process is initiated. This process uses rising edge detection technology to track the rising process of the energy envelope and records the precise moment when the envelope rises from below the threshold to above the threshold, or records the moment when the envelope reaches its peak value as the pulse arrival time.

[0147] In this embodiment, to further improve the accuracy of time estimation, the output of a matched filter is used as the detection signal. The output of the matched filter will produce a sharp correlation peak at the pulse arrival time, and the peak position corresponds to the optimal pulse arrival time estimate. The time position of the correlation peak can be determined by finding the maximum value of the matched filter output. To achieve subsampling accuracy, a parabolic interpolation method is used to refine the peak position. The interpolation method uses the values ​​of the peak point and its neighboring sampling points to fit a continuous function, and the extreme point of the function is the precise peak time.

[0148] Since the positioning pulse signal is repeatedly transmitted at a preset period, the pulse detection process detects multiple pulses within a continuous time period, recording the arrival time of each pulse to form a pulse arrival time sequence. This sequence is not only used for ranging calculation at the current moment, but also for verifying the correctness of the detection results by analyzing the time interval between adjacent pulses. The correctly detected pulse interval should be consistent with the transmission period. If the detected interval deviates too much from the expected period, it indicates that there may be missed detections or false alarms, requiring adjustment of detection parameters or correction of results.

[0149] Signal propagation delay is the time it takes for a sound wave to travel from transmission to reception. This parameter directly reflects the distance the sound wave travels and is the basis for ranging calculations. In practice, the signal propagation delay is calculated as the difference between the pulse arrival time and the transmission time. The pulse arrival time is obtained through pulse detection, while the transmission time is precisely recorded through a phase-locked mechanism. Because the pulse transmission time of the emergency positioning device is strictly aligned with a specific phase zero-crossing point of the continuous sound wave carrier signal, the receiving end can reconstruct the time reference of the transmission time using the received continuous carrier signal.

[0150] The specific method involves phase tracking of the received continuous carrier signal to identify zero-crossing moments that match the phase alignment parameters of the transmitting end. These zero-crossing moments correspond to potential pulse transmission moments. Based on a preset transmission period, the zero-crossing moment that best matches the arrival time of the detected pulse is selected from the sequence of zero-crossing moments as the transmission moment of that pulse. Matching is determined based on the reasonableness of the time difference, which should correspond to a reasonable propagation distance range. For example, in deep-sea saturation diving operations, the distance between the diving bell and the mother ship is typically between 50 and 500 meters, corresponding to a propagation delay of 0.033 to 0.333 seconds. Only pairings with time differences falling within a reasonable range are considered valid.

[0151] After calculating the signal propagation delay, multipath correction is needed. In an underwater environment, sound waves may reach the receiver via multiple paths, such as reflection from the sea surface, seabed, or thermocline refraction, resulting in the detection of multiple pulse peaks at the receiver. The pulse corresponding to the direct path arrives first and has the shortest propagation delay. Therefore, the peak with the shortest delay is selected as the direct path signal, and the distance is calculated based on the delay corresponding to this peak. To improve the stability of the delay measurement, the propagation delays of multiple consecutive pulses are filtered and smoothed. Moving average or median filtering methods are used to suppress the influence of random errors, resulting in a stable and reliable delay estimate.

[0152] Slant range calculation is the process of converting signal propagation time delay into spatial distance, a conversion based on the fundamental physical laws of sound wave propagation. In practice, slant range equals signal propagation time delay multiplied by the underwater speed of sound, calculated using the following formula: Where R is the slant range and c is the underwater sound speed. This represents the signal propagation delay. The method for obtaining the underwater sound speed is the same as in the previous steps, requiring accurate measurement or calculation based on environmental parameters. Since the sound speed varies with depth, and the sound propagation path between the diving bell and the mother ship may cross different depth layers, strictly speaking, a layered integral calculation of the propagation path should be performed. However, in engineering applications, a simplified method using equivalent sound speed is typically employed, using the sound speed value corresponding to the average depth of the propagation path. This simplified method provides sufficient accuracy when the depth variation is not too drastic.

[0153] The calculated slant range reflects the straight-line distance between the diving bell and the saturation diving support vessel. This distance is in three-dimensional space and is not equal to the horizontal distance or depth difference. To further improve ranging accuracy, a multi-pulse averaging method is used, which weights the ranging results of multiple consecutive pulses. The weights are determined based on the signal-to-noise ratio or correlation peak intensity of each measurement, with measurements having higher signal-to-noise ratios receiving greater weight. Averaging effectively suppresses random errors and improves the stability and reliability of ranging. The obtained slant range data, as an important observation, is input along with the radial velocity data into the subsequent data fusion algorithm to jointly determine the three-dimensional position and motion state of the diving bell.

[0154] This embodiment achieves dual-parameter synchronous measurement of the diving bell's motion state through time-frequency analysis and pulse detection technology, providing complete observation data for multimodal fusion positioning. The short-time Fourier transform method effectively reveals the time-varying frequency characteristics of the carrier signal, and the visualization of the time-spectrum makes Doppler frequency shift identification more intuitive and accurate, enabling the detection of minute velocity changes. Peak identification and parabolic interpolation techniques improve the accuracy of frequency measurement and velocity measurement, meeting the needs of emergency rescue for motion state monitoring. The advantage of Doppler velocity measurement lies in its ability to reflect the diving bell's motion trend in real time, providing crucial information for trajectory prediction and rescue decision-making. Pulse detection employs a strategy combining energy detection and matched filtering, reliably detecting pulse signals even under low signal-to-noise ratio conditions. The phase-locked mechanism provides a transmission time reference, eliminating the need for additional time synchronization equipment for delay measurement and simplifying system implementation complexity. Multipath suppression and smoothing filtering techniques effectively improve the stability of ranging. Dual-parameter synchronous measurement achieves the coordinated acquisition of position and velocity information; the two types of observations are strictly synchronized in time, providing a consistent data foundation for subsequent Kalman filter fusion. In summary, by fully leveraging the respective advantages of continuous carrier signals and pulse signals, the carrier signal provides high-precision velocity measurement, while the pulse signal provides accurate distance measurement. The complementary integration of the two enhances the overall performance of the positioning system, providing reliable target tracking capabilities for deep-sea emergency rescue.

[0155] In one embodiment of this invention, radial velocity and slant distance are fused to obtain fused data, and the fused data is then processed to obtain the diving bell's trajectory and current three-dimensional coordinates, including the following steps: S810. Construct an extended Kalman filter, using the three-dimensional position coordinates and three-dimensional velocity vector of the diving bell as state variables to establish a state-space model. S820: Extended Kalman filter with radial velocity as the velocity observation input and slant range as the distance observation input; S830. Using an extended Kalman filter, predict the prior state estimate for the current time step based on the state estimate and state transition matrix from the previous time step. S840. Using the extended Kalman filter, calculate the Kalman gain based on the current observation and prior state estimate, and update the current posterior state estimate. S850. Extract the position coordinate components from the posterior state estimate and use them as the current three-dimensional coordinates of the diving bell. S860: Connect the current three-dimensional coordinates of multiple consecutive moments in chronological order to construct the motion trajectory curve of the diving bell.

[0156] The Extended Kalman Filter (EKF) is a recursive filtering algorithm suitable for state estimation of nonlinear systems. It transforms the nonlinear problem into a locally linear problem by linearizing it at the current estimation point. In its implementation, the state variables are first defined by combining the three-dimensional position coordinates and three-dimensional velocity vector of the diving bell into a six-dimensional state vector. ,in The position coordinates of the diving bell in three-dimensional space are usually represented by a local coordinate system with the saturation diving support vessel as the origin, the x-axis pointing towards the bow, the y-axis pointing towards the side of the vessel, and the z-axis pointing vertically downwards towards the seabed. This represents the velocity components of the diving bell along the three coordinate axes.

[0157] The state-space model consists of two parts: the state transition equation and the observation equation. The state transition equation describes the evolution of the state variables over time. In the absence of external forces, the motion of the diving bell can be approximated as uniform motion or slowly changing motion influenced by ocean currents. The state transition equation can be expressed as follows: ,in Here is the state transition matrix. The noise is a process noise that follows a zero-mean Gaussian distribution.

[0158] The state transition matrix is ​​a 6×6 matrix whose structure reflects the relationship between position and velocity. The change in position coordinates is equal to the velocity multiplied by the time step, i.e.: ; The velocity component remains constant or changes slowly over a short period of time, that is: These relationships are written in matrix form, with the diagonal block of the state transition matrix being the identity matrix, and the partial derivatives of the position with respect to velocity in the off-diagonal blocks representing the time steps. .

[0159] Process noise covariance matrix The uncertainty in the state transition process is described, mainly stemming from factors such as ocean current disturbances and model simplification errors. The diagonal elements of the covariance matrix are set based on experience or measured data. The observation equation describes the relationship between the observed quantities and the state variables, with the radial velocity observation... It equals the projection of the velocity vector onto the direction the diving bell points towards the saturation diving support vessel, mathematically expressed as: The slant distance observation R is equal to the magnitude of the position vector, and its mathematical expression is: .

[0160] Both observation equations are nonlinear and need to be linearized at the current state estimate. This involves calculating the Jacobian matrix of the observation matrix, where each element represents the partial derivative of the observation with respect to the state variable. The observation noise covariance matrix is ​​also required. The uncertainties in the observation process are described. The radial velocity observation noise is set according to the measurement accuracy, and the slant range observation noise is set according to the ranging accuracy. After establishing a complete state-space model, the state estimates and error covariance matrix of the filter are initialized. The initial state estimates can be roughly estimated based on the first observation data, and the initial error covariance matrix is ​​set to a large value to reflect the initial uncertainty.

[0161] The input of the observations is the data source for the extended Kalman filter's state update, a process that requires organizing measurement data from different sensors into a standard observation vector form. In practice, the obtained radial velocity is used as the velocity observation, and the obtained slant range is used as the range observation; together, they constitute a two-dimensional observation vector. The input of the observations needs to be synchronized with the recursion period of the filter, typically set to match the transmission period of the positioning pulse signal, for example, updating once every 1 to 2 seconds. Since the radial velocity and slant range measurements are performed synchronously, the two observations correspond to the diving bell state at the same moment, and therefore can be combined into a single observation vector for joint processing.

[0162] In practical implementation, it is also necessary to verify the validity of the observation data and remove obviously abnormal measurements. Anomaly detection uses statistical testing methods to calculate the residual, i.e., the innovation, between the current observation and the predicted observation. If the magnitude of the innovation exceeds a preset threshold, the observation is determined to be an anomaly and will not participate in the filter update process. Anomalies may originate from multipath interference, transient noise, or accidental failures of the signal processing algorithm. Anomaly detection can improve the robustness of the filter and prevent state estimation divergence caused by abnormal observations.

[0163] For the observation data that has passed the validity test, time alignment processing is also required. Due to the certain delay in signal processing and data transmission, the actual time corresponding to the observation data may be slightly earlier than the time when the data arrives at the filter. The timestamp needs to be corrected according to the processing delay to ensure that the observation data and the state estimation are strictly in time correspondence.

[0164] In this embodiment, the observation input also needs to be dynamically adjusted in conjunction with the observation noise covariance matrix. When a signal quality degradation is detected (e.g., a decrease in signal-to-noise ratio), the observation noise variance is appropriately increased to reduce the confidence in the current observation, making the filter rely more on state prediction. When the signal quality is good, the observation noise variance is decreased to increase the confidence in the observation, enabling the filter to quickly track state changes. This adaptive noise adjustment mechanism improves the filter's adaptability in complex environments and ensures the stability and accuracy of state estimation.

[0165] State prediction is the first step in the recursive process of the extended Kalman filter. This step uses the system's dynamic model to make a prior estimate of the state at the current moment. In practice, it is based on the posterior state estimate from the previous moment. and state transition matrix The prior state estimate at the current time is calculated using the state transition equation. The calculation process actually predicts the current position and velocity based on the previous position and velocity, following a uniform motion model.

[0166] Simultaneously with state prediction, the error covariance matrix also needs to be updated. The formula for calculating the prior error covariance matrix is ​​as follows: ,in Let be the posterior error covariance matrix of the previous time step. This is the process noise covariance matrix. This calculation reflects the propagation and growth of state uncertainty. Due to model errors and process noise in the state transition process, the prior error covariance will be greater than the posterior error covariance, indicating an increase in prediction uncertainty.

[0167] The error covariance matrix is ​​a 6×6 symmetric positive definite matrix. Diagonal elements represent the variance of each state variable, and off-diagonal elements represent the covariance between state variables. The error covariance matrix quantifies the confidence level of the state estimate, providing weights for subsequent Kalman gain calculations. State prediction can also be used to generate predicted observations. Substituting prior state estimates into the observation equations, the predicted radial velocity and slant range are calculated. These predicted observations are then compared with actual observations to calculate innovation and assess the reasonableness of the observation data. If the deviation between actual and predicted observations is too large, it may indicate an inaccurate state model or abnormal observation data, requiring model correction or data removal. The state prediction step demonstrates the predictive capability of the filter. Even when observation data is temporarily missing, the filter can maintain the continuity of the state estimate based on the dynamic model, providing continuous positioning information for the system.

[0168] State update is the second step in the recursive process of the extended Kalman filter. This step uses the observation data at the current time to correct the prior state estimate, resulting in a more accurate posterior state estimate. In practice, the Jacobian matrix of the observation matrix is ​​first calculated. This matrix is ​​the partial derivative matrix of the observation equation with respect to the state variables, in the prior state estimates. Linearization calculations are performed at this point. For the radial velocity observation equation, the calculation of partial derivatives involves complex algebraic operations, such as... Other partial derivatives are calculated similarly; for the slant range observation equation, the partial derivatives are: ; The partial derivatives with respect to the velocity components are zero. All partial derivatives are combined into a 2×6 observation matrix. .

[0169] Then the innovation is calculated, which is the difference between the actual observation and the predicted observation. ,in For the observation equation, The predicted observation is calculated based on the prior state estimate. The innovation reflects new information provided by the observation data. If the innovation is close to zero, it means that the prediction is consistent with the observation and the state estimate is relatively accurate. If the innovation is large, it means that there is a deviation between the prediction and the observation, and the state estimate needs to be corrected significantly.

[0170] Next, we calculate the new information covariance matrix. This matrix integrates the uncertainties of state prediction and observation noise, and is used to normalize the innovation and evaluate its statistical significance. Then, the Kalman gain is calculated. The Kalman gain is a 6×2 matrix that reflects the degree of confidence in the observed data. Its value depends on the relative magnitudes of the prior error covariance and the observation noise covariance. When the prior error covariance is large and the observation noise is small, the Kalman gain is large, and the state update relies more on the observed data; conversely, when the prior error covariance is small and the observation noise is large, the Kalman gain is small, and the state update relies more on the predicted values.

[0171] Calculating the Kalman gain involves matrix inversion, requiring the numerical stability of the innovation covariance matrix to avoid computational errors caused by ill-conditioned matrices. Finally, the state estimate is updated based on the Kalman gain and the innovation. Simultaneously update the error covariance matrix. ,in The identity matrix is ​​used. The updated state estimate and error covariance matrix are the posterior estimates for the current moment. This estimate integrates the prediction information from the dynamic model and the measurement information from the observation data. It minimizes the estimation variance through optimal weighting and is the best estimate of the diving bell's state at the current moment.

[0172] Location coordinate extraction is the process of converting the state vector output by the filter into directly usable positioning information. In specific implementations, this is achieved by extracting the obtained posterior state estimates. Extract the first three components, namely the position coordinate components. The x-coordinate represents the current three-dimensional coordinates of the diving bell. This coordinate system is defined in a local coordinate system with the saturation diving support vessel as the origin. The x-coordinate represents the distance of the diving bell relative to the bow direction, with a positive value indicating it is forward of the bow and a negative value indicating it is aft of the stern. The y-coordinate represents the distance relative to the side of the ship, with a positive value indicating it is on the starboard side and a negative value indicating it is on the port side. The z-coordinate represents the depth, with a positive value indicating it is below the surface and a larger value indicating greater depth.

[0173] If it is necessary to transform the coordinates to other coordinate systems, such as geographic or geodetic coordinate systems, a coordinate transformation is required. Based on the position and heading information of the saturation diving support vessel, local coordinates are converted to global coordinates through rotation and translation transformations. The extraction of position coordinates also includes the extraction of uncertainty information, from the posterior error covariance matrix. Extract the submatrix corresponding to the position components, i.e. the first 3×3 block. The diagonal elements of this submatrix are the variances of each coordinate component. Taking the square root of the variances gives the standard deviation, which reflects the uncertainty of the position estimation.

[0174] Location uncertainty information is crucial for assessing positioning quality and guiding rescue decisions. High uncertainty indicates low positioning accuracy, requiring caution in using the results; conversely, low uncertainty indicates high positioning accuracy, allowing for reliable reliance on the information. The extracted current 3D coordinates serve as the best estimate of the diving bell's position at that moment, and are output to other modules of the emergency rescue system. For example, they can be displayed on a monitoring interface for rescue personnel to view, or input into a path planning algorithm to calculate the optimal rescue route.

[0175] Constructing the motion trajectory involves connecting discrete location points into a continuous curve, which visually displays the diving bell's historical movement path and current trend. In practice, the current three-dimensional coordinates at multiple consecutive moments are stored chronologically in the trajectory database, with each coordinate point accompanied by a timestamp recording the measurement time. As time progresses, the trajectory database accumulates more and more coordinate points, forming a spatiotemporal sequence of the diving bell's motion.

[0176] To construct a smooth trajectory curve, interpolation is performed on discrete coordinate points. Common interpolation methods include linear interpolation, cubic spline interpolation, or Bézier curve fitting. Linear interpolation is simple and fast, connecting adjacent coordinate points with straight line segments, making it suitable for real-time display. Cubic spline interpolation generates a smoother curve with continuous second derivatives, more realistically reflecting the diving bell's trajectory, but it has higher computational complexity. In practical applications, the appropriate interpolation method can be selected based on display requirements: linear interpolation is used for real-time monitoring interfaces, while cubic spline interpolation is used for post-analysis and report generation.

[0177] The trajectory curve can be visualized in three-dimensional space, using different colors or line styles to represent trajectories over different time periods. For example, cool colors represent earlier trajectories, warm colors represent more recent trajectories, and bold lines represent trajectories near the current position. The trajectory curve can also be projected onto a two-dimensional plane for display. For example, an xy projection can be displayed on a horizontal plane to show the horizontal drift path of the diving bell; an xz or yz projection can be displayed on a vertical cross-section to show the depth changes of the diving bell.

[0178] By analyzing the shape and trend of the trajectory curve, the movement pattern of the diving bell can be inferred, such as whether it is directionally drifting due to stable ocean currents or lingering in a certain area. The trajectory curve can also be used for trajectory prediction; by fitting a mathematical model of the trajectory curve, the future position can be extrapolated, providing predictive information for rescue decisions. The trajectory data can also be exported as a standard format file for post-event analysis, report writing, or sharing with other systems, providing complete location records and movement trajectory information for the entire emergency rescue process.

[0179] This embodiment achieves optimal fusion of multi-source observation data through an extended Kalman filter, effectively overcoming the limitations of single observation methods and improving the accuracy and reliability of diving bell positioning. The establishment of a state-space model unifies position and velocity estimation within the same framework, achieving a complete description of the motion state. The six-dimensional state vector comprehensively reflects the three-dimensional position and velocity of the diving bell, providing rich information for rescue decision-making. The joint processing of radial velocity and slant range observations leverages their respective advantages. Radial velocity provides direct measurement of motion direction and velocity magnitude, while slant range provides distance information. Their complementary fusion effectively constrains the solution space of state estimation, avoiding estimation ambiguity caused by single observations. The recursive processing framework of the extended Kalman filter enables real-time updates of the state estimate. Each time new observation data is acquired, filtering and updating are performed immediately, outputting the optimal state estimate for the current moment, meeting the real-time requirements of emergency rescue. The adaptive calculation of the Kalman gain dynamically adjusts the observation weights based on prediction uncertainty and observation noise. It increases dependence on observations when observation quality is good and increases dependence on model predictions when observation quality is poor. This adaptive mechanism improves the robustness of the filter, enabling it to operate stably in complex and variable marine environments. The recursive update of the error covariance matrix provides a quantitative assessment of the uncertainty in state estimation, offering a basis for the reliability analysis of positioning results. When the uncertainty exceeds a threshold, an alarm can be triggered, reminding rescuers to use positioning information with caution. After data fusion processing, the three-dimensional positioning accuracy of the diving bell is improved. The construction of the motion trajectory provides rescuers with intuitive visualization information; by observing the trajectory curve, they can quickly understand the diving bell's motion history and current trend. The trajectory prediction function enables rescuers to plan rescue routes in advance and optimize rescue strategies. In summary, a complete processing flow from raw observation data to precise positioning results has been achieved, providing high-precision and high-reliability target positioning capabilities for deep-sea emergency rescue, improving the efficiency and success rate of rescue operations, and providing key technical support for ensuring the safety of divers.

[0180] Reference Figure 3 This application also provides a multimodal fusion positioning system for emergency communication of diving bells, comprising: The diving bell is equipped with an underwater acoustic communication device, an emergency positioning device, and a strobe light; A saturation diving support vessel, connected to a diving bell, equipped with an acoustic receiver and underwater observation equipment.

[0181] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described multimodal fusion positioning method for emergency communication of diving bells.

[0182] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0183] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0184] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0185] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0186] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0187] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0188] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0189] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0190] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A multimodal fusion positioning method for emergency communication of diving bells, characterized in that, This method is applied to saturation diving support vessels, which are connected to a diving bell via an umbilical cable. The diving bell is equipped with an underwater acoustic communication device and an emergency positioning device. The saturation diving support vessel is equipped with an acoustic receiver. The method includes: In response to the detection of an umbilical cable breakage signal, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic wave carrier signal at a first center frequency, while the emergency positioning device is controlled to transmit an initial positioning pulse signal at a second center frequency at a preset period. The transmission time of the initial positioning pulse signal is aligned with the preset phase zero-crossing point of the initial continuous acoustic wave carrier signal, forming a dual-channel acoustic signal with a phase-locked relationship. Underwater mixed acoustic signals are acquired by an acoustic receiver, and the spectrum of the underwater mixed acoustic signals is analyzed to obtain the characteristic parameters of the environmental background noise. The underwater mixed acoustic signal is input into a preset bistable stochastic resonance system model, and the barrier height parameter of the bistable stochastic resonance system model is adjusted according to the environmental background noise characteristic parameters to obtain the effective signal waveform with enhanced signal-to-noise ratio. The effective signal waveform is separated to obtain a continuous acoustic carrier signal and a positioning pulse signal; Doppler frequency shift analysis was performed on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel, and the time difference of arrival was calculated on the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel. The radial velocity and slant distance data are fused to obtain fused data, and the fused data is then solved to obtain the diving bell's motion trajectory and current three-dimensional coordinates.

2. The method according to claim 1, characterized in that, The diving bell is also equipped with a strobe light, and the saturation diving support vessel is also equipped with underwater observation equipment. After calculating the diving bell's trajectory and current three-dimensional coordinates, the method also includes: Determine if the slope distance is less than a preset distance threshold; When the slant distance is less than the preset distance threshold, the strobe light is controlled to read the depth information collected by the pressure sensor inside the diving bell, and the depth information is mapped to the pulse emission interval time of the strobe light, so that the strobe light flashes according to the pulse emission interval time; The underwater observation equipment captures the light signal emitted by the strobe light, and calculates the time difference between adjacent light pulses based on the light signal. The current precise depth of the diving bell is obtained by reverse decoding based on the time difference. The depth coordinate component in the current 3D coordinates is replaced with the current accurate depth to obtain the corrected 3D coordinates.

3. The method according to claim 1, characterized in that, Controlling the underwater acoustic communication device to transmit an initial continuous acoustic carrier signal at a first center frequency, and simultaneously controlling the emergency positioning device to transmit an initial positioning pulse signal at a second center frequency at a preset period, including: According to the preset time-frequency slice interleaving protocol, the underwater acoustic communication device is controlled to transmit an initial continuous acoustic wave carrier signal at the first center frequency. The time-frequency slice interleaving protocol is used to define the timing and phase relationship between the initial continuous acoustic wave carrier signal and the initial positioning pulse signal. According to the time-frequency slice interleaving protocol, the emergency positioning device is controlled to transmit positioning pulse signals at specific phase zero-crossing points of the continuous acoustic carrier signal.

4. The method according to claim 1, characterized in that, The bistable stochastic resonance system model includes a barrier height parameter. Underwater mixed acoustic signals are input into the preset bistable stochastic resonance system model, and the barrier height parameter of the bistable stochastic resonance system model is adjusted according to the environmental background noise characteristic parameters to obtain an effective signal waveform with enhanced signal-to-noise ratio, including: Calculate the environmental noise energy spectral density based on the characteristic parameters of environmental background noise; The target barrier height parameters are determined based on the environmental noise energy spectral density and the underwater mixed acoustic signal. The barrier height parameter of the bistable stochastic resonance system model is adjusted to the target barrier height parameter; Using underwater mixed acoustic signals as input signals and ambient background noise as driving force, the signals are processed through a bistable stochastic resonance system model to obtain an effective signal waveform with enhanced signal-to-noise ratio.

5. The method according to claim 4, characterized in that, Based on the environmental noise energy spectral density and the underwater mixed acoustic signal, the target barrier height parameters are determined, including: Extract the effective signal components from the underwater mixed acoustic signal and calculate the signal amplitude of the effective signal components; The noise intensity is obtained by calculating the integral value of the environmental noise energy spectral density within a preset frequency band. The signal-to-noise ratio level is determined based on the ratio of signal amplitude to noise intensity. Based on the preset mapping table, find the target barrier height parameter corresponding to the signal-to-noise ratio level.

6. The method according to claim 1, characterized in that, The effective signal waveform is separated to obtain a continuous acoustic carrier signal and a positioning pulse signal, including: The effective signal waveform is bandpass filtered to extract the first frequency band signal corresponding to the first center frequency, resulting in the separated continuous acoustic carrier signal; and The second frequency band signal corresponding to the second center frequency is extracted to obtain the separated positioning pulse signal.

7. The method according to claim 6, characterized in that, Doppler frequency shift analysis was performed on the continuous acoustic carrier signal to obtain the radial velocity of the diving bell relative to the saturated diving support vessel. The time difference of arrival was calculated from the positioning pulse signal to obtain the slant distance between the diving bell and the saturated diving support vessel, including: A short-time Fourier transform is performed on the continuous acoustic wave carrier signal to obtain the time-spectrum diagram; Identify frequency peaks in the time-spectrum graph and calculate the frequency offset of the frequency peaks relative to the first center frequency. The radial velocity of the diving bell relative to the saturated diving support vessel is calculated based on the frequency offset and Doppler effect formula. Perform pulse detection on the positioning pulse signal to identify the pulse arrival time; Calculate the signal propagation delay based on the pulse arrival time and the preset transmission time; Calculate the slant distance between the diving bell and the saturation diving support vessel based on the signal propagation delay and underwater sound speed.

8. The method according to claim 1, characterized in that, Data fusion of radial velocity and slant distance is performed to obtain fused data. This fused data is then processed to obtain the diving bell's trajectory and current three-dimensional coordinates, including: An extended Kalman filter is constructed, and the three-dimensional position coordinates and three-dimensional velocity vector of the diving bell are used as state variables to establish a state-space model; Extend the Kalman filter by using radial velocity as the velocity observation input and slant range as the distance observation input. By using an extended Kalman filter, the prior state estimate for the current time step is predicted based on the state estimate and state transition matrix from the previous time step. By using the extended Kalman filter, the Kalman gain is calculated based on the current observation and prior state estimate, and the posterior state estimate is updated to the current time. The position coordinate components are extracted from the posterior state estimate and used as the current three-dimensional coordinates of the diving bell. By connecting the current three-dimensional coordinates of multiple consecutive moments in chronological order, the motion trajectory curve of the diving bell is constructed.

9. A multimodal fusion positioning system for emergency communication of a diving bell, applied to the method described in any one of claims 1 to 8, characterized in that, include: The diving bell is equipped with an underwater acoustic communication device, an emergency positioning device, and a strobe light; A saturation diving support vessel, connected to a diving bell, equipped with an acoustic receiver and underwater observation equipment.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multimodal fusion positioning method for emergency communication of diving bells as described in any one of claims 1 to 8.

Images