Parameter intelligent adjustment method of depth sounder and storage medium

Abstract

The application relates to the technical field of parameter intelligent adjustment of a depth sounding sonar, and discloses a parameter intelligent adjustment method of a depth sounding sonar and a storage medium, wherein the method comprises the following steps: receiving original data of a current period of a multi-beam depth sounding sonar, calculating a quality factor based on echo data of each beam and judging the validity of the data according to a mean value; if the data is valid, using a one-dimensional Kalman filter set to predict a depth value and a variance to calculate a depth threshold, and adjusting a detection range, a form of a transmitted signal and a transmission power according to the depth threshold; and if the data is invalid, constructing a hidden Markov model based on beam echo data and using a Viterbi algorithm to reconstruct a depth sounding sequence to reset the threshold. The application can realize closed-loop autonomous control of multi-parameter cooperation, and improve the operation stability and data quality of an unmanned platform depth sounding sonar in a complex environment.

Description

Technical Field

[0001] This invention relates to the field of intelligent parameter adjustment technology for depth sounding sonar, and particularly to an intelligent parameter adjustment method and storage medium for depth sounding sonar. Background Technology

[0002] Multibeam bathymetry sonar is a core piece of equipment for acquiring high-precision seabed topographic data, widely used in marine resource exploration, waterway surveying, and underwater engineering. Its performance is highly dependent on the coordinated settings of several key parameters, including depth threshold, measurement range, transmission signal type, and transmission power. In traditional manned vessel operations, experienced operators can manually adjust these parameters based on real-time water depth changes, topographic relief characteristics, and hydrological conditions to ensure continuous acquisition and high-quality output of bathymetry data. With the rapid development of unmanned marine platforms, the demand for automated applications of multibeam bathymetry sonar is increasing.

[0003] Existing technologies already include solutions for automatically adjusting a single parameter based on preset logic, such as simply switching the range based on the echo intensity threshold, or selecting the transmission pulse width according to a fixed water depth range.

[0004] However, in the face of complex and ever-changing underwater environments, existing technologies cannot autonomously determine the validity of depth measurement data and cannot form an adaptive control loop, which affects the operational stability of unmanned platforms under complex working conditions. Summary of the Invention

[0005] Based on this, it is necessary to address the problem of intelligent parameter adjustment of existing depth sounding sonars, and propose an intelligent parameter adjustment method and storage medium for depth sounding sonars.

[0006] A method for intelligent parameter adjustment of a depth-sounding sonar, the method comprising: Receive raw data from the current cycle of the multibeam echo sounder; wherein the raw data includes at least the echo sequence of each beam, the received channel signal, and the echo data of each beam. The quality factor of each beam is calculated based on the echo data of each beam. The mean of all beam quality factors is used to determine whether the current bathymetry data is a valid feedback of the actual terrain. If the feedback is a valid response to the actual terrain, the depth measurement sequence of the current period is input into a one-dimensional Kalman filter bank that is independently configured for each beam. The Kalman filtering algorithm is used to predict the depth values ​​of each beam in the next period, and the predicted depth value sequence and its prediction variance are obtained. Calculate the depth threshold for the next period based on the predicted depth value sequence and its prediction variance; Based on depth thresholds and depth measurement sequences, the detection range and transmitted signal format are adjusted; simultaneously, based on the spectral center energy analysis of the received channel signal, the transmitted power is adjusted.

[0007] Furthermore, the step of calculating the quality factor of each beam based on the echo data of each beam includes: Obtain the time of arrival of terrain target echoes from each beam echo data. t and its time estimate standard deviation ; Calculate the quality factor of each beam using the following formula: in, d For the corresponding depth value, This represents the standard deviation of the depth estimate corresponding to the depth value. For time-domain signal-to-noise ratio, For quality factor, This represents a common logarithm with base 10.

[0008] Furthermore, the step of determining whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors includes: Calculate the arithmetic mean of the quality factors of all beams in the current cycle; If the arithmetic mean is greater than or equal to a preset threshold, the current depth measurement data is determined to be a valid feedback of the actual terrain. If the arithmetic mean is less than the preset threshold, the current depth measurement data is determined to be invalid.

[0009] Furthermore, the one-dimensional Kalman filter bank comprises parallel-processed one-dimensional Kalman filters equal to the total number of sonar beams, with each one-dimensional Kalman filter corresponding to the depth time series of one beam. The depth threshold includes an upper depth threshold and a lower depth threshold. The step of calculating the depth threshold for the next period based on the predicted depth value sequence and its prediction variance includes: Upper depth threshold: Lower depth threshold: in, For the first k The depth prediction value for each cycle, For the first k The prediction variance for each period, The coefficient is set according to the expected confidence level. For the first k +1 cycle upper depth threshold, For the first k +1 cycle depth threshold.

[0010] Furthermore, in the step of adjusting the detection range and the transmitted signal form based on the depth threshold and depth measurement sequence, the step of adjusting the detection range includes: Obtain the maximum lower threshold value and current range of all beams in the current period; wherein, the maximum lower threshold value is the maximum value among the lower depth thresholds of all beams; A first range threshold and a second range threshold are set according to the current range; wherein the first range threshold is greater than the second range threshold; If the maximum lower threshold is less than the first range threshold and greater than the second range threshold, then the current range is maintained. If the maximum lower threshold is less than or equal to the second range threshold, then the current range is reduced. If the maximum lower threshold is greater than or equal to the first range threshold, then the current range is expanded.

[0011] Furthermore, in the step of adjusting the detection range and the transmitted signal format based on the depth threshold and depth measurement sequence, the step of adjusting the transmitted signal format includes: Obtain the water depth value corresponding to the depth measurement sequence; Based on the preset correspondence table between water depth values ​​and signal transmission formats, the corresponding signal transmission format is selected according to the magnitude of the water depth value.

[0012] Furthermore, the step of adjusting the transmit power based on the spectral center energy analysis of the received channel signal includes: Perform a Fourier transform on any valid received channel signal to obtain the spectrum; Calculate the energy value at the center frequency of the signal based on the spectrum. If the energy value at the center frequency decreases for no less than a preset number of consecutive cycles exceeding a preset threshold, the emission power of the current sound source level is reduced; otherwise, the emission power of the current sound source level is maintained.

[0013] Furthermore, after the step of determining whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors, the method further includes: If the feedback is not a valid response to the actual terrain, then extract the beam echo data for the current period. The beam echo data is modeled as a hidden Markov model; where the discrete depth points in each beam direction are hidden states and the echo intensity is the observed value. Using the echo intensity of each candidate depth point as the observation value, the emission probability of each hidden state is calculated according to the Gaussian distribution function; where the higher the echo intensity, the greater the emission probability. Using the negative correlation between depth changes between adjacent beams as a constraint, the transition probability between states is calculated according to the Gaussian distribution function; where the smaller the depth change between adjacent beams, the greater the transition probability. The Viterbi algorithm is used for recursive calculation to find the globally optimal depth path that maximizes the cumulative probability. The globally optimal depth path is used as a new depth measurement sequence and input into the one-dimensional Kalman filter bank to re-execute depth value prediction and depth threshold calculation.

[0014] Furthermore, the time of arrival t and its time estimate standard deviation It is obtained by extracting the echo data of each beam using the amplitude bottom detection method or the phase bottom detection method.

[0015] A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the following steps: Receive raw data from the current cycle of the multibeam echo sounder; wherein the raw data includes at least the echo sequence of each beam, the received channel signal, and the echo data of each beam. The quality factor of each beam is calculated based on the echo data of each beam. The mean of all beam quality factors is used to determine whether the current bathymetry data is a valid feedback of the actual terrain. If the feedback is a valid response to the actual terrain, the depth measurement sequence of the current period is input into a one-dimensional Kalman filter bank that is independently configured for each beam. The Kalman filtering algorithm is used to predict the depth values ​​of each beam in the next period, and the predicted depth value sequence and its prediction variance are obtained. Calculate the depth threshold for the next period based on the predicted depth value sequence and its prediction variance; Based on depth thresholds and depth measurement sequences, the detection range and transmitted signal format are adjusted; simultaneously, based on the spectral center energy analysis of the received channel signal, the transmitted power is adjusted.

[0016] The beneficial effects of this invention are as follows: By receiving multi-source raw data including depth sounding sequences, channel signals, and beam echoes, and calculating the quality factors of each beam based on the beam echo data, a quantitative criterion for data validity is constructed using the mean of all beam quality factors, thereby accurately distinguishing between real terrain feedback and noise interference. When determining data validity, the depth sounding sequence is input into a parallel-configured one-dimensional Kalman filter bank. The Kalman filtering algorithm is used to predict the depth values ​​of each beam and output the prediction variance. Then, the upper and lower depth thresholds are dynamically calculated based on the predicted depth value sequence and its prediction variance, so that the threshold width can be adaptively adjusted according to the changes in terrain prediction uncertainty, ensuring both noise resistance under gentle terrain and inclusiveness under complex terrain. Based on the dynamically generated depth thresholds and depth sounding sequences, the detection range and transmitted signal form are jointly adjusted to ensure optimal matching between detection coverage and resolution. At the same time, by using the spectral center energy analysis of the received channel signal, the signal amplitude limiting risk is monitored in real time and the transmitted power is adjusted in reverse, forming a closed-loop control link from sensing to execution.

[0017] It solves the technical problems of unmanned platforms being unable to autonomously coordinate and adjust in multi-parameter coupled scenarios and the easy loss of depth threshold lock-up, improves the operational stability of the depth sounding system in complex hydrological environments, the continuity of data acquisition, and the adaptive capability of parameter settings, and realizes intelligent closed-loop control of the entire process.

[0018] By constructing a complete technical logic of data quality assessment, routine tracking and adjustment, and autonomous reset after lockout, the single-parameter independent control is upgraded to a multi-parameter collaborative optimization system solution. This not only achieves high-precision dynamic tracking of depth threshold, but also enhances the system's robustness and self-recovery capability under extreme conditions by introducing spectrum analysis and hidden Markov model path planning, ensuring the efficiency and reliability of unmanned platform depth sounding operations. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] in: Figure 1 This is an application environment diagram of an intelligent parameter adjustment method for depth sounding sonar in one embodiment; Figure 2 This is a flowchart of an intelligent parameter adjustment method for a depth sounding sonar in one embodiment; Figure 3 This is a structural block diagram of an electronic device in one embodiment. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Figure 1 This is a diagram illustrating the application environment of intelligent parameter adjustment for a depth-sounding sonar in one embodiment. (Refer to...) Figure 1 This intelligent parameter adjustment method for depth sounding sonar is applied to an intelligent parameter adjustment system for depth sounding sonar. The intelligent parameter adjustment system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; the mobile terminal can be at least one of a mobile phone, tablet, or laptop. The server 120 can be a standalone server or a server cluster consisting of multiple servers.

[0023] like Figure 2 As shown, in one embodiment, a method for intelligent parameter adjustment of a depth-sounding sonar is provided. This method can be applied to both terminals and servers; this embodiment uses terminal application as an example. The intelligent parameter adjustment method for the depth-sounding sonar specifically includes the following steps: S1: Receive the raw data of the current cycle of the multibeam echo sounder; wherein the raw data includes at least the echo sequence of each beam, the received channel signal, and the echo data of each beam. S2: Calculate the quality factor of each beam based on the echo data of each beam; S3: Determine whether the current depth sounding data is a valid feedback of the actual terrain based on the mean of all beam quality factors; S4: If the feedback is effective for the real terrain, the depth measurement sequence of the current period is input into a one-dimensional Kalman filter bank that is independently configured for each beam. The Kalman filtering algorithm is used to predict the depth values ​​of each beam in the next period, and the predicted depth value sequence and its prediction variance are obtained. S5: Calculate the depth threshold for the next period based on the predicted depth value sequence and its prediction variance; S6: Based on the depth threshold and depth measurement sequence, adjust the detection range and the form of the transmitted signal; at the same time, adjust the transmission power based on the spectral center energy analysis of the received channel signal.

[0024] As described in step S1 above, the raw data of the multibeam bathymetry sonar for the current period is received. This raw data includes at least the bathymetry sequence of each beam, the received channel signal, and the echo data of each beam. The raw data can refer to the complete data packet acquired and uploaded to the processing unit by the sonar within one detection cycle (Ping). The bathymetry sequence is topographic data composed of depth point sets calculated by the bottom detection algorithm of each beam; the channel signal is the raw time-domain waveform of the received channel before beamforming processing; the beam echo data is a two-dimensional matrix of echo intensity with the beam number as the horizontal axis and the sampling time as the vertical axis, retaining the scattering information of the entire water column. These data originate from the receiving beamforming module and data acquisition card of the sonar system. The bathymetry sequence directly reflects the seabed morphology at the current moment, the channel signal is used to extract spectral features to assess signal integrity, and the beam echo data provides a comprehensive data foundation for subsequent quality factor (QF) calculation and multi-parameter coordinated adjustment. By receiving raw data from multiple heterogeneous sources, the system can have sufficient information input in both normal tracking and abnormal reset modes, thereby improving the system's perception capability in complex hydrological environments.

[0025] As described in step S2 above, the quality factor of each beam is calculated based on the echo data of each beam. The quality factor (QF) is a quantitative indicator that measures the accuracy and reliability of single-beam bathymetry data; it is defined as the logarithm of the time-domain signal-to-noise ratio. The quality factor is calculated by extracting the arrival time of the topographic target echoes and their estimated standard deviation from the echo data of each beam. Specifically, the arrival time of the seabed echoes is first identified from the echo data using the amplitude bottom detection method or the phase bottom detection method, and the uncertainty (i.e., standard deviation) of the time estimate is statistically analyzed, thereby deriving the depth value and its depth estimation standard deviation. The role of the quality factor is to transform the abstract signal noise level into a comparable value; a higher value indicates clearer echo data and smaller bathymetry error. For example, when the echo data of a certain beam is sharp and the background noise is low, its time estimation standard deviation is extremely small, and the calculated quality factor increases significantly; conversely, if the signal is submerged in noise, the standard deviation increases, and the quality factor decreases. The beam quality factor sequences generated in this step provide a direct basis for the next step of judging the validity of the overall data, realizing the transformation from analog signal characteristics to digital confidence index.

[0026] Before calculating the arithmetic mean of all beam quality factors, optionally, outlier removal processing can be performed on each calculated beam quality factor: if the quality factor of a certain beam is lower than a preset outlier threshold, the quality factor of that beam is marked as invalid and removed when calculating the arithmetic mean, so as to avoid serious outliers of a single beam having too much impact on the overall decision.

[0027] As described in step S3 above, the mean of the quality factors of all beams is used to determine whether the current bathymetry data is a valid feedback of the actual terrain. Within the current detection period (Ping), the quality factor (QF) of each beam calculated in the aforementioned steps is obtained, and these discrete quality factor values ​​are statistically averaged to generate a single index characterizing the overall quality of the entire bathymetry line data. The quality factor QF is a quantitative parameter that measures the signal-to-noise ratio and depth estimation accuracy of a single beam echo data; a higher QF value indicates a more reliable bathymetry result for that beam. By calculating the arithmetic mean, statistical principles can be used to eliminate abnormally high or low values ​​caused by transient reverberation, multipath effects, or local noise interference in individual beams, thereby avoiding misleading judgments about the overall system state due to single-point errors. For example, if a detection includes 512 beams, and 5 beams have extremely low quality factors due to surface bubble interference, while the quality factors of the remaining 507 beams are stable at a high level, directly taking the arithmetic mean can effectively smooth out the influence of these 5 anomalies, ensuring that the final average still objectively reflects the true detection situation of the seabed terrain. The judgment process is a comprehensive evaluation of the quality of all beam data within the current period based on statistical principles. Specifically, the arithmetic mean of the quality factors of all beams within the current period is calculated, and this mean is compared with a preset effective threshold. This threshold is set based on theoretical analysis and field test data, and is usually set as a critical value that can distinguish between real terrain echoes and random noise. If the mean is greater than or equal to the threshold, the current bathymetry data is determined to be a valid feedback of real terrain, indicating that the depth threshold setting is reasonable, and the system enters normal tracking mode; if the mean is less than the threshold, the data is determined to be invalid, suggesting that the depth threshold may have deviated from the real seabed, and the system needs to activate the reset mechanism. For example, if the threshold is set to 2, when the calculated average quality factor is 2.5, the system confirms the data as valid; if it is only 1.2, it is determined to be invalid. By introducing the mean quality factor as a global criterion, misjudgments caused by individual beam anomalies can be effectively eliminated, ensuring that only high-quality terrain data can drive subsequent prediction and adjustment processes, thereby avoiding interference from invalid data to the control closed loop.

[0028] As described in step S4 above, if the feedback is valid and reflects the actual terrain, the current period's bathymetry sequence is input into a one-dimensional Kalman filter bank independently configured for each beam. The Kalman filtering algorithm is then used to predict the depth values ​​of each beam in the next period, resulting in a predicted depth value sequence and its prediction variance. The one-dimensional Kalman filter bank is a computational array consisting of a number of parallel-running one-dimensional Kalman filters equal to the total number of sonar beams. Each filter is specifically responsible for processing the depth time series of a fixed beam number. Since the seabed depth pointed to by the same beam during adjacent detection cycles typically exhibits continuity and gradual variation, the Kalman filtering algorithm utilizes this temporal correlation. Through a prediction-update recursive mechanism, combining the state estimate from the previous moment with the observation value at the current moment (i.e., the confirmed valid bathymetry sequence mentioned above), it calculates the optimal estimate of the depth at the next moment (predicted depth value) and the uncertainty of this estimate (prediction variance). The predicted depth value reflects the changing trend of the terrain, while the prediction variance quantifies the confidence interval of the prediction result. The prediction variance is a scalar value of the error covariance output by the one-dimensional Kalman filter in the prediction step, used to quantify the uncertainty of the depth prediction value. The smaller the value, the more accurate the prediction; conversely, the larger the value, the greater the prediction uncertainty. For example, for a gentle seabed, the prediction variance is small, and the filter assigns a higher weight to the predicted value; for steep terrain, the prediction variance increases, and the filter dynamically adjusts its gain to quickly track the changes. The predicted depth value sequence and prediction variance output in this step not only achieve smooth tracking of the seabed terrain but also provide core parameters for the subsequent dynamic calculation of the depth threshold, ensuring that the threshold setting can both cover the real terrain and suppress noise interference.

[0029] As described in step S5 above, the depth threshold for the next period is calculated based on the predicted depth value sequence and its prediction variance. The depth threshold includes an upper depth threshold and a lower depth threshold, which are boundary values ​​used to limit the bottom detection search range for the next period. The depth threshold is calculated based on the predicted depth value sequence and its corresponding prediction variance, according to a set confidence level coefficient. Specifically, the calculation logic is as follows: The upper and lower thresholds are determined by adding or subtracting a tolerance range consisting of the product of the prediction variance and the confidence coefficient, centered on the depth prediction value of the previous period. This setting allows the threshold width to adaptively change with terrain complexity: in flat terrain with low prediction variance, the threshold narrows to improve noise immunity; in complex terrain with high prediction variance, the threshold widens to prevent missed detection of true echoes. For example, if the predicted depth of a beam is 50 meters, the standard deviation of the prediction variance is 0.5 meters, and the confidence coefficient is set to 3, then the threshold range can be set to [48.5 meters, 51.5 meters]. By dynamically generating thresholds based on prediction results, this method overcomes the limitation of fixed thresholds in adapting to varying terrains, achieving intelligent and adaptive depth tracking. In practical engineering implementation, the one-dimensional Kalman filter bank can be implemented in parallel by a digital signal processor (DSP) or a field-programmable gate array (FPGA). For a typical 512-beam depth sounding sonar system, each filter only needs to maintain two state variables (depth estimate and error covariance), with a computational load of approximately dozens of floating-point operations per prediction-update cycle. The total computational load is within the real-time processing capabilities of conventional embedded platforms.

[0030] As described in step S6 above, the detection range and transmitted signal format are adjusted based on the depth threshold and bathymetry sequence; simultaneously, the transmitted power is adjusted based on the spectral center energy analysis of the received channel signal. This step includes three parallel sub-adjustment processes aimed at achieving coordinated optimization of multiple parameters. First, the detection range is adjusted based on the maximum lower threshold value of all beams within the current cycle. By comparing the ratio of this maximum value to the current range, the range is dynamically maintained, reduced, or expanded to ensure that the detection range covers the seabed while retaining an appropriate margin, avoiding data overflow or resource waste. Second, the transmitted signal format is adjusted based on the water depth value reflected in the bathymetry sequence, by querying a preset water depth-signal format correspondence table. For example, in shallow water, a narrow pulse signal is automatically switched to improve resolution, while in deep water, a wide pulse or linear frequency modulated signal is switched to increase the detection distance. Finally, the adjustment of the transmission power is based on the energy analysis of the center frequency spectrum. The spectrum is obtained by performing a Fourier transform on the channel signal, and the energy change trend at the center frequency is monitored. If the energy drops below a threshold for several consecutive cycles, it indicates that the signal may be experiencing amplitude limiting distortion, and the system automatically reduces the transmission power. Conversely, if the energy is too low, the power is increased. These three adjustment processes work together: the depth threshold determines the benchmark for range adjustment, the depth measurement sequence guides the selection of signal form, and the spectrum analysis ensures the optimal state of the transmission energy. Through this multi-dimensional coordinated adjustment, the system can maintain optimal detection performance and data quality under different water depths, different bottom conditions, and different noise environments, solving the technical challenge of autonomously coordinating the adjustment of multiple parameters in unmanned platforms.

[0031] In one embodiment, step S2, which calculates the quality factor of each beam based on the echo data of each beam, includes: S201: Obtain the arrival time of terrain target echoes in each beam echo data. t and its time estimate standard deviation ; S202: Calculate the quality factor of each beam according to the following formula: in, d For the corresponding depth value, This represents the standard deviation of the depth estimate corresponding to the depth value. For time-domain signal-to-noise ratio, For quality factor, This represents a common logarithm with base 10.

[0032] As described in step S201 above, the arrival time t of the terrain target echo in each beam echo data and its time estimation standard deviation are obtained. δt Porta Timet This can refer to the time interval from the moment the sound wave is emitted until the peak or characteristic point of the echo from the seabed topographic target is received, which directly corresponds to the one-way propagation time between the sonar and the seabed; the time estimation standard deviation. δt This is a quantitative representation of the uncertainty in the time of arrival measurement, reflecting the dispersion of the detection algorithm's determination of the echo arrival time under noise interference. The time of arrival t and its time estimation standard deviation δt are extracted by preprocessing the echo data of each beam using either the amplitude-bottom detection method or the phase-bottom detection method. The amplitude-bottom detection method determines t by finding the maximum value of the echo envelope or the zero-crossing point of its derivative, while the phase-bottom detection method uses the phase jump characteristics of the echo data to lock t, and then calculates δt by statistically analyzing the fluctuations of multiple samples or adjacent samples. For example, in a certain detection, for the i-th beam, the peak value of the detected echo envelope appears 0.05 seconds after transmission, i.e., t=0.05s. Within 10 consecutive sampling windows, the fluctuation range of this peak time results in a calculated standard deviation δt of 0.002s. Through this high-precision time-domain feature extraction, the signal-to-noise ratio characteristics of the echo data can be accurately captured, providing reliable raw data support for subsequent quality assessment.

[0033] As described in step S202 above, the quality factor of each beam is calculated according to the following formula: Where d is the corresponding depth value, and δd is the standard deviation of the depth estimate corresponding to the depth value. For time-domain signal-to-noise ratio, For quality factor, This represents a common logarithm with base 10.

[0034] Among them, the time domain signal-to-noise ratio Essentially, it is the reciprocal of the relative precision of the echo arrival time, due to depth. With time Proportional ( (where c is the speed of sound), therefore the relative error in time Equivalent to the relative error of depth Quality factor This is the result after taking the common logarithm of the relative accuracy. This logarithmic transformation not only normalizes the dimensions but also compresses the signal-to-noise ratio of different orders of magnitude into a stable numerical range, making the measurement quality in shallow and deep water areas comparable. Specifically, the extracted time of arrival (t) is first converted to a depth value (d) using the sound velocity profile, and the standard deviation of the time estimate (δt) is converted to the standard deviation of the depth estimate (δd). Then, the ratio of the two is calculated. Finally, taking the logarithm yields... For example, if a certain beam measures depth... Depth estimation standard deviation ,but Then the quality factor is calculated. If another deep-water beam measures , ,That Also 100, The value is also 2, indicating that although the absolute errors are different, their relative measurement accuracy is consistent, and the quality factor can objectively reflect this equivalence. Through... and Through collaborative computing, the system transforms complex echo waveform characteristics into a single quantitative indicator, effectively eliminating the interference of water depth scale changes on quality assessment and ensuring the scientific validity and robustness of subsequent data validity judgments.

[0035] In one embodiment, step S3, which determines whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors, includes: S301: Calculate the arithmetic mean of the quality factors of all beams in the current cycle; S302: If the arithmetic mean is greater than or equal to a preset threshold, then the current depth measurement data is determined to be a valid feedback of the real terrain; S303: If the arithmetic mean is less than the preset threshold, the current depth measurement data is determined to be invalid.

[0036] As described in step S301 above, the arithmetic mean of the quality factors of all beams in the current period is calculated. The quality factor (QF) of each beam calculated in the aforementioned steps within the current detection period is obtained, and these discrete quality factor values ​​are statistically averaged to generate a single index characterizing the overall quality of the entire sounding line data. The quality factor QF is a quantitative parameter measuring the signal-to-noise ratio and depth estimation accuracy of a single beam echo data; a higher QF value indicates more reliable sounding results for that beam. By calculating the arithmetic mean, statistical principles can be used to eliminate abnormally high or low values ​​caused by transient reverberation, multipath effects, or local noise interference in individual beams, thereby avoiding misleading judgments about the overall system state due to single-point errors. For example, if a detection includes 512 beams, and 5 beams have extremely low quality factors due to surface bubble interference, while the quality factors of the remaining 507 beams are stable at a high level, directly taking the arithmetic mean can effectively smooth out the influence of these 5 anomalies, ensuring that the final average value still objectively reflects the true detection situation of the seabed topography.

[0037] As described in step S302 above, if the arithmetic mean is greater than or equal to a preset threshold, the current bathymetry data is determined to be a valid feedback of the actual terrain. The preset threshold is a fixed value derived theoretically and calibrated through field experiments, used to delineate the boundary between effective tracking and loss of lock-on control. In this embodiment, the preset threshold is set to 2. When the calculated arithmetic mean falls into this range (i.e., ≥2), it indicates that the echo data of most beams in the current cycle is clear, the standard deviation of depth estimation is small, and the system is in a normal bottom tracking state. The current bathymetry sequence truly reflects the seabed topographic features. At this time, the normal tracking mode will be maintained, allowing the subsequent Kalman filter prediction and parameter adaptive adjustment process to continue. For example, when the water depth in a certain sea area changes gently, the echo intensity of each beam is high and stable, and the calculated average quality factor is about 3.5, which is significantly higher than the preset threshold of 2, the system determines that the data is valid and directly inputs the current bathymetry sequence into a one-dimensional Kalman filter bank for depth prediction in the next cycle.

[0038] Loss of lock refers to the failure of the depth threshold to effectively track the actual seabed topography, resulting in bathymetry data deviating from the true value or the inability to detect valid echoes. Loss of lock reset refers to the process of re-searching for the actual topography using a backup method (such as Viterbi path planning based on beam echo data) after detecting a loss of lock, thereby restoring effective tracking of the depth threshold. Detection range refers to the maximum detection distance range set by the sonar system, corresponding to the length of the receiver sampling time window; its value directly affects the depth coverage of the data acquisition. Source level refers to the intensity of sound waves radiated by the sonar transducer in a specific direction, measured in decibels (dB), used to characterize the power level of the transmitted signal.

[0039] As described in step S303 above, if the arithmetic mean is less than a preset threshold, the current bathymetry data is deemed invalid. When the arithmetic mean falls below the preset threshold (i.e., <2), it means that a large number of beams failed to detect valid seabed echoes in the current period, or the detected signal noise was too high, causing the depth estimation error to exceed the allowable range. It is determined that the current depth threshold has lost lock or control, and the existing bathymetry sequence no longer represents the real terrain, but is feedback from noise or false targets. This will immediately block the conventional Kalman filter prediction process to prevent the erroneous prediction value from further deteriorating the threshold setting and trigger the lock-out reset mechanism. For example, when the sonar passes through a strong thermocline or encounters severe wave interference, the echo data is severely distorted, causing the quality factors of each beam to generally decrease, and the arithmetic mean drops to 1.2, which is lower than the threshold of 2. The current data is then deemed invalid, and the Viterbi dynamic path planning algorithm based on the beam echo data is started to re-search for the optimal terrain curve to restore threshold tracking.

[0040] A data validity evaluation model based on the principle of group consistency was constructed by calculating the arithmetic mean of all beam quality factors and comparing it with a preset threshold. The statistical characteristics of the arithmetic mean effectively suppress the interference of individual bad beams on the overall decision, improving the system's robustness in non-uniform noise environments. Simultaneously, the introduction of the preset threshold provides a clear and reproducible hard decision boundary for system behavior, making the switching logic between normal tracking and lock-out reset modes clear and deterministic. This automatic decision mechanism based on quantitative indicators not only solves the misjudgment problem caused by traditional methods relying on human experience or single-beam judgment, but also serves as a prerequisite for subsequent collaborative operations such as depth threshold prediction, range adjustment, and power control, ensuring the stability and data reliability of the unmanned platform's multibeam bathymetry sonar throughout the entire operational cycle.

[0041] In one embodiment, the one-dimensional Kalman filter bank comprises parallel-processed one-dimensional Kalman filters equal to the total number of sonar beams, each one-dimensional Kalman filter corresponding to the depth time series of one beam, and the depth threshold includes an upper depth threshold and a lower depth threshold; step S5, which calculates the depth threshold for the next period based on the predicted depth value sequence and its prediction variance, includes: S501: Upper Depth Threshold: S502: Lower Depth Threshold: in, This is the depth prediction value for the k-th period. Let Variance be the prediction variance for the k-th period. The coefficient is set according to the expected confidence level. This is the upper depth threshold for the (k+1)th period. This is the depth threshold for the (k+1)th cycle.

[0042] As described in steps S501-S502 above, a parallel one-dimensional Kalman filter bank is constructed, with the same number of sonar beams as the total number of beams. Each one-dimensional Kalman filter corresponds to the depth time series of one beam. The one-dimensional Kalman filter bank is a collection of multiple independently operating one-dimensional Kalman filters, the number of which is strictly equal to the total number of beams N in the multi-beam bathysound sonar. Each filter is specifically responsible for processing the depth observation sequence of a specific numbered beam in the time dimension, performing prediction and update operations in parallel without interference. This structure utilizes the continuity and gradual variation of seabed topography changes along the same beam direction between adjacent detection periods (Ping). For example, for a bathysound sonar system with 512 beams, the system will instantiate 512 independent one-dimensional Kalman filters. The i-th filter only receives and processes the historical bathysound data of the i-th beam, thereby decomposing the complex multidimensional state estimation problem into N low-complexity one-dimensional estimation problems, significantly reducing the computational load and improving the real-time response speed. This parallel processing mechanism ensures that depth trend predictions for all beams are completed synchronously within each detection cycle, providing a unified time reference for subsequent threshold generation.

[0043] The depth threshold for the next cycle is calculated based on the predicted depth value sequence and its prediction variance. The depth threshold includes an upper depth threshold and a lower depth threshold, which are boundary values ​​used to limit the bottom detection search range for the next cycle. The depth threshold is calculated based on the predicted depth value sequence and its corresponding prediction variance, according to a set confidence level coefficient. The specific calculation logic is as follows: The upper and lower thresholds are determined by adding or subtracting a tolerance range consisting of the product of the prediction variance and the confidence coefficient, centered on the depth prediction value of the previous cycle. This setting allows the threshold width to adaptively change with terrain complexity: in flat terrain with low prediction variance, the threshold narrows to improve noise immunity; in complex terrain with high prediction variance, the threshold widens to prevent missed detection of true echoes. For example, if a beam's predicted depth is 50 meters, the standard deviation of the prediction variance is 0.5 meters, and the confidence coefficient is set to 3, then the threshold range can be set to [48.5 meters, 51.5 meters]. By dynamically generating thresholds based on prediction results, this method overcomes the shortcomings of fixed thresholds in adapting to varying terrains, and achieves intelligent and adaptive depth tracking.

[0044] The upper and lower depth thresholds for the next period are calculated based on the predicted depth value sequence and its prediction variance. The depth thresholds include the upper depth threshold. and lower depth threshold Both are boundary values ​​dynamically generated based on the statistical confidence interval principle. Specifically, the upper depth threshold is based on the formula... The calculated lower depth threshold is based on the formula. Calculated. Here. This represents the depth prediction value output after Kalman filtering in the k-th cycle, which represents the optimal estimation center of the seabed topography at the next moment. This represents the prediction variance for the k-th period, quantifying the degree of uncertainty of the predicted value. A coefficient set according to the expected confidence level is used to control the width of the threshold window to cover the real terrain within a specific probability range.

[0045] In one embodiment, step S6, which adjusts the probe range and the transmitted signal form based on a depth threshold and a depth measurement sequence, includes the following step: S601: Obtain the maximum lower threshold value and current range of all beams in the current period; wherein, the maximum lower threshold value is the maximum value among the lower depth thresholds of all beams; S602: Set a first range threshold and a second range threshold according to the current range; wherein, the first range threshold is greater than the second range threshold; S603: If the maximum lower threshold is less than the first range threshold and greater than the second range threshold, then maintain the current range; S604: If the maximum lower threshold is less than or equal to the second range threshold, then reduce the current range; S605: If the maximum lower threshold value is greater than or equal to the first range threshold, then expand the current range.

[0046] As described in step S601 above, the maximum lower threshold and current range of all beams within the current period are obtained. The maximum lower threshold can refer to the maximum value among the lower depth thresholds corresponding to all beams processed in parallel within the current measurement period (Ping). Since the lower depth threshold is calculated by adding the prediction variance to the depth value predicted by Kalman filtering, it represents the deepest possible location of the current terrain within the statistical confidence interval. Obtaining this maximum value aims to determine the actual maximum depth boundary detected in a conservative estimation manner, preventing insufficient range due to individual beams detecting deep pits or steep slopes. The current range can refer to the maximum detection range set by the sonar system in the current period, usually corresponding to the time window length of the receiver analog-to-digital converter (ADC) or the upper limit of the number of sampling points. For example, if the current range R set by the sonar system is 50 meters, and the calculated lower depth thresholds among the 512 beams are distributed between 30 meters and 42 meters, then the maximum value of 42 meters is extracted as the maximum lower threshold value in this step. By monitoring the relative relationship between the maximum lower threshold and the current range in real time, accurate data input can be provided for subsequent dynamic range adjustments, ensuring that the adjustment strategy is based on real terrain feedback rather than fixed empirical values.

[0047] As described in step S602 above, a first range threshold and a second range threshold are set according to the current range; wherein, the first range threshold is greater than the second range threshold. The first range threshold and the second range threshold are two critical judgment values ​​calculated based on the current range according to a preset proportional coefficient, used to construct the hysteresis interval of range adjustment to avoid frequent oscillations of the system in the critical state. The first range threshold is usually set to a higher proportional value of the current range (e.g., 0.875 times), representing the warning line that the range is about to be insufficient; the second range threshold is usually set to a lower proportional value of the current range (e.g., 0.75 times), representing the contraction line that the range is too redundant. Specifically, if the current range is Then the first range threshold Second range threshold This dual-threshold mechanism introduces hysteresis characteristics, allowing the range expansion and contraction operations to be triggered at different depth levels, effectively filtering out misjudgments caused by minor terrain fluctuations or measurement noise.

[0048] As described in step S603 above, if the maximum lower threshold is less than the first range threshold and greater than the second range threshold, the current range is maintained. When the maximum lower threshold... satisfy This indicates that the currently set range can fully cover the seabed topography (without overflow risk) while retaining a suitable amount of redundancy to ensure the stability of signal processing, and no adjustment is required. For example, the current range... The system calculates a first threshold of 35 meters and a second threshold of 30 meters. If the maximum lower threshold is 32 meters, falling within the (30, 35) range, the system determines that the current range configuration is optimal and maintains it. The data acquisition continues without changing the current range and proceeds to the next cycle. Maintaining the current range avoids unnecessary parameter switching that could cause transient system responses, ensuring the continuity and consistency of the depth sounding data sequence. This is the steady-state maintenance component in the adaptive range adjustment strategy.

[0049] As described in step S604 above, if the maximum lower threshold is less than or equal to the second range threshold, the current range is reduced. When the maximum lower threshold... This indicates that the actual maximum detected depth is much smaller than the current range limit. An excessively large range will cause the Time Variable Gain (TVG) curve to be stretched, the quantization resolution of the ADC (Analog-to-Digital Converter) to be less utilized in the effective signal segment, and an unnecessary increase in data processing volume. At this time, the system will perform a range reduction operation, reducing the current range... Update to a new, smaller value. In practice, the reduced new range can be set as a fixed percentage of the original range (e.g., ...). Alternatively, the range can be directly approximated to the maximum lower threshold with a safety margin. For example, if the current range is 100 meters, the second threshold is 75 meters, and the maximum lower threshold is only 60 meters, the system will reduce the range to 75 meters. By reducing the range in a timely manner, the amplification gain of the receiving channel for valid echo data can be improved, the signal-to-noise ratio of weak signals can be enhanced, and the storage and transmission overhead of invalid data segments can be reduced, thus optimizing the overall system energy efficiency.

[0050] As described in step S605 above, if the maximum lower threshold is greater than or equal to the first range threshold, the current range is expanded. When the maximum lower threshold... When this occurs, it indicates that the deepest point of the seabed topography has approached or reached the limit of the current measurement range. If not adjusted in time, depth overflow is highly likely to occur in the next cycle, resulting in the loss of some seabed topography data or its incorrect labeling as noise. At this time, the system will perform a range expansion operation to increase the current measurement range. Update to a larger value. In practice, the expanded new range can be set to a fixed multiple of the original range (e.g., ...). For example, if the current measurement range is 40 meters, the first threshold is 35 meters, and the maximum lower threshold reaches 36 meters, the system will expand the measurement range to 45 meters. This proactive expansion mechanism utilizes the statistical characteristics of Kalman filter predictions to actively expand the detection window before the data actually overflows, ensuring the integrity and reliability of bathymetry data under complex and undulating terrain.

[0051] In one embodiment, step S6, which adjusts the probe range and the transmitted signal format based on a depth threshold and a depth measurement sequence, includes the following step regarding the transmitted signal format: S611: Obtain the water depth value corresponding to the depth measurement sequence; S612: Based on the preset correspondence table between water depth values ​​and transmission signal formats, select the corresponding transmission signal format according to the magnitude of the water depth value.

[0052] As described in step S611 above, the water depth value corresponding to the bathymetry sequence is obtained. The water depth value can refer to the seabed topographic depth data effectively detected by the bathymetry sonar in the current period, derived from a bathymetry sequence confirmed as valid after quality factor evaluation. Specifically, this water depth value can be the arithmetic mean of the depths of all effective beams in the current period, or it can be a statistical value (such as the median or mode) representing the typical depth of the current operating area, used to characterize the current operating water depth environment of the sonar. The acquisition of this water depth value relies on the aforementioned reception of raw data, calculation of the quality factor, and judgment of data validity, ensuring that the water depth data used as the basis for decision-making is authentic and reliable, and eliminating noise or outlier interference. For example, when the sonar detects that the average quality factor of 480 out of 512 beams in a certain period is greater than the preset threshold of 2, the system will extract the depth readings of these 480 effective beams, calculate their average value to obtain the current water depth value of 18.5 meters. This water depth extraction method based on effective feedback sequences ensures that subsequent signal form adjustment strategies are established on the basis of accurate environmental perception, avoiding parameter misadjustment caused by erroneous depth data.

[0053] As described in step S612 above, based on the preset correspondence table between water depth values ​​and transmitted signal formats, the corresponding transmitted signal format is selected according to the magnitude of the water depth value. The preset correspondence table between water depth values ​​and transmitted signal formats is a lookup table pre-stored in the sonar control system's memory. This table establishes mapping rules between discrete water depth intervals and optimal transmitted signal parameters (including pulse type, pulse width, modulation method, etc.). Transmitted signal formats mainly include continuous wave (CW) pulses and linear frequency modulation (LFM) signals. Different signal formats have different time-frequency characteristics and are suitable for different water depth scenarios: shallow water environments require narrow pulses to obtain high distance resolution and avoid reverberation interference; deep water environments require wide pulses or LFM signals with a large time-width-bandwidth product to ensure sufficient detection energy and signal-to-noise ratio. The correspondence table is set according to the sonar equation and the acoustic propagation characteristics of the ocean, dividing the water depth into multiple intervals, with each interval matched with a specific signal format. The corresponding relationship table is set based on the sonar equation and the characteristics of ocean acoustic propagation: In shallow water (usually less than 10 meters), the sound propagation distance is short and the reverberation is strong. Using narrow pulses (such as 50 microsecond CW) can improve the range resolution to avoid the transmission signal and the seabed echo overlapping in time. In medium and deep water (10 to 25 meters), the propagation loss increases, and the pulse width needs to be increased to ensure the echo signal-to-noise ratio, but continuous wave pulses can still meet the resolution requirements. In deep water (greater than 25 meters), due to the significant propagation loss, LFM signals with a large time-bandwidth product need to be used. Pulse compression technology can be used to improve the detection range without reducing the range resolution. For example, the following corresponding rules are set: when the water depth is less than or equal to 5 meters, a CW pulse with a pulse width of 50 microseconds is selected to maximize the longitudinal resolution in shallow water areas; when the water depth is between 6 and 10 meters, a CW pulse with a pulse width of 150 microseconds is selected to balance resolution and detection capability; when the water depth is between 11 and 25 meters, a CW pulse with a pulse width of 500 microseconds is selected to increase transmission energy; when the water depth is greater than 25 meters, an LFM signal with a pulse width of 5 milliseconds is selected, and pulse compression technology is used to significantly improve the detection range without reducing the range resolution. Through the lookup table mechanism, the system can complete the decision and switching of signal form in milliseconds without complex real-time iterative calculations, which not only meets the stringent real-time requirements of unmanned platforms but also achieves precise matching between the transmitted waveform and the hydrological environment.

[0054] In one embodiment, step S6, which adjusts the transmit power based on the spectral center energy analysis of the received channel signal, includes: S621: Perform a Fourier transform on any valid received channel signal to obtain the spectrum; S622: Calculate the energy value at the center frequency of a signal based on the spectrum; S623: If the energy value at the center frequency decreases for no less than a preset number of cycles for more than a preset threshold, then reduce the emission power of the current sound source level; otherwise, maintain the emission power of the current sound source level.

[0055] As described in step S621 above, a Fourier transform is performed on the channel signal to obtain the spectrum. The Fourier transform is a mathematical process that converts a channel signal in the time domain into a frequency domain representation. Its function is to separate the mixed frequency components, thereby clearly presenting the energy distribution of the signal at different frequencies. Specifically, the processor receives the original channel signal sequence from the sonar receiver and performs a Fast Fourier Transform (FFT) operation on the sequence. Since the transmitted signal of a multibeam sonar typically has a defined center frequency... In the spectrum obtained after Fourier transform, the energy is mainly concentrated near the center frequency. For example, for a sonar system with a center frequency of 200kHz, the transformed spectrum will show a significant peak at 200kHz, while the energy of noise and other interference frequencies is distributed at the sidelobes or noise floor level. Through this frequency domain transformation, amplitude fluctuation interference caused by phase superposition or instantaneous pulses in the time domain waveform can be eliminated, and frequency domain features reflecting the essential strength of the signal can be extracted.

[0056] As described in step S622 above, the energy value at the signal's center frequency is calculated based on the spectrum. The energy value at the signal's center frequency can refer to the value corresponding to the nominal center frequency of the sonar transmission signal within the obtained spectrum. The integral of the square of the spectral amplitude or the peak amplitude within its adjacent narrowband range is obtained by directly reading the index corresponding to the spectral array. The numerical value of the location. This energy value. Its function is to serve as a key indicator for determining whether the transmitted signal has entered the nonlinear limiting region, and to locate the center frequency after acquiring the spectrum. At the corresponding frequency point, extract the amplitude value at that point and calculate its energy (usually the square of the amplitude). Within the linear amplification range where the signal is not clipped, as the transmit power increases or the propagation conditions stabilize, The waveform exhibits a strict linear positive correlation with the input signal strength. Once the receiving front-end or transmitting transducer enters a saturation-limiting state, the top of the waveform is clipped, leading to an increase in high-frequency harmonic components and a stagnation or even decrease in the energy growth of the fundamental frequency (i.e., the center frequency). For example, when the sound source level is increased from 210dB to 215dB, if the system is operating normally... It should be increased by approximately 1.78 times (corresponding to a 5dB gain); however, if it is detected at this time... If the increase is only 1.2 times or remains the same, it indicates that the signal has been distorted. In practical applications, the preset number can be pre-set according to the dynamic characteristics of the marine environment, preferably 10 detection cycles; the preset threshold can be set to an energy value decrease of more than 3dB or a deviation from the theoretical linear growth value of more than 20%. The above values ​​are only examples, and can be adaptively adjusted according to the specific parameters of the sonar system and the noise environment of the working water area in actual implementation.

[0057] As described in step S623 above, if the energy value at the center frequency decreases for at least a preset number of cycles exceeding a preset threshold, the transmission power of the current sound source level is reduced; otherwise, the transmission power of the current sound source level is maintained. The preset number can refer to the number of consecutive pings (cycles) used to smooth transient interference and ensure decision reliability. Its setting can be adjusted according to the dynamic characteristics of the marine environment, and is typically set to 10 cycles. The preset threshold can refer to the critical proportion or decibel value for judging abnormal energy decreases, for example, set to a decrease in energy value exceeding 3 dB from the previous cycle or a deviation exceeding 20% ​​from the theoretical linear growth value. The main execution body for this step is the sonar parameter automatic control module, which executes the process by recording the most recent N cycles using a sliding window mechanism. The sequence is analyzed, and its rate of change is calculated in real time. The current period's... Compare with the data from the previous period; if... If the sonar exhibits a decreasing trend for at least 10 consecutive cycles and the cumulative decrease exceeds a preset threshold (e.g., 5%), it is determined that the current transmit power is too high, causing received signal clipping distortion. A control command is then generated to reduce the current sound source level by one level (e.g., reducing the current sound source level by 2%, i.e., approximately 4dB for a 220dB sound source). Conversely, if E_{fc} remains stable or fluctuates normally with depth, without triggering the continuous decrease condition, the current sound source level is maintained. For example, in a certain detection mission, the sonar entered a shallow, highly reflective area. The center frequency energy E_fc of 12 consecutive pings continuously decreased from 0.8V² to 0.5V², a decrease of 37.5%, exceeding the set 20% threshold. The system immediately and automatically reduced the transmit power from 220dB to 216dB, thereby eliminating waveform clipping.

[0058] In one embodiment, after step S3, which determines whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors, the method further includes: S401: If the feedback is not a valid response to the actual terrain, extract the beam echo data for the current cycle. S402: Model the beam echo data as a hidden Markov model; wherein, the discrete depth points in each beam direction are hidden states, and the echo intensity is the observed value; S403: Using the echo intensity of each candidate depth point as the observation value, calculate the emission probability of each hidden state according to the Gaussian distribution function; where the higher the echo intensity, the greater the emission probability. S404: Using the negative correlation between the depth changes between adjacent beams as a constraint, calculate the transition probability between states according to the Gaussian distribution function; where the smaller the depth change between adjacent beams, the greater the transition probability. S405: Apply the Viterbi algorithm to perform recursive calculations and find the globally optimal depth path that maximizes the cumulative probability; S406: The globally optimal depth path is used as a new depth measurement sequence and input into the one-dimensional Kalman filter bank to re-execute depth value prediction and depth threshold calculation.

[0059] As described in step S401 above, if the feedback is not a valid response to the actual terrain, the beam echo data for the current period is extracted. A valid response not indicating actual terrain may mean that the echo data of most beams in the current period is unclear, i.e., the arithmetic mean of all calculated beam quality factors is less than a preset threshold (e.g., 2), indicating that the current depth threshold has been lost, the conventional bathymetry sequence is unreliable, and the system automatically switches to the loss-of-lock reset mode to extract the beam echo data for the current period from the original data packet. This beam echo data is a two-dimensional matrix, where the horizontal axis corresponds to each sonar beam number, the vertical axis corresponds to the depth sampling point in each beam direction, and the values ​​in the matrix represent the echo intensity at each location. Even if the terrain target is not correctly tracked, causing the bathymetry sequence to fail, the beam echo data usually still retains complete terrain echo intensity distribution information, which provides a data basis for subsequent re-searching of the actual terrain.

[0060] As described in step S402 above, the beam echo data is modeled as a Hidden Markov Model (HMM); where discrete depth points in each beam direction are hidden states, and echo intensity is the observed value. The HMM is a key architecture for transforming terrain detection into a probabilistic graphical model. In this model, the hidden state is defined as the possible true seabed depth point in each beam direction, which cannot be directly observed due to noise or error thresholds; the observed value is defined as the echo intensity of the corresponding depth point in the beam echo data, which is a physical quantity directly acquired by the sensor. For the ... Each beam, at each discrete sampling point within its depth range. All are considered a potential hidden state. The grayscale value or energy value of that point in the image. That is, the corresponding observed value.

[0061] As described in step S403 above, the echo intensity of each candidate depth point is used as the observed value, and the emission probability of each hidden state is calculated according to the Gaussian distribution function; where the higher the echo intensity, the greater the emission probability. Emission probability refers to the likelihood of observing the current echo intensity given a certain hidden state (i.e., assuming a certain depth point is the real seabed). This step uses the Gaussian distribution function to construct the emission probability model, with its mean set as the theoretical strong echo center and its variance determined according to the environmental noise level. Specifically, for any candidate depth point, the higher its echo intensity in the beam echo data, the greater the probability that the point is a reflection from the real seabed, and therefore the higher the calculated emission probability value; conversely, if the echo intensity is weak or close to the background noise level, the emission probability approaches zero. For example, when the normalized echo intensity of a certain depth point is 0.9, substituting it into the Gaussian distribution function yields a higher emission probability value, indicating that the point is very likely part of the terrain; while when the intensity is only 0.1, the emission probability decreases significantly.

[0062] As described in step S404 above, the negative correlation between depth changes between adjacent beams is used as a constraint, and the transition probability between states is calculated according to the Gaussian distribution function; where the smaller the depth change between adjacent beams, the greater the transition probability. The transition probability describes the likelihood of jumping from a certain depth state of one beam to another depth state of an adjacent beam, and is used to characterize the smoothness and continuity constraints of the terrain. This step also uses the Gaussian distribution function for modeling, but its variable is the depth change between adjacent beams (i.e., the depth difference). Since natural seabed topography is usually continuously changing, the smaller the depth difference between adjacent beams, the higher the probability of conforming to the physical characteristics of the terrain, and therefore the greater the corresponding transition probability; if the depth difference is huge (such as a vertical cliff jump, unless such special terrain is known to exist), the transition probability drops sharply. Specifically, let the first... The state depth of the beam is , No. The state depth of the beam is Then the transition probability and They exhibit a negatively correlated Gaussian distribution. For example, if the depth difference between adjacent beams is only 0.5 meters, the calculated transfer probability is close to the peak value; if the difference reaches 10 meters, the probability value is extremely low.

[0063] To meet real-time processing requirements, candidate depth points of each beam can be downsampled before executing the Viterbi algorithm. For example, only the depth points corresponding to the local maxima of echo intensity can be selected as candidate states, thereby compressing the search space from the full depth range to a subset of several candidate peaks, significantly reducing the algorithm complexity and enabling it to run in real time on embedded platforms.

[0064] As described in step S405 above, the Viterbi algorithm is applied for recursive calculation to find the globally optimal depth path that maximizes the cumulative probability. The Viterbi algorithm is a dynamic programming algorithm used to find the hidden state sequence most likely to generate the observation sequence in a Hidden Markov Model. In this step, the algorithm starts from the first beam and recursively calculates the emission probability and transition probability as described above, beam by beam. For each candidate depth point in each beam, the algorithm calculates the maximum cumulative probability of reaching that point and records its predecessor node (i.e., which depth point in the previous beam caused this maximum probability). Through this recursive process, the algorithm avoids the combinatorial explosion problem of exhaustively searching all possible paths and efficiently searches the entire beam echo space. Finally, the endpoint with the maximum cumulative probability is found at the last beam, and by backtracking the predecessor node records, a globally optimal depth path is obtained that runs through all beams and simultaneously achieves the optimal conditions of strongest echo intensity and most continuous terrain.

[0065] As described in step S406 above, the globally optimal depth path is used as a new sounding sequence and input into the one-dimensional Kalman filter bank to re-execute depth value prediction and depth threshold calculation. The globally optimal depth path is the high-precision sounding sequence reconstructed by the Viterbi algorithm, which replaces the original invalid erroneous data. This new sequence is used as the observation input and re-injected into the one-dimensional Kalman filter bank configured independently for each beam. Using this reliable initial value, the Kalman filter immediately updates its internal state estimate and error covariance, thereby accurately predicting the depth value for the next cycle and recalculating the upper and lower depth thresholds based on the prediction results. This process achieves seamless switching and autonomous reset from the unlocked state to the normal tracking state.

[0066] In one embodiment, the arrival time t and its time estimation standard deviation are... The arrival time (t) is extracted from the echo data of each beam using either the amplitude bottom detection method or the phase bottom detection method. The arrival time (t) refers to the time interval from the time of transmission to the reception of the peak or characteristic point of the echo from the seabed topographic target, directly corresponding to the one-way propagation time between the sonar and the seabed. The time estimation standard deviation (δt) is a quantitative representation of the uncertainty in the arrival time measurement, reflecting the degree of dispersion in the detection algorithm's determination of the echo arrival time under noise interference. The arrival time (t) and its time estimation standard deviation (δt) are obtained by preprocessing the echo data of each beam and then using either the amplitude bottom detection method or the phase bottom detection method. Specifically, the amplitude bottom detection method determines t by finding the maximum value of the echo envelope or the zero-crossing point of its derivative, while the phase bottom detection method uses the phase jump characteristics of the echo data to lock t, and then statistically analyzes the fluctuations of multiple samples or adjacent samples to calculate δt. For example, in a certain detection, for the i-th beam, the system detected that the peak of the echo envelope appeared 0.05 seconds after transmission, i.e., t=0.05s. However, within 10 consecutive sampling windows, the fluctuation range of this peak time resulted in a calculated standard deviation δt of 0.002s.

[0067] Figure 3 An internal structural diagram of an electronic device in one embodiment is shown. This electronic device can specifically be a terminal or a server, and more specifically, a computer device. Figure 3 As shown, the electronic device includes a processor, a memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program. When executed by the processor, this computer program enables the processor to implement an intelligent parameter adjustment method for the depth-sounding sonar. The internal memory may also store a computer program, which, when executed by the processor, enables the processor to implement the intelligent parameter adjustment method for the depth-sounding sonar. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0068] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the following steps: Receive raw data from the current cycle of the multibeam echo sounder; wherein the raw data includes at least the echo sequence of each beam, the received channel signal, and the echo data of each beam. The quality factor of each beam is calculated based on the echo data of each beam. The mean of all beam quality factors is used to determine whether the current bathymetry data is a valid feedback of the actual terrain. If the feedback is a valid response to the actual terrain, the depth measurement sequence of the current period is input into a one-dimensional Kalman filter bank that is independently configured for each beam. The Kalman filtering algorithm is used to predict the depth values ​​of each beam in the next period, and the predicted depth value sequence and its prediction variance are obtained. Calculate the depth threshold for the next period based on the predicted depth value sequence and its prediction variance; Based on depth thresholds and depth measurement sequences, the detection range and transmitted signal format are adjusted; simultaneously, based on the spectral center energy analysis of the received channel signal, the transmitted power is adjusted.

[0069] It solves the technical problems of unmanned platforms being unable to autonomously coordinate and adjust in multi-parameter coupled scenarios and the easy loss of depth threshold lock-up, improves the operational stability of the depth sounding system in complex hydrological environments, the continuity of data acquisition, and the adaptive capability of parameter settings, and realizes intelligent closed-loop control of the entire process.

[0070] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0071] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of protection of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the appended claims.

Claims

1. A method for intelligent parameter adjustment of a depth-sounding sonar, characterized in that, The method includes: Receive raw data from the current cycle of the multibeam echo sounder; wherein the raw data includes at least the echo sequence of each beam, the received channel signal, and the echo data of each beam. The quality factor of each beam is calculated based on the echo data of each beam. The mean of all beam quality factors is used to determine whether the current bathymetry data is a valid feedback of the actual terrain. If the feedback is a valid response to the actual terrain, the depth measurement sequence of the current period is input into a one-dimensional Kalman filter bank that is independently configured for each beam. The Kalman filtering algorithm is used to predict the depth values ​​of each beam in the next period, and the predicted depth value sequence and its prediction variance are obtained. Calculate the depth threshold for the next period based on the predicted depth value sequence and its prediction variance; Based on depth thresholds and depth measurement sequences, the detection range and transmitted signal format are adjusted; simultaneously, based on the spectral center energy analysis of the received channel signal, the transmitted power is adjusted.

2. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, The step of calculating the quality factor of each beam based on the echo data of each beam includes: Obtain the time of arrival of terrain target echoes from each beam echo data. t and its time estimate standard deviation ; Calculate the quality factor of each beam using the following formula: in, d For the corresponding depth value, This represents the standard deviation of the depth estimate corresponding to the depth value. For time-domain signal-to-noise ratio, For quality factor, This represents a common logarithm with base 10.

3. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, The step of determining whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors includes: Calculate the arithmetic mean of the quality factors of all beams in the current cycle; If the arithmetic mean is greater than or equal to a preset threshold, the current depth measurement data is determined to be a valid feedback of the actual terrain. If the arithmetic mean is less than the preset threshold, the current depth measurement data is determined to be invalid.

4. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, The one-dimensional Kalman filter bank contains one-dimensional Kalman filters that are processed in parallel, with the same number of sonar beams. Each one-dimensional Kalman filter corresponds to the depth time series of one beam, and the depth threshold includes an upper depth threshold and a lower depth threshold. The step of calculating the depth threshold for the next period based on the predicted depth value sequence and its prediction variance includes: Upper depth threshold: Lower depth threshold: in, For the first k The depth prediction value for each cycle, For the first k The prediction variance for each period, The coefficient is set according to the expected confidence level. For the first k +1 cycle upper depth threshold, For the first k +1 cycle depth threshold.

5. The intelligent parameter adjustment method for depth-sounding sonar according to claim 4, characterized in that, The step of adjusting the detection range and the transmitted signal form based on the depth threshold and depth measurement sequence includes the following steps: Obtain the maximum lower threshold value and current range of all beams in the current period; wherein, the maximum lower threshold value is the maximum value among the lower depth thresholds of all beams; A first range threshold and a second range threshold are set according to the current range; wherein the first range threshold is greater than the second range threshold; If the maximum lower threshold is less than the first range threshold and greater than the second range threshold, then the current range is maintained. If the maximum lower threshold is less than or equal to the second range threshold, then the current range is reduced. If the maximum lower threshold is greater than or equal to the first range threshold, then the current range is expanded.

6. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, The step of adjusting the detection range and transmission signal format based on depth threshold and depth measurement sequence includes the following steps regarding the transmission signal format: Obtain the water depth value corresponding to the depth measurement sequence; Based on the preset correspondence table between water depth values ​​and signal transmission formats, the corresponding signal transmission format is selected according to the magnitude of the water depth value.

7. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, The step of adjusting the transmit power based on the spectral center energy analysis of the received channel signal includes: Perform a Fourier transform on any valid received channel signal to obtain the spectrum; Calculate the energy value at the center frequency of the signal based on the spectrum. If the energy value at the center frequency decreases for no less than a preset number of consecutive cycles exceeding a preset threshold, the emission power of the current sound source level is reduced; otherwise, the emission power of the current sound source level is maintained.

8. The intelligent parameter adjustment method for depth-sounding sonar according to claim 1, characterized in that, After the step of determining whether the current bathymetry data is a valid feedback of the actual terrain based on the mean of all beam quality factors, the method further includes: If the feedback is not a valid response to the actual terrain, then extract the beam echo data for the current period. The beam echo data is modeled as a hidden Markov model; where the discrete depth points in each beam direction are hidden states and the echo intensity is the observed value. Using the echo intensity of each candidate depth point as the observation value, the emission probability of each hidden state is calculated according to the Gaussian distribution function; where the higher the echo intensity, the greater the emission probability. Using the negative correlation between depth changes between adjacent beams as a constraint, the transition probability between states is calculated according to the Gaussian distribution function; where the smaller the depth change between adjacent beams, the greater the transition probability. The Viterbi algorithm is used for recursive calculation to find the globally optimal depth path that maximizes the cumulative probability. The globally optimal depth path is used as a new depth measurement sequence and input into the one-dimensional Kalman filter bank to re-execute depth value prediction and depth threshold calculation.

9. The intelligent parameter adjustment method for depth-sounding sonar according to claim 2, characterized in that, The time of arrival t and its time estimate standard deviation It is obtained by extracting the echo data of each beam using the amplitude bottom detection method or the phase bottom detection method.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the intelligent parameter adjustment method for the depth sounding sonar as described in any one of claims 1 to 9.

Images