Real-time marine environmental element monitoring system

Abstract

The present application relates to the technical field of environmental monitoring, and particularly relates to a real-time marine environmental element monitoring system, which comprises a collecting unit, a constructing unit, a processing unit and a separating unit; the collecting unit collects environmental data, node parameters and target signals in real time by adapting the collection frequency to the marine depth, combining threshold triggering and external instructions; the constructing unit trains an interference prediction model by simulating the refraction propagation path of interference signals, extracting signal characteristics and fusing multi-dimensional data based on the above data through ray acoustic simulation; the processing unit inputs relevant data into the model to output accurate interference results; and the separating unit strips interference signals from mixed signals through frequency domain conversion, dynamic filtering, time domain reconstruction and verification in combination with the interference results. The present application realizes accurate prediction and separation of interference signals, significantly improves the real-time performance, accuracy and reliability of marine environmental element monitoring, and provides strong technical support for marine resource development.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring technology, and in particular to a real-time marine environmental element monitoring system. Background Technology

[0002] The marine environment is a complex environment characterized by multiple layers, multiple factors, and dynamic changes. Its salinity, current velocity, and other elements vary gradients with depth and fluctuate constantly due to weather, tides, and runoff. Real-time and accurate monitoring of the marine environment is essential for the rational development and efficient utilization of marine resources. In existing technologies, because the ocean is a continuous medium for signal transmission, and signals such as sound waves, vibrations, and pressure changes propagate rapidly between different depths, the nodes for collecting marine environmental data are essentially various sensors. The environmental signals these nodes need to capture propagate through the seawater medium to other depths, leading to interference signals at other depths. Furthermore, the ocean medium exhibits uneven stratification in the vertical direction. Because interference signals refract at different layers, their actual propagation distance is greater than the vertical distance between nodes. The attenuation effect of interference signals is positively correlated with the actual propagation distance, resulting in predicted interference intensity greater than the true value. Additionally, the overlapping spectra of signals collected by different nodes make it difficult to distinguish between interference and target signals, thus hindering direct analysis of the collected signals to obtain reliable monitoring results.

[0003] Therefore, it is necessary to propose a real-time marine environmental element monitoring system to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time marine environmental element monitoring system to address the problem that the ocean is a continuous medium for signal transmission, and signals such as sound waves, vibrations, and pressure changes propagate rapidly between different depths. The nodes that collect marine environmental data are essentially various sensors. The environmental signals these nodes need to capture propagate through the seawater medium to other depths, leading to interference signals when nodes at other depths collect signals. Furthermore, the ocean medium has uneven stratification in the vertical direction, causing interference signals to refract at different layers. Therefore, the actual propagation distance of the interference signal is greater than the vertical distance between nodes, and the attenuation effect of the interference signal is positively correlated with the actual propagation distance. The predicted interference intensity is greater than the true value. In addition, the spectrum of the signals collected by each node overlaps, making it difficult to distinguish between the interference signal and the target signal. Therefore, it is difficult to directly analyze the collected signals to obtain reliable monitoring results.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A real-time marine environmental element monitoring system, comprising: The acquisition unit is used to acquire environmental data and node parameters of all vertically deployed monitoring nodes in the target marine area in real time, and to acquire the target signals that each monitoring node needs to monitor. The building unit is used to construct an interference prediction model based on environmental data and node parameters; The processing unit is used to input environmental data, node parameters and target signals into the interference prediction model to output the interference results of the monitoring node; The separation unit is used to separate the interference signal from the mixed signal collected by the monitoring node and combine the interference results to obtain the target signal that the monitoring node needs to monitor in the mixed signal.

[0006] Preferably, the acquisition unit is used to acquire environmental data within the target marine area in real time, node parameters of all vertically deployed monitoring nodes, and target signals to be monitored by each monitoring node, specifically for: Based on the depth data of the target ocean area, an initial acquisition frequency is assigned to the monitoring nodes, and the node parameters of the monitoring nodes are recorded according to the acquisition frequency. Pre-set target signal thresholds and environmental data thresholds. When a monitoring node detects a target signal threshold or an environmental data threshold trigger, the node's acquisition frequency is automatically adjusted by a factor to acquire node parameters. The system collects target signals from monitoring nodes at preset time intervals. When an external instruction to adjust the monitoring target is received, the system automatically collects the corresponding target signal after the adjustment.

[0007] Preferably, the construction unit is used to construct an interference prediction model based on environmental data and node parameters, specifically for: The adjacent monitoring nodes in the vertical direction of the target monitoring node are identified as interference sources. The characteristics of the target signal of the target monitoring node and the target signal of the adjacent monitoring node are extracted. The overlapping region of the target signal spectrum is obtained based on the characteristics. The characteristics include frequency range, amplitude fluctuation range, initial phase value and spectral peak value. Real-time acquisition of vertical layered medium parameters in the target ocean area; dynamic three-dimensional coordinates of the target monitoring node and adjacent monitoring nodes are obtained through real-time feedback data from monitoring nodes. Based on the layered medium parameters and dynamic three-dimensional coordinates, the refraction and propagation path of the interference signal in different layered media is simulated using ray acoustics. The actual propagation length of the path is calculated. Based on the actual propagation length and the corresponding layered medium parameters, the attenuation of the interference signal is obtained. The interference signal is the target signal of the adjacent monitoring node. Based on the overlapping region of the spectrum of the target signal and the interference signal, the difference features are extracted, and the difference features are input into a pre-set neural network model to output the spectrum difference results between the target signal and the interference signal. By fusing the attenuation of the interference signal and the spectral difference results, the feature vector of the interference signal is predicted, and an interference prediction model is constructed.

[0008] Preferably, the step of simulating the refraction and propagation path of interference signals in different layered media using ray acoustics based on the layered medium parameters and dynamic three-dimensional coordinates is specifically used for: Based on the parameters of the layered media, the sound velocity value of each layered medium is calculated using Wilson's formula. Using the dynamic three-dimensional coordinates of adjacent monitoring nodes as the starting point, and based on the real-time vertical depth of the target monitoring node, the emission direction of the interference signal ray is set, and the emission angle and initial sound velocity of the ray are determined. Based on the order of the layered medium from the adjacent monitoring node to the target monitoring node, and combining the sound velocity value, emission direction, emission angle of the sound ray and the initial sound velocity, the refraction angle of the sound ray of the interference signal in each layered medium is calculated layer by layer to determine the sound ray propagation trajectory in that layered medium layer. Based on the determined sound ray propagation trajectory in each layered medium, the straight length of the sound ray propagation trajectory is calculated. The straight lengths of the sound ray propagation trajectories in all layered media are summed to obtain the actual propagation length of the interference signal from the adjacent monitoring node to the target monitoring node. Based on the sound ray refraction angle, sound ray propagation trajectory, and actual propagation length of each layered medium, the refraction and propagation path of the interference signal in different layered media is obtained.

[0009] Preferably, the step of predicting the feature vector of the interference signal by fusing the interference signal attenuation amount and the spectral difference results, and constructing an interference prediction model, is specifically used for: The interference signal attenuation and spectral difference results are integrated to generate an input feature set; The feature set is used as input, and the real perturbation features are used as labels to train a lightweight neural network model. After receiving new input data, the model predicts the amplitude, phase, and spectral distribution of the interference signal, generates feature vectors, and constructs the interference prediction model.

[0010] Preferably, the processing unit is used to input environmental data, node parameters, and target signals into the interference prediction model to output the interference results of the monitoring node, specifically for: Acquire real-time target signals, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes; Extracting features from the target signal; The features, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes are input into the interference prediction model, and the feature vector of the interference signal is output. The interference results are obtained and output based on feature vector analysis.

[0011] Preferably, the separation unit is used to separate the interference signal from the mixed signal based on the mixed signal collected by the monitoring node and the interference result, so as to obtain the target signal that the monitoring node needs to monitor in the mixed signal, specifically used for: The mixed signal collected by the monitoring node and the interference result output by the interference prediction model are obtained. The interference result includes frequency, amplitude, phase and interference intensity. The mixed signal includes the signal of the target monitoring node and the signal of the adjacent monitoring node. The mixed signal is frequency domain transformed to generate its spectrum. Based on the spectral distribution characteristics in the interference results, the spectral interval corresponding to the interference signal is located in the mixed signal spectrum. By combining the frequency, amplitude, and phase parameters of the interference results, a separation filter is constructed to filter and separate the spectral range located in the mixed signal; The stripped mixed signal is reconstructed in the time domain to restore its time domain characteristics and obtain the initial target signal. Based on the interference intensity in the interference results, the initial target signal is verified to obtain the target signal that the monitoring node needs to monitor.

[0012] The technical effects and advantages of the present invention in the above technical solution are as follows: The acquisition unit of this invention adapts its acquisition frequency to ocean depth, dynamically responds to threshold triggering, and responds to external commands to accurately acquire environmental data, node parameters, and target signals. The construction unit relies on layered medium parameters and dynamic three-dimensional coordinates of nodes. It uses ray acoustics to simulate the refraction and propagation path of interference signals, extracts signal features, and fuses multi-dimensional data to train a lightweight interference prediction model. This effectively solves the problems of inaccurate measurement of interference signal propagation distance and difficulty in distinguishing overlapping signal spectra caused by layered ocean media. The processing unit and the separation unit work together to achieve precise removal of interference signals through frequency domain conversion, dynamic separation filtering, and time domain reconstruction, using the accurate interference results output by the model. Finally, it obtains the target signal without interference, improving the real-time performance, accuracy, and reliability of marine environmental element monitoring. Attached Figure Description

[0013] Figure 1 This is a structural diagram of the real-time marine environmental element monitoring system of the present invention. Detailed Implementation

[0014] 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.

[0015] like Figure 1 As shown, this embodiment provides a real-time marine environmental element monitoring system, which includes: The acquisition unit is used to acquire environmental data and node parameters of all vertically deployed monitoring nodes in the target marine area in real time, and to acquire the target signals that each monitoring node needs to monitor. The building unit is used to construct an interference prediction model based on environmental data and node parameters; The processing unit is used to input environmental data, node parameters and target signals into the interference prediction model to output the interference results of the monitoring node; The separation unit is used to separate the interference signal from the mixed signal collected by the monitoring node and combine the interference results to obtain the target signal that the monitoring node needs to monitor in the mixed signal.

[0016] In one embodiment of the present invention, the acquisition unit is used to acquire environmental data within the target marine area, node parameters of all vertically deployed monitoring nodes, and target signals to be monitored by each monitoring node in real time, specifically for: Based on the depth data of the target ocean area, an initial acquisition frequency is assigned to the monitoring nodes, and the node parameters of the monitoring nodes are recorded according to the acquisition frequency. Pre-set target signal thresholds and environmental data thresholds. When a monitoring node detects a target signal threshold or an environmental data threshold trigger, the node's acquisition frequency is automatically adjusted by a factor to acquire node parameters. The system collects target signals from monitoring nodes at preset time intervals. When an external instruction to adjust the monitoring target is received, the system automatically collects the corresponding target signal after the adjustment.

[0017] In this embodiment of the invention, based on the characteristics of the ocean environment's depth gradient variation, to avoid insufficient surface data and redundant deep-layer data due to the same acquisition frequency, initial acquisition frequencies are assigned to monitoring nodes according to different depths in the ocean area. For example, surface 0-50m, mid-layer 50-200m, deep 200m; initial acquisition frequencies for monitoring nodes are 10Hz for the surface, 5Hz for the mid-layer, and 2Hz for the deep layer. Node parameters are recorded according to the acquisition frequencies, including power supply voltage, operating temperature, signal reception strength, attitude offset, etc. When acquiring environmental data and target signals, target signal thresholds and environmental data thresholds are preset, such as pressure signal fluctuations greater than 0.5kPa, vibration signal frequencies falling within the 50-500Hz range, and environmental abrupt change thresholds set to salinity changes greater than 1‰ within 10 minutes. Environmental parameters include seawater salinity, temperature, current velocity, density, and other medium parameters. After the thresholds are set, when a monitoring node detects a target signal threshold or an environmental data threshold triggers, the acquisition frequency of the monitoring node is automatically adjusted by a multiple. Specifically… The monitoring nodes collect target signals and environmental data in real time, and compare them synchronously with the corresponding bound thresholds. If the target signal or environmental data reaches or exceeds the target signal threshold, a frequency adjustment command is triggered, automatically increasing the monitoring node's frequency to twice the initial frequency. For example, if the surface node is adjusted from 10Hz to 20Hz, the timing module is started to record the continuous collection duration after the frequency adjustment, which is 5 minutes by default. After the continuous collection duration ends, the monitoring node's frequency is automatically restored to the initial set frequency. If a threshold trigger is detected again during continuous collection, the timing is restarted, extending the frequency collection duration by twice. For target signal collection, a preset time interval is used to collect the target signal from the monitoring node according to the time interval. When an external command to adjust the monitoring target is received, the corresponding target signal after adjustment is automatically collected, ensuring the continuity of target signal collection while responding promptly to external monitoring needs and ensuring that the collected target signal matches the actual monitoring requirements.

[0018] In one embodiment of the present invention, the construction unit is used to construct an interference prediction model based on environmental data and node parameters, specifically for: The adjacent monitoring nodes in the vertical direction of the target monitoring node are identified as interference sources. The characteristics of the target signal of the target monitoring node and the target signal of the adjacent monitoring node are extracted. The overlapping region of the target signal spectrum is obtained based on the characteristics. The characteristics include frequency range, amplitude fluctuation range, initial phase value and spectral peak value. Real-time acquisition of vertical layered medium parameters in the target ocean area; dynamic three-dimensional coordinates of the target monitoring node and adjacent monitoring nodes are obtained through real-time feedback data from monitoring nodes. Based on the layered medium parameters and dynamic three-dimensional coordinates, the refraction and propagation path of the interference signal in different layered media is simulated using ray acoustics. The actual propagation length of the path is calculated. Based on the actual propagation length and the corresponding layered medium parameters, the attenuation of the interference signal is obtained. The interference signal is the target signal of the adjacent monitoring node. Based on the overlapping region of the spectrum of the target signal and the interference signal, the difference features are extracted, and the difference features are input into a pre-set neural network model to output the spectrum difference results between the target signal and the interference signal. By fusing the attenuation of the interference signal and the spectral difference results, the feature vector of the interference signal is predicted, and an interference prediction model is constructed.

[0019] Based on the layered medium parameters and dynamic three-dimensional coordinates, the method of simulating the refraction and propagation path of interference signals in different layered media using ray acoustics is specifically used for: Based on the parameters of the layered media, the sound velocity value of each layered medium is calculated using Wilson's formula. Using the dynamic three-dimensional coordinates of adjacent monitoring nodes as the starting point, and based on the real-time vertical depth of the target monitoring node, the emission direction of the interference signal ray is set, and the emission angle and initial sound velocity of the ray are determined. Based on the order of the layered medium from the adjacent monitoring node to the target monitoring node, and combining the sound velocity value, emission direction, emission angle of the sound ray and the initial sound velocity, the refraction angle of the sound ray of the interference signal in each layered medium is calculated layer by layer to determine the sound ray propagation trajectory in that layered medium layer. Based on the determined sound ray propagation trajectory in each layered medium, the straight length of the sound ray propagation trajectory is calculated. The straight lengths of the sound ray propagation trajectories in all layered media are summed to obtain the actual propagation length of the interference signal from the adjacent monitoring node to the target monitoring node. Based on the sound ray refraction angle, sound ray propagation trajectory, and actual propagation length of each layered medium, the refraction and propagation path of the interference signal in different layered media is obtained.

[0020] The interference signal feature vector is predicted by fusing the interference signal attenuation amount and spectral difference results, and an interference prediction model is constructed, which is specifically used for: The interference signal attenuation and spectral difference results are integrated to generate an input feature set; The feature set is used as input, and the real perturbation features are used as labels to train a lightweight neural network model. After receiving new input data, the model predicts the amplitude, phase, and spectral distribution of the interference signal, generates feature vectors, and constructs the interference prediction model.

[0021] In this embodiment of the invention, an interference prediction model is constructed. This model is primarily used to extract the target signal (interference signal) from adjacent monitoring nodes in the vertical direction when acquiring target signals from target monitoring nodes. This allows the model to extract the target signal from the mixed signal of the target monitoring node. By identifying adjacent monitoring nodes in the vertical direction as interference sources, the model extracts the characteristics of the target signal and the target signals of adjacent monitoring nodes. These characteristics include frequency range, amplitude fluctuation range, initial phase value, and spectral peak value. The aim is to clarify the basic attributes of the target signal and the interference signal. Through the characteristics of the target signal, the spectral overlap region is obtained. Real-time acquisition of the target ocean area in the vertical direction... The layered medium parameters include seawater salinity, seawater temperature, seawater density, and seawater turbidity. Seawater salinity affects the speed of sound and the degree of signal attenuation. Seawater temperature is directly related to the calculation of the speed of sound, such as the input data for Wilson's formula, and varies with ocean depth. Seawater density affects the damping coefficient and attenuation rate of signal propagation. When the scattering of optical signals or sound waves is involved, seawater turbidity parameters need to be collected to correct the scattering loss of the signal propagation path. Real-time feedback data from monitoring nodes, including the node's own positioning data, distance data between nodes, and environmental data, is used to obtain the dynamic three-dimensional coordinates of the target monitoring node and adjacent monitoring nodes. Specifically, the monitoring nodes receive real-time feedback data on their own... The system uses positioning data (including vertical depth and horizontal offset), relative distance data between nodes, and environmental data. Environmental data is used to compensate for errors in the positioning sensors, thus correcting the node's own positioning data. Combining the corrected self-positioning data with the relative distance data between nodes, a 3D coordinate conversion algorithm is used to initially calculate the 3D coordinates of the target monitoring node and its neighboring nodes. After obtaining the 3D coordinates, the initial coordinates are verified using distance measurement data exchanged between nodes. Finally, the dynamic 3D coordinates of the target monitoring node and its neighboring nodes are output. Based on the layered medium parameters and the dynamic 3D coordinates, X-ray acoustics is used to simulate the refraction and propagation path of interference signals in different layered media. Specifically, based on the layered medium parameters, Wilson's formula is used to calculate the refraction and propagation path of interference signals in each layered medium. The sound velocity value is calculated using the stratified seawater temperature and salinity as input parameters. These parameters are obtained through the stratified medium parameters. The sound velocity of each stratified medium is calculated through a nonlinear combination of temperature and salinity. Then, using the dynamic three-dimensional coordinates of adjacent monitoring nodes as the starting point, and based on the real-time vertical depth of the target monitoring node, the emission direction of the interference signal ray is set, and the emission angle and initial sound velocity of the ray are determined. This is to provide initial physical conditions for ray acoustic simulation of the interference signal refraction and propagation path, ensuring that subsequent layer-by-layer calculations of the refraction angle, determination of the propagation trajectory, and derivation of the actual propagation length have corresponding starting direction and initial velocity information. This is based on the order of the stratified medium from the adjacent monitoring nodes to the target monitoring node.By combining the sound velocity, emission direction, emission angle of the sound ray, and initial sound velocity, the refraction angle of the interference signal's sound ray in each layer of the medium is calculated layer by layer to determine the sound ray propagation trajectory within that layer. Based on the determined sound ray propagation trajectory, the straight-line length of the sound ray propagation trajectory is calculated. Then, the straight-line lengths of the sound ray propagation trajectories in all layers are summed to obtain the actual propagation length of the interference signal from the adjacent monitoring node to the target monitoring node. Combining the sound ray refraction angle, sound ray propagation trajectory, and actual propagation length of each layer, the refracted propagation path of the interference signal is obtained. Through the actual propagation length and the corresponding layered medium parameters, the interference signal attenuation is obtained. This interference signal attenuation refers to the reduction in the distance between the target signal (i.e., the interference signal) at the adjacent monitoring node. During the propagation of the signal through the vertically layered ocean medium to the target monitoring node, the quantified value of the signal amplitude attenuation caused by factors such as propagation distance and the characteristics of the layered medium reflects the actual amplitude strength of the interference signal when it reaches the target node. By extracting the difference features from the spectral overlap region between the target signal and the interference signal, these features are distinguishable characteristics exhibited by the target signal at the target monitoring node and the interference signal at adjacent monitoring nodes within the spectral overlap region. Specific features include the phase difference change rate, spectral peak time delay, difference in amplitude rise / fall slope, and difference in spectral harmonic distribution. These difference features are then input into a lightweight convolutional neural network, which outputs a spectral feature matching degree to quantify the two types of interference signals. The degree of overlap of the signals in terms of spectral peaks and frequency distribution, the quantized value of phase difference (used to output the mean and fluctuation range of the phase difference between the two types of signals at corresponding frequency points), the amplitude attenuation difference coefficient (used to reflect the difference in amplitude attenuation patterns of the two types of signals during propagation), the quantized difference between the attenuation of the interference signal and the amplitude change of the target signal, and the spectral separation threshold (used to output the minimum energy threshold that can achieve spectral separation of the two types of signals) are all considered. Finally, the interference signal attenuation and spectral difference results are fused to construct an interference prediction model. Specifically, the interference signal attenuation and spectral difference results are integrated to generate an input feature set. Feature integration is achieved by normalizing the interference signal attenuation and mapping it to the [0,1] interval, and normalizing the spectral difference... The quantization parameters in the results (such as spectral feature matching degree, phase difference quantization value, amplitude attenuation difference coefficient, and spectral separation threshold) are standardized. If the interference signal attenuation is a single-dimensional value and the spectral difference result is a multi-dimensional parameter set, the interference signal attenuation is expanded into a one-dimensional feature vector, which is then aligned with the multi-dimensional vector of the spectral difference result to obtain a basic feature vector of equal length. The feature vector of the interference signal attenuation and the feature vector of the spectral difference result are concatenated sequentially to obtain the initial fused features, which are then dimensionality-reduced. The fused features are then organized according to the interference signal attenuation-related features and the spectral difference-related features to obtain a feature set. This feature set is used as input, and the real interference features are used as labels to train a lightweight neural network model.The true interference features are the set of actual physical characteristics of interference signals from adjacent monitoring nodes when they reach the target monitoring node. After receiving new input data, the model predicts the amplitude, phase, and spectral distribution of the interference signal, generates feature vectors, and constructs the interference prediction model.

[0022] In one embodiment of the present invention, the processing unit is configured to input environmental data, node parameters, and target signals into an interference prediction model to output interference results for the monitoring node, specifically for: Acquire real-time target signals, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes; Extracting features from the target signal; The features, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes are input into the interference prediction model, and the feature vector of the interference signal is output. The interference results are obtained and output based on feature vector analysis.

[0023] In this embodiment of the invention, in the constructed interference prediction model, the characteristic layered medium parameters of the target signal and the dynamic three-dimensional coordinates of the monitoring nodes are input into the interference prediction model, the feature vector of the interference signal is output, and the feature vector is parsed to obtain the interference result and output it. The parsing process is to receive the feature vector of the interference signal output by the interference prediction model (including core dimension data such as amplitude, phase, and spectral distribution), split the feature vector, extract the corresponding quantization parameters (such as specific amplitude values, phase angle range, spectral peak position and frequency range, etc.), and combine the layered medium parameters and the spatial relationship of the nodes to perform a rationality verification and sorting of the quantization parameters to obtain the interference result. The interference result includes interference type, frequency parameters, amplitude parameters, phase parameters and interference intensity, etc.

[0024] The separation unit is used to separate the interference signal from the mixed signal based on the mixed signal collected by the monitoring node and the interference result, so as to obtain the target signal that the monitoring node needs to monitor in the mixed signal. Specifically, it is used for: The mixed signal collected by the monitoring node and the interference result output by the interference prediction model are obtained. The interference result includes frequency, amplitude, phase and interference intensity. The mixed signal includes the signal of the target monitoring node and the signal of the adjacent monitoring node. The mixed signal is frequency domain transformed to generate its spectrum. Based on the spectral distribution characteristics in the interference results, the spectral interval corresponding to the interference signal is located in the mixed signal spectrum. By combining the frequency, amplitude, and phase parameters of the interference results, a separation filter is constructed to filter and separate the spectral range located in the mixed signal; The stripped mixed signal is reconstructed in the time domain to restore its time domain characteristics and obtain the initial target signal. Based on the interference intensity in the interference results, the initial target signal is verified to obtain the target signal that the monitoring node needs to monitor.

[0025] In this embodiment of the invention, after obtaining the interference result and the mixed signal of the target monitoring node, a fast Fourier transform algorithm is used to convert the continuous mixed signal in the time domain into frequency domain data, generating a mixed signal spectrum containing the correspondence between frequency and amplitude. In the mixed signal spectrum, the spectral range of the interference signal is located, and then a separation filter is constructed using frequency, amplitude, and phase parameters to filter and strip the spectral range. After stripping, an inverse fast Fourier transform is used to convert it from the frequency domain back to the time domain, synchronously compensating for the phase shift and amplitude loss during the frequency domain conversion process, restoring the amplitude characteristics and timing rules of the signal changing over time, and finally obtaining a continuous and complete initial target signal. Based on the interference intensity quantization value in the interference result, a corresponding verification threshold is set, and the actual amplitude and spectral characteristics of the initial target signal are compared with the theoretical value or historical benchmark value of the target signal to check whether there is an interference frequency band exceeding the threshold. If the deviation is within the threshold, the verification is deemed passed, and the target signal is output; if it fails, the separation filter parameters are adjusted and filtered again until a satisfactory target signal is obtained.

[0026] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A real-time marine environmental element monitoring system, characterized in that, The system includes: The acquisition unit is used to acquire environmental data and node parameters of all vertically deployed monitoring nodes in the target marine area in real time, and to acquire the target signals that each monitoring node needs to monitor. The building unit is used to construct an interference prediction model based on environmental data and node parameters; The processing unit is used to input environmental data, node parameters and target signals into the interference prediction model to output the interference results of the monitoring node; The separation unit is used to separate the interference signal from the mixed signal based on the mixed signal collected by the monitoring node and the interference result, so as to obtain the target signal that the monitoring node needs to monitor in the mixed signal; The construction unit is used to construct an interference prediction model based on environmental data and node parameters, specifically for: The adjacent monitoring nodes in the vertical direction of the target monitoring node are identified as interference sources. The characteristics of the target signal of the target monitoring node and the target signal of the adjacent monitoring node are extracted. The overlapping region of the target signal spectrum is obtained based on the characteristics. The characteristics include frequency range, amplitude fluctuation range, initial phase value and spectral peak value. Real-time acquisition of vertical layered medium parameters in the target ocean area; dynamic three-dimensional coordinates of the target monitoring node and adjacent monitoring nodes are obtained through real-time feedback data from monitoring nodes. Based on the layered medium parameters and dynamic three-dimensional coordinates, the refraction and propagation path of the interference signal in different layered media is simulated using ray acoustics. The actual propagation length of the path is calculated. Based on the actual propagation length and the corresponding layered medium parameters, the attenuation of the interference signal is obtained. The interference signal is the target signal of the adjacent monitoring node. Based on the overlapping region of the spectrum of the target signal and the interference signal, the difference features are extracted, and the difference features are input into a pre-set neural network model to output the spectrum difference results between the target signal and the interference signal. By fusing the attenuation of the interference signal and the spectral difference results, the feature vector of the interference signal is predicted, and an interference prediction model is constructed.

2. The real-time marine environmental element monitoring system according to claim 1, characterized in that, The acquisition unit is used to acquire environmental data and node parameters of all vertically deployed monitoring nodes within the target marine area in real time, and to acquire the target signals that each monitoring node needs to monitor. Specifically, it is used for: Based on the depth data of the target ocean area, an initial acquisition frequency is assigned to the monitoring nodes, and the node parameters of the monitoring nodes are recorded according to the acquisition frequency. Pre-set target signal thresholds and environmental data thresholds. When a monitoring node detects a target signal threshold or an environmental data threshold trigger, the node's acquisition frequency is automatically adjusted by a factor to acquire node parameters. The system collects target signals from monitoring nodes at preset time intervals. When an external instruction to adjust the monitoring target is received, the system automatically collects the corresponding target signal after the adjustment.

3. The real-time marine environmental element monitoring system according to claim 1, characterized in that, Based on the layered medium parameters and dynamic three-dimensional coordinates, the method of simulating the refraction and propagation path of interference signals in different layered media using ray acoustics is specifically used for: Based on the parameters of the layered media, the sound velocity value of each layered medium is calculated using Wilson's formula. Using the dynamic three-dimensional coordinates of adjacent monitoring nodes as the starting point, and based on the real-time vertical depth of the target monitoring node, the emission direction of the interference signal ray is set, and the emission angle and initial sound velocity of the ray are determined. Based on the order of the layered medium from the adjacent monitoring node to the target monitoring node, and combining the sound velocity value, emission direction, emission angle of the sound ray and the initial sound velocity, the refraction angle of the sound ray of the interference signal in each layered medium is calculated layer by layer to determine the sound ray propagation trajectory in that layered medium. Based on the determined sound ray propagation trajectory in each layered medium, the straight length of the sound ray propagation trajectory is calculated. The straight lengths of the sound ray propagation trajectories in all layered media are summed to obtain the actual propagation length of the interference signal from the adjacent monitoring node to the target monitoring node. Based on the sound ray refraction angle, sound ray propagation trajectory, and actual propagation length of each layered medium, the refraction and propagation path of the interference signal in different layered media is obtained.

4. The real-time marine environmental element monitoring system according to claim 1, characterized in that, The interference signal feature vector is predicted by fusing the interference signal attenuation amount and spectral difference results, and an interference prediction model is constructed, which is specifically used for: The interference signal attenuation and spectral difference results are integrated to generate an input feature set; The feature set is used as input, and the real perturbation features are used as labels to train a lightweight neural network model. After receiving new input data, the model predicts the amplitude, phase, and spectral distribution of the interference signal, generates feature vectors, and constructs the interference prediction model.

5. The real-time marine environmental element monitoring system according to claim 1, characterized in that, The processing unit is used to input environmental data, node parameters, and target signals into the interference prediction model to output the interference results of the monitoring nodes, specifically for: Acquire real-time target signals, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes; Extracting features from the target signal; The features, layered medium parameters, and dynamic three-dimensional coordinates of the monitoring nodes are input into the interference prediction model, and the feature vector of the interference signal is output. The interference results are obtained and output based on feature vector analysis.

6. The real-time marine environmental element monitoring system according to claim 1, characterized in that, The separation unit is used to separate the interference signal from the mixed signal based on the mixed signal collected by the monitoring node and the interference result, so as to obtain the target signal that the monitoring node needs to monitor in the mixed signal. Specifically, it is used for: The mixed signal collected by the monitoring node and the interference result output by the interference prediction model are obtained. The interference result includes frequency, amplitude, phase and interference intensity. The mixed signal includes the signal of the target monitoring node and the signal of the adjacent monitoring node. The mixed signal is frequency domain transformed to generate its spectrum. Based on the spectral distribution characteristics in the interference results, the spectral interval corresponding to the interference signal is located in the mixed signal spectrum. By combining the frequency, amplitude, and phase parameters of the interference results, a separation filter is constructed to filter and separate the spectral range located in the mixed signal; The stripped mixed signal is reconstructed in the time domain to restore its time domain characteristics and obtain the initial target signal. Based on the interference intensity in the interference results, the initial target signal is verified to obtain the target signal that the monitoring node needs to monitor.

Images