River sound tomography multi-peak signal matching method, system and equipment based on two channels and medium

Abstract

The invention relates to a river sound tomography multi-peak signal matching method, system and device based on two channels and a medium, and the method comprises the steps: carrying out the two-channel discrete sampling and time sequence synchronous correction of a river sound tomography signal, and obtaining a two-channel discrete signal sequence; according to a signal-to-noise ratio adaptive threshold, performing multi-scale sliding window peak detection on the dual-channel discrete signal sequence to obtain a peak point in each channel and a corresponding time position; modeling a graph matching optimization model with constraints, and carrying out peak value pairing by adopting a multi-stage dynamic programming matching strategy to obtain an effective peak value pair; and constructing a propagation time difference sequence based on the effective peak pairs and the corresponding propagation time differences, performing quality inspection and abnormal value elimination on the propagation time difference sequence, and outputting propagation time difference data. According to the method, the problems of poor adaptability and low matching precision of a traditional acoustic tomography signal matching method in a complex river environment are solved, and the reliability and availability of propagation time difference data are improved.

Description

Technical Field

[0001] This invention relates to the field of acoustic signal processing technology, and in particular to a method, system, device and medium for matching multi-peak signals in river acoustic tomography based on dual channels. Background Technology

[0002] River acoustic tomography is a non-contact monitoring method that retrieves key hydrological parameters such as flow velocity, temperature, and flow rate based on the propagation characteristics of sound waves. It has advantages such as wide spatial coverage and continuous monitoring capability. However, in natural river environments, due to factors such as rapid water flow, complex riverbed boundaries, and high concentrations of suspended solids, sound wave signals are prone to multipath effects, signal attenuation, and severe noise interference during propagation. This results in multiple peaks in the received signal, low signal-to-noise ratio, and waveform distortion, significantly increasing the difficulty of signal processing.

[0003] Currently, traditional acoustic tomography signal matching methods are mostly derived from marine acoustic tomography techniques, such as peak matching based on the maximum cross-correlation coefficient or ray tracing matching. These methods perform well in relatively stable aquatic environments such as the ocean, but their applicability is limited in dynamic river environments with many interference factors. Specifically, under low signal-to-noise ratio conditions, single peak detection is prone to missed detections or false detections; when multiple peaks coexist, the matching process is prone to data ambiguity, making it difficult to guarantee the correctness of cross-channel peak pairs; in addition, existing methods have poor adaptability to the non-stationary characteristics of signals and time-varying noise, making it difficult to meet the needs of real-time, high-precision river monitoring. Especially in large-scale, high-resolution hydrological observation scenarios, traditional methods have low processing efficiency and high mismatch rates, leading to a decrease in the reliability of inversion data and failing to meet the requirements of real-time data and accuracy in modern smart water conservancy systems.

[0004] Therefore, there is an urgent need in this field for a multi-peak signal matching method that can adapt to the complex acoustic environment of rivers, has strong anti-interference capabilities, and high matching accuracy. Summary of the Invention

[0005] (a) Technical problems to be solved In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method, system, device and medium for matching multi-peak signals in river acoustic tomography based on dual channels, which solves the technical problems of poor adaptability and low matching accuracy of traditional acoustic tomography signal matching methods in river environments.

[0006] (II) Technical Solution To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, embodiments of the present invention provide a method for matching multi-peak signals in river acoustic tomography based on a dual-channel system, comprising: The acquired river acoustic tomography signal was discretized and sampled in two channels, and the two channels of discrete signals were time-synchronized and corrected to obtain a synchronized two-channel discrete signal sequence. Based on the time-domain statistical characteristics of the dual-channel discrete signal sequence, the signal-to-noise ratio adaptive threshold is obtained, and multi-scale sliding window peak detection is performed on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel. The peak matching problem between the two channels is modeled as a graph matching optimization model, and a multi-stage dynamic programming matching strategy is used to perform peak pairing to obtain effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint. Based on effective peak pairs and their corresponding propagation time differences, a propagation time difference sequence is constructed. A local outlier detection strategy based on a sliding window is used to perform quality checks and outlier removal on the propagation time difference sequence, and the output is propagation time difference data that can be used for river information inversion.

[0007] Optionally, the acquired river acoustic tomography signal is discretized and sampled in two channels, and the resulting two-channel discrete signals are time-synchronized and corrected to obtain a synchronized two-channel discrete signal sequence including: River acoustic tomography signals are acquired using a pre-set river acoustic tomography device, and the river acoustic tomography signals are pre-processed, including filtering, noise reduction, and baseline calibration. Based on the set sampling frequency, the preprocessed river acoustic tomography signal is discretized in two channels to obtain two-channel discrete signals; Based on the preset maximum allowable error of the system and the acoustic center frequency of the river acoustic tomography signal, the two-channel discrete signals are time-synchronized and corrected to obtain a synchronized dual-channel discrete signal sequence.

[0008] Optionally, based on the signal-to-noise ratio adaptive threshold obtained from the time-domain statistical characteristics of the dual-channel discrete signal sequence, multi-scale sliding window peak detection is performed on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel, including: For each channel's discrete signal sequence, a sliding window is used at multiple time scales to obtain the local amplitude statistics of the discrete signal sequence; Based on the mean and standard deviation of local amplitude statistics, a weighted fusion is performed using the global baseline threshold of the discrete signal sequence to generate a dynamic detection threshold that is adapted to the signal-to-noise ratio of the current discrete signal sequence. At each time scale, the signal amplitude within the sliding window is compared with the corresponding dynamic detection threshold, and candidate peak points are obtained based on the comparison results; Candidate peak points detected at different scales are fused and redundancy is removed, and spurious peak points that are unstable at multiple scales are eliminated to obtain the peak points and their corresponding time positions in each channel.

[0009] Optionally, the peak matching problem between the two channels is modeled as a graph matching optimization model, and a multi-stage dynamic programming matching strategy is used for peak pairing to obtain effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint, including: Using the set of peak points in the first channel as the starting set and the set of peak points in the second channel as the ending set, a bipartite graph model is constructed. The weight of the edges in the graph is determined by the amplitude difference between the corresponding two peak points. The maximum allowable time difference between peak pairs is used as a delay consistency constraint, and the range of amplitude ratios between peak pairs is set as an amplitude correlation constraint. Edges that do not meet the constraints are removed from the bipartite graph. The peak matching problem that satisfies the constraints is transformed into finding the matching combination with the optimal total weight in a bipartite graph; By using a multi-stage dynamic programming strategy, the matching process is divided into multiple overlapping stages in chronological order. The optimal matching pair is solved locally in each stage, and the global optimal peak matching sequence is gradually constructed through the state transition between stages. Backtrack and verify the globally optimal peak matching sequence to determine that each peak point is matched at most once, and output the final set of valid peak pairs.

[0010] Optionally, a multi-stage dynamic programming strategy is used to divide the matching process into multiple overlapping stages in chronological order. Within each stage, the optimal matching pair is locally solved, and a globally optimal peak matching sequence is gradually constructed through state transitions between stages, including: In the first stage, the peak points in the starting point set are traversed and compared with all peak points in the ending point set, and all peak point pairs with equal values ​​are obtained to form the initial peak matching sequence. In the second stage, based on the preset threshold constraints of time delay tolerance and amplitude similarity, peak point pairs that meet the threshold constraints are searched in the starting point set and the ending point set, and the obtained peak point pairs are added to the initial peak matching sequence to obtain the extended peak matching sequence. In the third stage, the extended peak matching sequence is used as the initial solution to perform global optimization on the matching path across the entire time axis. The optimization objectives are path smoothness and global matching degree, and the global optimal peak matching sequence is obtained by solving the problem.

[0011] Optionally, the globally optimal peak matching sequence is backtracked and validated to determine that each peak point is matched at most once, and the final set of valid peak pairs is output, including: Backtrack the globally optimal peak matching sequence and determine whether any peak point appears in multiple peak pairs; When the same peak point appears in multiple peak pairs, the peak pair with the best weight is retained according to the weight of the current peak point in the duplicate peak pair, and the remaining duplicate peak pairs are removed to obtain the deduplicated globally optimal peak matching sequence. The waveform matching verification is performed on the deduplicated global optimal peak matching sequence to obtain the signal waveform correlation coefficient of each peak pair at the corresponding time position in the dual-channel discrete signal sequence. If the correlation coefficient of the signal waveform is lower than the preset waveform similarity threshold, the current peak pair is determined to be a false match and is removed from the global optimal peak matching sequence to obtain the final set of valid peak pairs.

[0012] Optionally, based on effective peak pairs and their corresponding propagation time differences, a propagation time difference sequence is constructed. A sliding window-based local outlier detection strategy is then used to perform quality checks and outlier removal on the propagation time difference sequence. The output propagation time difference data that can be used for river information inversion includes: Arrange the propagation time differences corresponding to the effective peak pairs in chronological order to obtain the original propagation time difference sequence; The local outlier detection strategy based on sliding window sets the sliding window size and step size, and calculates the local outlier of each propagation time difference data in the original sequence within each sliding window. The local outlier factor is compared with the preset outlier threshold. If the local outlier factor of the propagation time difference data exceeds the outlier threshold, the current propagation time difference data is determined to be an outlier. Remove all propagation time difference data marked as outliers from the original propagation time difference sequence, and then perform linear interpolation to fill in the missing propagation time difference sequence after removing outliers, to obtain the propagation time difference sequence. The variance and signal-to-noise ratio of the propagation time difference sequence are tested for quality, and the propagation time difference sequence that meets the preset quality threshold in both variance and signal-to-noise ratio is identified as propagation time difference data that can be used for river information inversion.

[0013] Secondly, embodiments of the present invention provide a dual-channel river acoustic tomography multi-peak signal matching system, comprising: The signal preprocessing and synchronization module is used to perform dual-channel discretization sampling on the acquired river acoustic tomography signal and perform time-series synchronization correction on the obtained two-channel discrete signals to obtain a synchronized dual-channel discrete signal sequence. The adaptive multi-scale peak detection module is used to perform multi-scale sliding window peak detection on the dual-channel discrete signal sequence based on the signal-to-noise ratio adaptive threshold obtained from the time-domain statistical characteristics of the dual-channel discrete signal sequence, so as to obtain the peak points and corresponding time positions in each channel. The multi-stage constrained peak matching module is used to model the peak matching problem between two channels as a graph matching optimization model, and to use a multi-stage dynamic programming matching strategy to perform peak pairing, thereby obtaining effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint. The peak matching signal output module is used to construct a propagation time difference sequence based on effective peak pairs and their corresponding propagation time differences. It also employs a sliding window-based local outlier detection strategy to perform quality checks and outlier removal on the propagation time difference sequence, outputting propagation time difference data that can be used for river information inversion.

[0014] Thirdly, embodiments of the present invention provide a dual-channel river acoustic tomography multi-peak signal matching device, comprising: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the dual-channel river acoustic tomography multi-peak signal matching method described above.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method for matching multi-peak signals in river acoustic tomography based on a dual-channel architecture.

[0016] (III) Beneficial Effects This invention proposes a dual-channel river acoustic tomography multi-peak signal matching method, which has the following significant advantages compared to existing technologies: First, this invention effectively overcomes the limitations of traditional single-channel detection in complex river environments by introducing a dual-channel collaborative verification mechanism, thereby improving the accuracy of peak detection. Simultaneously, the intelligent matching algorithm based on a graph matching optimization model and a multi-stage dynamic programming strategy significantly improves the accuracy of peak pairing and resolves the matching ambiguity problem when multiple peaks coexist.

[0017] Next, for complex river environments, this invention adopts a multi-scale analysis framework to deal with sound wave multipath propagation. Based on multiple constraint mechanisms such as time delay consistency, amplitude correlation and waveform similarity, the physical rationality of the matching peak pairs is ensured, so that the system can still maintain high matching accuracy under low signal-to-noise ratio conditions.

[0018] Then, this invention constructs a complete quality control chain, from signal synchronization, peak detection, and matching verification to outlier detection and repair of the final propagation time difference sequence. This system makes the data processing process transparent and traceable, and can quickly identify the source of problems when data quality fluctuates, reducing the difficulty of system maintenance and reliance on manual intervention.

[0019] Finally, this invention can generate high-precision and high-reliability propagation time difference data by matching effective peak pairs, laying a solid data foundation for subsequent data analysis applications such as accurate velocity field inversion, flow estimation, and temperature monitoring, which is conducive to comprehensively improving the automation and intelligence level of hydrological monitoring. At the same time, this invention can adapt to monitoring tasks with different river widths, flow velocities, and hydrological conditions by adjusting parameters, making it promising for applications in water resource management, flood warning, and water ecological assessment. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a dual-channel river acoustic tomography multi-peak signal matching method according to an embodiment of the present invention. Figure 2 A schematic diagram illustrating the extraction of a set of multi-peak river acoustic tomography signals according to an embodiment of the present invention; Figure 3 A set of multi-peak matching example diagrams of dual-channel river acoustic tomography signals provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the inversion effect of river flow velocity and flow rate based on propagation time difference data, provided in an embodiment of the present invention. Detailed Implementation

[0021] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] refer to Figures 1 to 4As shown in the embodiment of the present invention, a method for matching multi-peak signals in river acoustic tomography based on dual channels is proposed. This method includes: discretizing the acquired river acoustic tomography signal into two channels and performing time-series synchronization correction on the two channels to obtain a synchronized dual-channel discrete signal sequence; using an adaptive signal-to-noise ratio threshold obtained from the temporal statistical characteristics of the dual-channel discrete signal sequence, performing multi-scale sliding window peak detection on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel; modeling the peak matching problem between the two channels as a graph matching optimization model, and employing a multi-stage dynamic programming matching strategy to perform peak pairing, obtaining effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint; constructing a propagation time difference sequence based on the effective peak pairs and their corresponding propagation time differences, and using a sliding window-based local outlier detection strategy to perform quality checks and outlier removal on the propagation time difference sequence, outputting propagation time difference data that can be used for river information inversion.

[0023] This embodiment effectively overcomes the limitations of traditional single-channel detection in complex river environments by introducing a dual-channel collaborative verification mechanism, thus improving the accuracy of peak detection. Simultaneously, an intelligent matching algorithm based on a graph matching optimization model and a multi-stage dynamic programming strategy significantly improves the accuracy of peak pairing and resolves matching ambiguities when multiple peaks coexist. Next, for complex river environments, this embodiment employs a multi-scale analysis framework to address acoustic multipath propagation. Multiple constraint mechanisms based on time delay consistency, amplitude correlation, and waveform similarity ensure the physical rationality of matched peak pairs, enabling the system to maintain high matching accuracy even under low signal-to-noise ratio conditions. Then, this embodiment constructs a complete quality control chain, from signal synchronization, peak detection, and matching verification to outlier detection and repair of the final propagation time difference sequence. This system makes the data processing process transparent and traceable, enabling rapid identification of the source of problems when data quality fluctuates, reducing the difficulty of system maintenance and reliance on manual intervention. Finally, this embodiment can generate high-precision and high-reliability propagation time difference data by matching effective peak pairs, which lays a solid data foundation for subsequent data analysis applications such as accurate velocity field inversion, flow estimation, and temperature monitoring. This is conducive to comprehensively improving the automation and intelligence level of hydrological monitoring. Furthermore, this embodiment can adapt to monitoring tasks with different river widths, flow velocities, and hydrological conditions by adjusting parameters, making it have broad application prospects in water resource management, flood warning, and water ecological assessment.

[0024] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0025] Specifically, refer to Figure 1 As shown, this embodiment proposes a dual-channel river acoustic tomography multi-peak signal matching method, which may include the following steps S100 to S400: S100. The acquired river acoustic tomography signal is discretized and sampled in two channels, and the two channels of discrete signals are time-synchronized and corrected to obtain a synchronized two-channel discrete signal sequence.

[0026] In this embodiment, a dual-channel collaborative verification mechanism combining dual-channel discretization sampling and time-series synchronous correction is employed for river acoustic tomography signals. Specifically, the received river acoustic tomography signals are first synchronously discretized using two independent acquisition channels to obtain two time-domain discrete sequences. Subsequently, time-series synchronous correction is performed based on the system's maximum allowable error and the acoustic center frequency of the river acoustic tomography signals to eliminate time deviations introduced by sensor position differences or inconsistent hardware responses, thereby obtaining a precisely aligned dual-channel discrete signal sequence. This method effectively overcomes the limitations of traditional single-channel detection in complex river environments, which is susceptible to noise interference, multipath effects, and signal attenuation. Through cross-verification and complementary enhancement of the dual signals, the accuracy of identifying effective peak values ​​and the robustness of detection are significantly improved.

[0027] In this embodiment, step S100 may include the following sub-steps S110 to S130: S110. Acquire river acoustic tomography signals using a preset river acoustic tomography device, and perform preprocessing on the river acoustic tomography signals, including filtering, noise reduction, and baseline calibration.

[0028] Furthermore, the raw river acoustic tomography signals acquired by the river acoustic tomography equipment typically contain environmental noise and interference introduced by the electronic equipment itself. Therefore, in this embodiment, after acquiring the raw river acoustic tomography signals, further data preprocessing is required. First, a bandpass filter is used to perform preliminary filtering of the signal to suppress out-of-band noise. Subsequently, a digital filter can be used for more refined noise reduction, such as using a wavelet transform denoising algorithm to effectively separate noise based on the differences in the characteristics of signal and noise at different scales. Finally, baseline calibration is performed on the denoised signal to eliminate DC bias or low-frequency drift that may exist during signal acquisition.

[0029] S120. Based on the set sampling frequency, the preprocessed river acoustic tomography signal is discretized in two channels to obtain two-channel discrete signals.

[0030] Furthermore, the mathematical expression for the river acoustic tomography signal in the two independent acquisition channels is as follows: (1) In equation (1), S i ( t () represents the acoustic tomography signals of the river from different channels; A i ( t () is a time-varying amplitude function, reflecting the attenuation of sound wave propagation; f 0 represents the center frequency of the sound wave; φ i ( t ) is a delay function that includes propagation delay information.

[0031] Subsequently, continuous dual-channel river acoustic tomography signals were analyzed at a sampling frequency. f s Discretization sampling is performed, and the mathematical expression for discretization sampling is: (2) In equation (2), T s The sampling interval is... T s =1 / f s ; n For sampling quantity, n =1, …… ,N.

[0032] S130. Based on the preset maximum allowable error of the system and the acoustic center frequency of the river acoustic tomography signal, perform time synchronization correction on the two-channel discrete signals to obtain the synchronized dual-channel discrete signal sequence.

[0033] Furthermore, the purpose of timing synchronization correction is to compensate for any minute time delays that may exist between the two acquisition channels. This delay could stem from non-ideal symmetry in the physical location of the sensors, minor differences in ADC startup time, or clock jitter. Therefore, the maximum allowable error of the system is typically preset to be much smaller than the sampling interval, for example, set to less than one-tenth of a sampling interval, depending on application requirements. Time series of two-channel discrete signals. t 1[ k ]and t 2[ k The following timing synchronization constraints should be met: (3) In equation (3), ∆ t sync This represents the maximum allowable error for the system.

[0034] S200. Based on the time-domain statistical characteristics of the dual-channel discrete signal sequence, the signal-to-noise ratio adaptive threshold is obtained, and multi-scale sliding window peak detection is performed on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel.

[0035] In this embodiment, considering the complex river environment, traditional single-scale peak detection methods are prone to missed or false detections due to the susceptibility of acoustic signals to interference from multipath effects, water flow noise, and riverbed reflections during propagation. Therefore, this embodiment employs a multi-scale analysis framework to address signal fluctuations and distortions caused by multipath propagation. By analyzing signal characteristics at different time scales, it enhances the ability to identify true signal peaks and suppresses spurious peaks caused by noise. (Reference) Figure 2 As shown, it is a schematic diagram of a set of examples of multi-peak river acoustic tomography signal extraction.

[0036] In this embodiment, step S200 may include the following sub-steps S210 to S240: S210. For each channel's discrete signal sequence, the local amplitude statistics of the discrete signal sequence are obtained using a sliding window at multiple time scales.

[0037] S220. Based on the mean and standard deviation of local amplitude statistics, a weighted fusion is performed using the global baseline threshold of the discrete signal sequence to generate a dynamic detection threshold that is compatible with the signal-to-noise ratio of the current discrete signal sequence.

[0038] Furthermore, the mathematical expression for the dynamic detection threshold is: (4) In equation (4), τ[ n Dynamic detection threshold; μ s [ n [This represents the local mean;] σ s [ n [ ] represents the local standard deviation; k is the standard deviation weighting coefficient; α τ is the baseline weighting coefficient. base This is the global baseline threshold.

[0039] S230. At each time scale, compare the signal amplitude within the sliding window with the corresponding dynamic detection threshold, and obtain candidate peak points based on the comparison results.

[0040] Furthermore, candidate peak points pThe conditions that are met can be expressed by formula (5): (5) In equation (5), s[n] is the discrete signal; τ is the dynamic detection threshold.

[0041] S240. The candidate peak points detected at different scales are fused and redundancy is removed, and pseudo-peak points that are unstable at multiple scales are eliminated to obtain the peak points and corresponding time positions in each channel.

[0042] S300: The peak matching problem between the two channels is modeled as a graph matching optimization model, and a multi-stage dynamic programming matching strategy is used to perform peak pairing to obtain effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint.

[0043] In this embodiment, the graph matching optimization model treats each peak as a graph node and models the relationships between possible corresponding peaks in different channels as weighted edges. This transforms the complex peak matching task into a structured graph theory optimization problem, effectively utilizing the temporal and amplitude relationships between peaks to enhance the robustness and interpretability of the matching process. Furthermore, by introducing a multi-stage decision-making mechanism, the global matching task is decomposed into multiple temporally dependent local matching sub-problems, significantly reducing the complexity of the problem-solving process while ensuring both global consistency and local rationality of the matching results. (Reference) Figure 3 As shown, it is an example diagram of multi-peak signal matching in a measured dual-channel river acoustic tomography.

[0044] In this embodiment, step S300 may include the following sub-steps S310 to S350: S310. Using the set of peak points of the first channel as the starting set and the set of peak points of the second channel as the ending set, construct a bipartite graph model. The weight of the edge in the graph is determined by the amplitude difference between the corresponding two peak points.

[0045] Furthermore, the weight calculation of the edges in the graph can be comprehensively designed by combining factors such as the signal-to-noise ratio of the peak points and local waveform characteristics, so as to more accurately reflect the matching possibility between different peak points.

[0046] S320. Based on the preset maximum allowable time difference between peak pairs as a delay consistency constraint, and setting the amplitude ratio range between peak pairs as an amplitude correlation constraint, edges that do not meet the constraint conditions are removed from the bipartite graph.

[0047] Furthermore, based on the constraints of time delay consistency and amplitude correlation, the search space for invalid matches can be effectively reduced, the efficiency of subsequent optimization processes can be improved, and the occurrence of peak matching pairs without physical meaning can be avoided.

[0048] S330. The peak matching problem that satisfies the constraints is transformed into finding the matching combination with the optimal total weight in a bipartite graph.

[0049] Furthermore, the peak matching problem between the two channels is modeled as a constrained graph matching optimization problem: (6) Constraints: (7) In equations (6) and (7), Due to time delay differences, For the difference in amplitude, x ij Let λ be a binary decision variable, representing the magnitude difference weighting system. N , M These represent the total number of peak values ​​in the two channels, respectively.

[0050] S340. Using a multi-stage dynamic programming strategy, the matching process is divided into multiple overlapping stages in chronological order. The optimal matching pair is solved locally in each stage, and the global optimal peak matching sequence is gradually constructed through the state transition between stages.

[0051] Furthermore, this embodiment effectively balances the local accuracy and global consistency of the peak matching process through a phased progressive optimization approach. In the first stage, obvious peak matching pairs are quickly identified through precise numerical comparison; the second stage introduces tolerance matching with dual constraints of time delay and amplitude to enhance the adaptability of peak matching; the third stage smooths and optimizes the matching path from a global perspective to ensure that the final matching result achieves optimal performance in terms of time continuity and waveform consistency. This method progressively refines the matching granularity, avoiding the high computational complexity of global search and effectively suppressing error accumulation caused by local optima through information transfer and correction between stages, thereby significantly improving the accuracy and robustness of peak matching. Specifically, step S340 may include the following sub-steps S341 to S343: S341. In the first stage, the peak points in the starting point set are traversed and compared with all the peak points in the ending point set. All peak point pairs with equal values ​​are obtained to form the initial peak matching sequence.

[0052] To further explain, the expression for the first stage of exact matching is: (8) In equation (8), p 1i , p 1j These are the peak points of the two channels. P 1i , P 1j This is the set of peak points for the two channels.

[0053] S342. In the second stage, based on the preset threshold constraints of time delay tolerance and amplitude similarity, peak point pairs that satisfy the threshold constraints are searched in the starting set and the ending set, and the obtained peak point pairs are added to the initial peak matching sequence to obtain the extended peak matching sequence.

[0054] To further explain, the expression for the second-stage tolerance matching is: (9) In equation (9), δ is the constraint threshold for time delay tolerance and amplitude similarity.

[0055] S343. In the third stage, the extended peak matching sequence is used as the initial solution to perform global optimization on the matching path on the entire time axis. The optimization objectives are path smoothness and global matching degree, and the global optimal peak matching sequence is obtained by solving.

[0056] S350. Backtrack and verify the globally optimal peak matching sequence to determine that each peak point is matched at most once, and output the final set of valid peak pairs.

[0057] Furthermore, to ensure the uniqueness and reliability of the matching results, this embodiment introduces a backtracking deduplication mechanism and waveform similarity verification after obtaining the initial globally optimal peak matching sequence. This eliminates potential one-to-many matches and false matches, thereby improving the accuracy and robustness of the peak pairing results. Specifically, step S350 may include the following sub-steps S351 to S355: S352. Backtrack the globally optimal peak matching sequence and determine whether any peak point appears in multiple peak pairs.

[0058] S353. When the same peak point appears in multiple peak pairs, the peak pair with the best weight is retained according to the weight of the current peak point in the repeated peak pairs, and the remaining repeated peak pairs are removed to obtain the global optimal peak matching sequence after deduplication.

[0059] S354. Perform waveform matching verification on the deduplicated global optimal peak matching sequence and obtain the signal waveform correlation coefficient of each peak pair at the corresponding time position in the dual-channel discrete signal sequence.

[0060] S355. If the correlation coefficient of the signal waveform is lower than the preset waveform similarity threshold, the current peak pair is determined to be a false match and the current peak pair is removed from the global optimal peak matching sequence to obtain the final set of valid peak pairs.

[0061] This embodiment employs an intelligent matching algorithm based on a graph matching optimization model and a multi-stage dynamic programming strategy, which significantly improves the accuracy of peak pairing and resolves the matching ambiguity problem when multiple peaks coexist. Simultaneously, based on multiple constraint mechanisms including time delay consistency, amplitude correlation, and waveform similarity, it effectively eliminates false matches that do not conform to physical laws, ensuring the physical rationality of matched peak pairs and enabling the system to maintain high matching accuracy even under low signal-to-noise ratio conditions.

[0062] S400. Based on the effective peak pairs and their corresponding propagation time differences, a propagation time difference sequence is constructed. A local outlier detection strategy based on a sliding window is used to perform quality checks and outlier removal on the propagation time difference sequence, and the output propagation time difference data that can be used for river information inversion is generated.

[0063] In this embodiment, a sliding window-based local outlier detection strategy is employed to perform quality checks and outlier removal on the propagation time difference (PTD) sequence, outputting PTD data suitable for river information inversion. By introducing local outlier detection, the system can adaptively identify data anomalies within a local range, avoiding misjudgments or omissions caused by unreasonable global threshold settings, further enhancing the robustness of monitoring data under complex hydrological conditions. The resulting high-quality PTD data lays a solid data foundation for subsequent data analysis applications such as accurate velocity field inversion, flow estimation, and temperature monitoring, contributing to a comprehensive improvement in the automation and intelligence level of hydrological monitoring. (Reference) Figure 4 The diagram shown illustrates the inversion effect of river flow velocity and discharge.

[0064] In this embodiment, step S400 may include the following sub-steps S410 to S450: S410. Arrange the propagation time differences corresponding to the effective peak pairs in chronological order to obtain the original propagation time difference sequence.

[0065] Furthermore, based on the matching peak pairs p 1i , p 1j The formula for calculating the propagation time difference is: (10) In equation (10), t 1i Peak point p 1i The point in time, t 2j Peak point p 2j The point in time, n 2j Peak point p 2j Time and locationn 1i Peak point p 1i Time and location T s The time interval used for discretization.

[0066] S420. A local outlier detection strategy based on a sliding window is proposed, which sets the sliding window size and step size, and calculates the local outlier of each propagation time difference data in the original sequence within each sliding window.

[0067] S430. Compare the local outlier factor with the preset outlier threshold. If the local outlier factor of the propagation time difference data exceeds the outlier threshold, then the current propagation time difference data is determined to be an outlier.

[0068] S440. Remove all propagation time difference data marked as outliers from the original propagation time difference sequence, and perform linear interpolation to fill in the missing propagation time difference sequence after removing outliers, to obtain the propagation time difference sequence.

[0069] S450. Perform quality checks on the variance and signal-to-noise ratio of the propagation time difference sequence, and determine the propagation time difference sequence whose variance and signal-to-noise ratio both meet the preset quality threshold as propagation time difference data that can be used for river information inversion.

[0070] Furthermore, this embodiment proposes a dual-channel river acoustic tomography multi-peak signal matching system, including: The signal preprocessing and synchronization module is used to perform dual-channel discretization sampling on the acquired river acoustic tomography signal and perform time-series synchronization correction on the obtained two-channel discrete signals to obtain a synchronized dual-channel discrete signal sequence.

[0071] The adaptive multi-scale peak detection module is used to perform multi-scale sliding window peak detection on the dual-channel discrete signal sequence based on the signal-to-noise ratio adaptive threshold obtained from the time-domain statistical characteristics of the dual-channel discrete signal sequence, thereby obtaining the peak points and corresponding time positions in each channel.

[0072] The multi-stage constrained peak matching module is used to model the peak matching problem between two channels as a graph matching optimization model, and adopts a multi-stage dynamic programming matching strategy to perform peak pairing, thereby obtaining effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint.

[0073] The peak matching signal output module is used to construct a propagation time difference sequence based on effective peak pairs and their corresponding propagation time differences. It also employs a sliding window-based local outlier detection strategy to perform quality checks and outlier removal on the propagation time difference sequence, outputting propagation time difference data that can be used for river information inversion.

[0074] Furthermore, this embodiment proposes a dual-channel river acoustic tomography multi-peak signal matching device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the dual-channel river acoustic tomography multi-peak signal matching method described above.

[0075] Specifically, the dual-channel river acoustic tomography multi-peak signal matching device is the hardware implementation of the above-described method. It constitutes a complete computing system, where the processor handles all computation and control tasks, and the memory stores the program instructions necessary for implementing the method, the acquired river acoustic tomography signals, intermediate results during processing, and the final multi-peak matching data. This device can be implemented as an embedded system, an industrial computer, or a server node, deployed at the river site or data center. It can process dual-channel acoustic tomography data in real-time or offline, automatically and accurately matching and identifying complex multi-peak signals, greatly improving the automation and reliability of hydrological monitoring.

[0076] Finally, this embodiment proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method for matching multi-peak signals in river acoustic tomography based on a dual-channel architecture.

[0077] Specifically, a computer-readable storage medium is a software carrier that carries the core logic of the method described above, transforming the method into a repeatable, distributable, and deployable physical product. The medium can take various forms, including but not limited to USB flash drives, solid-state drives (SSDs), read-only memory (ROM), random access memory (RAM), and optical discs. When the computer-executable instructions in the storage medium are loaded into a device with processing capabilities and run, the device is given the ability to execute the multi-peak signal matching method. Therefore, the dual-channel river acoustic tomography multi-peak signal matching method proposed in this embodiment can not only be implemented using dedicated equipment but can also be flexibly integrated into existing hydrological monitoring systems or computing platforms in software form, greatly enhancing the universality and ease of application of the technology.

[0078] In summary, this invention proposes a method, system, device, and medium for multi-peak signal matching in river acoustic tomography based on dual channels. This invention constructs a complete, robust, and high-precision multi-peak matching processing scheme for river acoustic tomography signals through dual-channel signal collaborative processing, adaptive multi-scale peak detection, constrained graph matching optimization, and automatic data quality control. This method effectively solves key technical challenges such as the susceptibility of acoustic signals to interference, the difficulty of multi-peak matching, and large fluctuations in data quality in complex river environments, significantly improving the reliability and usability of propagation time difference (PTD) data. Based on the PTD data obtained after multi-peak matching, high-precision applications such as river velocity field inversion, flow estimation, and temperature monitoring can be further realized, providing solid data support and technical assurance for water resource management, flood warning, and water ecological assessment, demonstrating significant practical value and broad application prospects.

[0079] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0080] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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.

[0081] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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.

[0082] It should be noted that in the description of this invention, the word "a" or "an" preceding a component does not exclude the existence of multiple such components. This invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. The use of terms such as first, second, third, etc., is merely for convenience and does not indicate any order. These terms can be understood as part of the component names.

[0083] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0084] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning of the basic inventive concept, can make other changes and modifications to these embodiments.

[0085] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of the invention.

Claims

1. A method for matching multi-peak signals in river acoustic tomography based on dual channels, characterized in that, include: The acquired river acoustic tomography signal was discretized and sampled in two channels, and the two channels of discrete signals were time-synchronized and corrected to obtain a synchronized two-channel discrete signal sequence. Based on the time-domain statistical characteristics of the dual-channel discrete signal sequence, the signal-to-noise ratio adaptive threshold is obtained, and multi-scale sliding window peak detection is performed on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel. The peak matching problem between the two channels is modeled as a graph matching optimization model, and a multi-stage dynamic programming matching strategy is used to perform peak pairing to obtain effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint. Based on effective peak pairs and their corresponding propagation time differences, a propagation time difference sequence is constructed. A local outlier detection strategy based on a sliding window is used to perform quality checks and outlier removal on the propagation time difference sequence, and the output is propagation time difference data that can be used for river information inversion.

2. The method as described in claim 1, characterized in that, The acquired river acoustic tomography signal was discretized and sampled in two channels. The resulting two-channel discrete signals were then time-synchronized and corrected to obtain a synchronized two-channel discrete signal sequence, including: River acoustic tomography signals are acquired using a pre-set river acoustic tomography device, and the river acoustic tomography signals are pre-processed, including filtering, noise reduction, and baseline calibration. Based on the set sampling frequency, the preprocessed river acoustic tomography signal is discretized in two channels to obtain two-channel discrete signals; Based on the preset maximum allowable error of the system and the acoustic center frequency of the river acoustic tomography signal, the two-channel discrete signals are time-synchronized and corrected to obtain a synchronized dual-channel discrete signal sequence.

3. The method as described in claim 1, characterized in that, Based on the signal-to-noise ratio adaptive threshold obtained from the time-domain statistical characteristics of the dual-channel discrete signal sequence, multi-scale sliding window peak detection is performed on the dual-channel discrete signal sequence to obtain the peak points and corresponding time positions in each channel, including: For each channel's discrete signal sequence, a sliding window is used at multiple time scales to obtain the local amplitude statistics of the discrete signal sequence; Based on the mean and standard deviation of local amplitude statistics, a weighted fusion is performed using the global baseline threshold of the discrete signal sequence to generate a dynamic detection threshold that is adapted to the signal-to-noise ratio of the current discrete signal sequence. At each time scale, the signal amplitude within the sliding window is compared with the corresponding dynamic detection threshold, and candidate peak points are obtained based on the comparison results; Candidate peak points detected at different scales are fused and redundancy is removed, and spurious peak points that are unstable at multiple scales are eliminated to obtain the peak points and their corresponding time positions in each channel.

4. The method as described in claim 1, characterized in that, The peak matching problem between two channels is modeled as a graph matching optimization model, and a multi-stage dynamic programming matching strategy is used for peak pairing to obtain effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint, including: Using the set of peak points in the first channel as the starting set and the set of peak points in the second channel as the ending set, a bipartite graph model is constructed. The weight of the edges in the graph is determined by the amplitude difference between the corresponding two peak points. The maximum allowable time difference between peak pairs is used as a delay consistency constraint, and the range of amplitude ratios between peak pairs is set as an amplitude correlation constraint. Edges that do not meet the constraints are removed from the bipartite graph. The peak matching problem that satisfies the constraints is transformed into finding the matching combination with the optimal total weight in a bipartite graph; By using a multi-stage dynamic programming strategy, the matching process is divided into multiple overlapping stages in chronological order. The optimal matching pair is solved locally in each stage, and the global optimal peak matching sequence is gradually constructed through the state transition between stages. Backtrack and verify the globally optimal peak matching sequence to determine that each peak point is matched at most once, and output the final set of valid peak pairs.

5. The method as described in claim 4, characterized in that, Using a multi-stage dynamic programming strategy, the matching process is divided into multiple overlapping stages in chronological order. Within each stage, the optimal matching pair is locally solved, and a globally optimal peak matching sequence is gradually constructed through state transitions between stages, including: In the first stage, the peak points in the starting point set are traversed and compared with all peak points in the ending point set, and all peak point pairs with equal values ​​are obtained to form the initial peak matching sequence. In the second stage, based on the preset threshold constraints of time delay tolerance and amplitude similarity, peak point pairs that satisfy the threshold constraints are searched in the starting set and the ending set, and the obtained peak point pairs are added to the initial peak matching sequence to obtain the extended peak matching sequence. In the third stage, the extended peak matching sequence is used as the initial solution to perform global optimization on the matching path across the entire time axis. The optimization objectives are path smoothness and global matching degree, and the global optimal peak matching sequence is obtained by solving the problem.

6. The method as described in claim 4, characterized in that, Backtracking and validating the globally optimal peak matching sequence determines that each peak point is matched at most once, and outputs the final set of valid peak pairs, including: Backtrack the globally optimal peak matching sequence and determine whether any peak point appears in multiple peak pairs; When the same peak point appears in multiple peak pairs, the peak pair with the best weight is retained according to the weight of the current peak point in the duplicate peak pair, and the remaining duplicate peak pairs are removed to obtain the deduplicated globally optimal peak matching sequence. The waveform matching verification is performed on the deduplicated global optimal peak matching sequence to obtain the signal waveform correlation coefficient of each peak pair at the corresponding time position in the dual-channel discrete signal sequence. If the correlation coefficient of the signal waveform is lower than the preset waveform similarity threshold, the current peak pair is determined to be a false match and removed from the global optimal peak matching sequence to obtain the final set of valid peak pairs.

7. The method as described in claim 1, characterized in that, Based on effective peak pairs and their corresponding propagation time differences, a propagation time difference sequence is constructed. A sliding window-based local outlier detection strategy is then used to perform quality checks and outlier removal on the propagation time difference sequence. The output propagation time difference data that can be used for river information inversion includes: Arrange the propagation time differences corresponding to the effective peak pairs in chronological order to obtain the original propagation time difference sequence; The local outlier detection strategy based on sliding window sets the sliding window size and step size, and calculates the local outlier of each propagation time difference data in the original sequence within each sliding window. The local outlier factor is compared with the preset outlier threshold. If the local outlier factor of the propagation time difference data exceeds the outlier threshold, the current propagation time difference data is determined to be an outlier. Remove all propagation time difference data marked as outliers from the original propagation time difference sequence, and then perform linear interpolation to fill in the missing propagation time difference sequence after removing outliers, to obtain the propagation time difference sequence. The variance and signal-to-noise ratio of the propagation time difference sequence are tested for quality, and the propagation time difference sequence that meets the preset quality threshold in both variance and signal-to-noise ratio is identified as propagation time difference data that can be used for river information inversion.

8. A dual-channel river acoustic tomography multi-peak signal matching system, characterized in that, include: The signal preprocessing and synchronization module is used to perform dual-channel discretization sampling on the acquired river acoustic tomography signal and perform time-series synchronization correction on the obtained two-channel discrete signals to obtain a synchronized dual-channel discrete signal sequence. The adaptive multi-scale peak detection module is used to perform multi-scale sliding window peak detection on the dual-channel discrete signal sequence based on the signal-to-noise ratio adaptive threshold obtained from the time-domain statistical characteristics of the dual-channel discrete signal sequence, and to obtain the peak points and corresponding time positions in each channel. The multi-stage constrained peak matching module is used to model the peak matching problem between two channels as a graph matching optimization model, and to use a multi-stage dynamic programming matching strategy to perform peak pairing, thereby obtaining effective peak pairs that satisfy the time delay consistency constraint and amplitude correlation constraint. The peak matching signal output module is used to construct a propagation time difference sequence based on effective peak pairs and their corresponding propagation time differences. It also employs a sliding window-based local outlier detection strategy to perform quality checks and outlier removal on the propagation time difference sequence, outputting propagation time difference data that can be used for river information inversion.

9. A dual-channel river acoustic tomography multi-peak signal matching device, characterized in that, include: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the dual-channel river acoustic tomography multi-peak signal matching method as described in any one of claims 1-7.

10. A computer-readable storage medium storing computer-executable instructions thereon, characterized in that, When executed by a processor, the computer-executable instructions implement a dual-channel river acoustic tomography multi-peak signal matching method as described in any one of claims 1-7.

Images