Multi-source collaborative regional water environment monitoring system and method based on artificial intelligence

Abstract

The application relates to the technical field of water quality measurement, and discloses a multi-source cooperative regional water environment monitoring system and a monitoring method based on artificial intelligence, which comprises a multi-modal detection module and a central processing module; the central processing module acquires echo envelope signals received by an acoustic sensing unit, calculates the energy attenuation rate of the echo envelope signals in the time domain dimension, determines the distribution parameters of suspended solids and bubbles in a detection area based on a dynamic coupling model to establish an optical transmittance compensation matrix, calibrates response signals collected by an optical sensing unit by using the compensation matrix, and offsets the measurement baseline deviation caused by the scattering of multi-phase substances, the application realizes in-situ physical compensation of an optical detection path through acoustic characteristics, reshapes the optical measurement boundary under a high-turbidity water body environment, eliminates the destructive effect of complex working conditions on the law of spectral analysis, and ensures that the monitoring system outputs objective and true results under extreme water quality conditions.

Description

Technical Field

[0001] This invention belongs to the field of water quality measurement technology, and in particular relates to a multi-source collaborative regional water environment monitoring system and monitoring method based on artificial intelligence. Background Technology

[0002] Currently, multi-source sensor arrays are used to acquire water body physicochemical parameters as the basic means of watershed ecological supervision and water affairs hub scheduling. Existing monitoring systems typically integrate optical detection components, electrochemical detection components, and acoustic detection components. By collecting discrete physical signals such as dissolved oxygen, ammonia nitrogen, heavy metals, and turbidity, these signals are aggregated to the edge gateway, and statistical models or artificial intelligence algorithms are used to perform backend compensation on the raw data, thereby outputting water quality assessment results.

[0003] However, in high-turbidity water environments such as tidal estuaries or mixed industrial wastewater discharge zones, the physical scattering effect of the water medium causes the underlying measurement mechanism to fail. This is because the detection process based on spectral analysis relies on the physical premise that photons have a deterministic optical path in the medium. However, in the presence of high concentrations of suspended matter or dense microbubbles, the detection beam undergoes multiple diffuse scatterings, causing random folding and nonlinear extension of the photon's flight path before reaching the receiving unit. This physical uncertainty in the optical path introduces a large number of hysteretic photons carrying path noise into the original signal captured by the receiver, resulting in nonlinear drift of the measurement baseline. Existing improvements focus on optimizing the sensor probe packaging structure or shape, using physical isolation or fluid guidance to mitigate interference. Hardware improvements are insufficient to adapt to transient, unsteady, multiphase flow environments. In addition to hardware structural improvements, existing software control logic also has shortcomings. For example, Chinese invention patent application CN118688182A discloses a water TOC measurement method based on acoustically calibrated underwater plasma optical detection, which uses underwater acoustic signals to extract features to calibrate the power instability of laser-induced plasma sources. To address this problem, the conventional approach tends to increase the complexity of the model in the backend to fit distorted data. However, this approach cannot reverse the information entropy loss during the physical transmission stage. When the spectral response energy drops below the level of the environmental background noise, pure mathematical domain compensation is insufficient to restore the true physical attenuation coefficient, leading to false alarms in extreme fluid environments.

[0004] Therefore, the technical problem to be solved by this invention is how to overcome the signal distortion caused by the uncertainty of the detection path in high-turbidity water, and how to achieve the blocking of the energy conversion of noise photons at the source of physical detection by reconstructing the hardware sampling timing boundary of the detection component through cross-modal physical quantities, and outputting a high-purity measurement true value with physical self-consistency. Summary of the Invention

[0005] This invention provides a multi-source collaborative regional water environment monitoring system based on artificial intelligence, comprising: The multimodal detection module is used to acquire water body detection parameter data; the multimodal detection module includes an acoustic sensing unit and an optical sensing unit; The central processing module, connected to the multimodal detection module, is used to perform source compensation based on water dynamic parameters. Internally, the central processing module includes: an echo feature extraction unit, used to acquire the high-frequency backscattered echo envelope signal received by the acoustic sensing unit and calculate the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain; a dynamic parameter mapping unit, used to input the energy attenuation rate into a preset dynamic coupling model, compare the energy attenuation rate with a preset fluid scattering feature library to remove the contribution weights of bubbles and solid suspended particles to acoustic energy dissipation, calculate the suspended matter distribution density and bubble content in the water body of the detection area, and establish an optical transmittance compensation matrix; and a measurement benchmark calibration unit, used to calibrate the fluorescence response electrical signal acquired by the optical sensing unit using the optical transmittance compensation matrix, correct the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles, and reconstruct the true values ​​of detection parameters conforming to the linear boundary of the Lambert-Beer law.

[0006] Preferably, the central processing module is used to adjust the detection frequency through the drive waveform adaptive unit; the internal logic of the drive waveform adaptive unit includes: acquiring real-time flow velocity data of the detection area and comparing the real-time flow velocity data with a preset flow velocity threshold; when the real-time flow velocity data exceeds the preset flow velocity threshold, switching the driving mode of the multi-mode detection module to a combination of wideband continuous scanning mode and moving average feature extraction mode, using the spectral coverage characteristics of the wideband carrier to offset the phase deviation caused by the transient eddy current of the fluid, and making the sampling frequency increase linearly with the increase of the flow velocity gradient, thereby maintaining the sampling spatial continuity of the physical detection domain in the unsteady flow field.

[0007] Preferably, the central processing module further includes a window contamination monitoring module, which monitors the attenuation characteristics of the reflected echo from the surface of the detection window received by the acoustic sensing unit, so as to calculate the amount of physical contamination adhering to the surface of the detection window of the optical sensing unit and correct the optical transmittance compensation matrix accordingly.

[0008] Preferably, the central processing module is used to determine the microbubble density in the detection area based on the amplitude fluctuation frequency of the high-frequency backscattered echo envelope signal, and output a compensation control signal for the intensity of the excitation light source of the optical sensing unit.

[0009] Preferably, the system includes a first monitoring subsystem and a second monitoring subsystem distributed in a distributed manner, with the second monitoring subsystem located downstream of the first monitoring subsystem; the first monitoring subsystem is used to calculate the gradient change rate of its output characteristic signal and send an early warning control signal to the second monitoring subsystem to trigger the second monitoring subsystem to shorten the transmission interval of its internal acoustic sensing unit.

[0010] Preferably, the second monitoring subsystem is used to increase the sampling point density of the physical detection domain after receiving the early warning control signal, and to allocate the computational weight of its internal detection tasks according to the gradient change rate of the characteristic signal.

[0011] Preferably, the multimodal detection module further includes an electrochemical sensing unit; the central processing module is used to perform correlation calculations between the calibrated fluorescence response electrical signal and the conductivity data collected by the electrochemical sensing unit to correct the cross-interference of dissolved ions in the water on the optical detection results.

[0012] Preferably, the central processing module is used to determine the particle size distribution of suspended particles in the water by calculating the second derivative of the energy attenuation rate, and adjust the receiving bandwidth of the optical sensing unit according to the particle size distribution.

[0013] Preferably, the multimodal detection module further includes a self-circulating cleaning module; the central processing module is used to send a high-pressure jet control signal to the self-circulating cleaning module when the calculated energy attenuation rate exceeds a preset blocking threshold.

[0014] A multi-source collaborative regional water environment monitoring method based on artificial intelligence includes the following steps: Step S1101: Obtain multimodal water body detection parameter data; Step S1102: The high-frequency backscattered echo envelope signal received by the acoustic sensing unit is obtained through the echo feature extraction unit, and the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain dimension is calculated. Step S1103: The energy attenuation rate is input into the preset dynamic coupling model through the dynamic parameter mapping unit to determine the distribution density of suspended matter and bubble content in the water body of the detection area, and to establish an optical transmittance compensation matrix. Step S1104: The fluorescence response electrical signal collected by the optical sensing unit is calibrated by the optical transmittance compensation matrix through the measurement reference calibration unit, the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles is corrected, and the true value of the detection parameter conforming to the linear boundary of the Lambert-Beer law is reconstructed.

[0015] Compared with existing technologies, the multi-source collaborative regional water environment monitoring system based on artificial intelligence of this invention has the following advantages: 1. In multi-source collaborative regional water environment monitoring, the number density distribution of suspended particles in the target water body is inverted by extracting the envelope characteristics of the backscattered echo generated by the ultrasonic transducer in the high-frequency band, and the mean free path of the probe photons in the current medium is calculated in real time using this physical quantity. This mechanism establishes a hardware modulation closed loop at the physical detection source, which modulates the multispectral sampling time sequence by the acoustic field attenuation coefficient, so that the integral sampling window of the optical detection component is controlled within the ballistic photon flight time threshold derived from the acoustic characteristics. This cross-modal physical quantity collaborative logic forcibly cuts off the interference photons generated by multiple diffuse scatterings at the sensor hardware level, thereby reshaping the linear measurement boundary that conforms to the Lambert-Beer law under extreme turbidity conditions, avoiding the baseline drift caused by the uncertainty of the physical optical path of traditional monitoring equipment.

[0016] 2. The edge computing gateway dynamically adjusts the synchronization window parameters between the ultrasonic transducer and the multispectral detection component based on the real-time flow velocity parameters output by the fluid dynamics sensing component. When the fluid velocity exceeds a preset threshold, the system automatically switches to a combination of wideband continuous scanning mode and moving average feature vector extraction mode to offset the phase shift caused by transient eddies and irregular bubbles in the acousto-optic coupling model. This dynamic interlocking and feedback compensation of multi-source physical information enables the system to autonomously change the hardware drive waveform according to the fluid dynamics state, thereby improving the measurement signal-to-noise ratio of the device in complex multiphase flow environments without changing the physical probe structure.

[0017] 3. The edge computing gateway monitors the attenuation rate of multiple reflections from the surface of the viewing window received by the ultrasonic transducer component, characterizing the physical contamination level of the optical viewing window in real time, and constructs an optical transmittance compensation matrix accordingly. This mechanism enables in-situ self-calibration of the optical measurement path by the acoustic signal, and algebraically fuses the collected fluorescence response electrical signal with the compensation matrix to offset the detection energy loss caused by biological attachment or physical shielding. This in-situ compensation mechanism based on physical reflection characteristics ensures the objective accuracy of the measurement results and extends the effective operating cycle of the system in harsh field environments. Attached Figure Description

[0018] Figure 1 This is a flowchart of the multi-source collaborative monitoring and optical reference calibration process for the water environment according to the present invention; Figure 2 This is a diagram of the architecture of the multimodal perception and AI-driven adaptive compensation system of this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, bottom, transverse, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between the components in a specific state (as shown in the accompanying drawings), and are only for the convenience of describing this invention, not to require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated.

[0021] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0022] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is 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, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0023] A multi-source collaborative regional water environment monitoring system based on artificial intelligence includes: The multimodal detection module is used to acquire water body detection parameter data; the multimodal detection module includes an acoustic sensing unit and an optical sensing unit; The central processing module, connected to the multimodal detection module, is used to perform source compensation based on water dynamic parameters. Internally, the central processing module includes: an echo feature extraction unit, used to acquire the high-frequency backscattered echo envelope signal received by the acoustic sensing unit and calculate the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain; a dynamic parameter mapping unit, used to input the energy attenuation rate into a preset dynamic coupling model, compare the energy attenuation rate with a preset fluid scattering feature library to remove the contribution weights of bubbles and solid suspended particles to acoustic energy dissipation, calculate the suspended matter distribution density and bubble content in the water body of the detection area, and establish an optical transmittance compensation matrix; and a measurement benchmark calibration unit, used to calibrate the fluorescence response electrical signal acquired by the optical sensing unit using the optical transmittance compensation matrix, correct the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles, and reconstruct the true values ​​of detection parameters conforming to the linear boundary of the Lambert-Beer law.

[0024] Preferably, the central processing module is used to adjust the detection frequency through the drive waveform adaptive unit; the internal logic of the drive waveform adaptive unit includes: acquiring real-time flow velocity data of the detection area and comparing the real-time flow velocity data with a preset flow velocity threshold; when the real-time flow velocity data exceeds the preset flow velocity threshold, switching the driving mode of the multi-mode detection module to a combination of wideband continuous scanning mode and moving average feature extraction mode, using the spectral coverage characteristics of the wideband carrier to offset the phase deviation caused by the transient eddy current of the fluid, and making the sampling frequency increase linearly with the increase of the flow velocity gradient, thereby maintaining the sampling spatial continuity of the physical detection domain in the unsteady flow field.

[0025] Preferably, the central processing module further includes a window contamination monitoring module, which monitors the attenuation characteristics of the reflected echo from the surface of the detection window received by the acoustic sensing unit, so as to calculate the amount of physical contamination adhering to the surface of the detection window of the optical sensing unit and correct the optical transmittance compensation matrix accordingly.

[0026] Preferably, the central processing module is used to determine the microbubble density in the detection area based on the amplitude fluctuation frequency of the high-frequency backscattered echo envelope signal, and output a compensation control signal for the intensity of the excitation light source of the optical sensing unit.

[0027] Preferably, the system includes a first monitoring subsystem and a second monitoring subsystem distributed in a distributed manner, with the second monitoring subsystem located downstream of the first monitoring subsystem; the first monitoring subsystem is used to calculate the gradient change rate of its output characteristic signal and send an early warning control signal to the second monitoring subsystem to trigger the second monitoring subsystem to shorten the transmission interval of its internal acoustic sensing unit.

[0028] Preferably, the second monitoring subsystem is used to increase the sampling point density of the physical detection domain after receiving the early warning control signal, and to allocate the computational weight of its internal detection tasks according to the gradient change rate of the characteristic signal.

[0029] Preferably, the multimodal detection module further includes an electrochemical sensing unit; the central processing module is used to perform correlation calculations between the calibrated fluorescence response electrical signal and the conductivity data collected by the electrochemical sensing unit to correct the cross-interference of dissolved ions in the water on the optical detection results.

[0030] Preferably, the central processing module is used to determine the particle size distribution of suspended particles in the water by calculating the second derivative of the energy attenuation rate, and adjust the receiving bandwidth of the optical sensing unit according to the particle size distribution.

[0031] Preferably, the multimodal detection module further includes a self-circulating cleaning module; the central processing module is used to send a high-pressure jet control signal to the self-circulating cleaning module when the calculated energy attenuation rate exceeds a preset blocking threshold.

[0032] A multi-source collaborative regional water environment monitoring method based on artificial intelligence includes the following steps: Step S1101: Obtain multimodal water body detection parameter data; Step S1102: The high-frequency backscattered echo envelope signal received by the acoustic sensing unit is obtained through the echo feature extraction unit, and the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain dimension is calculated. Step S1103: The energy attenuation rate is input into the preset dynamic coupling model through the dynamic parameter mapping unit to determine the distribution density of suspended matter and bubble content in the water body of the detection area, and to establish an optical transmittance compensation matrix. Step S1104: The fluorescence response electrical signal collected by the optical sensing unit is calibrated by the optical transmittance compensation matrix through the measurement reference calibration unit, the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles is corrected, and the true value of the detection parameter conforming to the linear boundary of the Lambert-Beer law is reconstructed.

[0033] Example 1: In continuous in-situ water environment monitoring in the multiphase mixing zone of a tidal estuary, the dynamic backwater effect of upstream runoff and ocean tides causes the water to carry high concentrations of irregularly suspended solid particles and dense microbubbles, forming a high-turbidity multiphase scattering medium. The excitation beam emitted by the optical sensing unit in the multimodal detection module undergoes multiple diffuse scatterings in this medium, resulting in random folding and nonlinear extension of the actual optical path before the photons reach the receiver. This deviates from the physical premise of spectral analysis based on a single deterministic optical path, causing the original signal captured by the optical sensing unit to be mixed with path noise. Delayed photons induce a nonlinear shift in the optical measurement baseline. The echo feature extraction unit within the central processing module acquires the high-frequency backscattered echo envelope signal received by the acoustic sensing unit, calculates the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain, and the dynamic parameter mapping unit inputs the energy attenuation rate into the dynamic coupling model. The energy attenuation rate is compared with a preset fluid scattering feature library, and the contribution weights of bubbles and suspended solid particles to acoustic energy dissipation are removed. The distribution density of suspended matter and the bubble content in the water body of the detection area are calculated, and an optical transmittance compensation matrix is ​​established accordingly. The time-driven waveform adaptive unit continuously acquires real-time flow velocity data of the detection area and compares it with a preset flow velocity threshold. When the real-time flow velocity data exceeds the preset threshold, the driving mode of the multi-mode detection module is switched to a combination of wideband continuous scanning mode and moving average feature extraction mode. This relies on the spectral coverage characteristics of the wideband carrier to offset the phase deviation caused by transient eddies in the fluid, causing the sampling frequency to increase linearly with the flow velocity gradient. This maintains the sampling spatial continuity of the physical detection domain in the unsteady flow field. The transient eddies in the fluid cause an overall delay in the sound wave propagation path. This leads to local phase loss at the acoustic receiver. Instead of interfering with or eliminating the distorted eddy currents on a physical scale, this system controls the acoustic sensing unit to transmit a broadband carrier containing multiple discrete frequency points. By utilizing the inconsistency of spatial phase shift fading of different frequency components in the same physically distorted flow field, and in conjunction with a moving average algorithm on the receiving side, the echo feature vectors of the multi-frequency channels are superimposed in the incoherent spatial domain. This smooths out the deep fading and phase jump phenomenon of single-frequency signals caused by eddy currents at the electrical signal solution level, achieving an algorithmic equivalent stripping of the phase deviation of the physical flow field.

[0034] The measurement reference calibration unit inside the central processing module uses an optical transmittance compensation matrix to calibrate the fluorescence response electrical signal acquired by the optical sensing unit, corrects the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles, extracts acoustic features to invert the physical microenvironment of the medium, translates physical parameters into an in-situ physical compensation reference for the optical detection link, blocks the energy conversion of physical space noise caused by multiphase flow interference into the electrical signal level, and outputs the true value of the detection parameters that conforms to the linear boundary of the Lambert-Beer law. During the aforementioned synchronous measurement process, considering that the acoustic sensing unit and the optical sensing unit are tightly encapsulated in the same detection panel and located within an equivalent hydrodynamic boundary layer, the central processing module acquires the signal transmitted from the acoustic sensing unit into its own detection window. The residual reflected echo amplitude received after the side-emitted ultrasonic wave is used to calculate the internal reflection energy attenuation caused by the thickening of the acoustic impedance layer of dirt on the outer side of the window. Based on the pre-set acoustic-optical isomorphic window attachment decay cross-calibration table, the current micron-level physical contamination attachment thickness of adjacent optical windows is equivalently mapped. Then, a direct cancellation operation is performed on the corresponding transmission coefficient set of the optical transmittance compensation matrix. For the correlation calculation of fluorescence response electrical signal and conductivity data, the central processing module is pre-set with an algebraic reconstruction model based on the Stern-Walmer fluorescence quenching mechanism. Before actual operation, based on the fact that the active component that triggers fluorescence quenching is usually a specific soluble anion, the system pre-establishes the overall total conductivity of the water body in the target area through multiple sampling titrations. The system employs an empirical polynomial to determine the background concentration of dominant fluorescence quenching ions. Upon acquiring real-time conductivity, the central processing module uses this polynomial to inversely convert the overall conductivity scalar into an equivalent specific quenching ion concentration parameter. This converted parameter is then injected as the quencher variable into the Stern-Wolmer reconstruction model for baseline calculation. The electrochemical sensing unit continuously monitors the initial conductivity of the water. When the initial conductivity exceeds a set trigger threshold, the central processing module retrieves the fluorescence quenching constant, which is linearly positively correlated with the initial conductivity, based on the system's built-in compensation mapping table. The central processing module then multiplies the original fluorescence signal acquired by the optical sensing unit by the product of the fluorescence quenching constant and the initial conductivity, and adds the background amplitude of the original fluorescence signal. The calculated algebraically reconstructed fluorescence response signal eliminates cross-interference from dissolved ions. The specific trigger threshold is determined by extracting the 95th percentile from the three-month historical statistical distribution curve of background dissolved salt concentration in the target monitoring water area. When the system operates as the first monitoring subsystem in a distributed watershed architecture, the central processing module immediately calls the real-time flow velocity data from the multimodal detection module in parallel when the gradient change rate of the detected characteristic signal exceeds the threshold. Combining this with the physical distance vector between the first monitoring subsystem and the downstream second monitoring subsystem pre-stored in memory, the physical lag period of the abnormal pollution plume reaching the downstream section is estimated using the one-dimensional advection diffusion equation. This lag period value is then packaged and sent out along with the early warning control signal.The second monitoring subsystem is instructed to trigger a shortened transmission interval based on the calculated timing window delay, ensuring that the hardware-encrypted acquisition cycle accurately covers the dynamic physical timeline of the contamination plume's passage.

[0035] Example 2: This example addresses the challenge of optical sensing baseline drift in high-turbidity multiphase flow environments. A simulated circulating water flume test platform in a tidal estuary is constructed to obtain in-situ physical compensation parameters. This platform is equipped with flow velocity regulation and bubble generation devices, independently controlling water flow velocity, suspended solids concentration, and microbubble volume fraction. Gaussian white noise with a signal-to-noise ratio of 20dB is injected into the signal receiving front end of the acoustic sensing unit, superimposed with 50Hz power frequency interference harmonics to simulate the electromagnetic disturbance environment in real industrial waters. A preset flow velocity threshold and sampling frequency balance the resolution of multiphase flow transient feature capture and the computational load of the central processing module. When the water flow velocity gradient in the detection area increases, the fluid... The transient eddy current intensifies the spatial distribution changes at the phase interface. To avoid spatial Nyquist sampling aliasing, when the real-time flow velocity data exceeds a preset flow velocity threshold of 1.5 m / s, the sampling frequency increases linearly with the flow velocity gradient. Based on this, four experimental groups were constructed. Control group 1 was a single optical detection architecture without an acoustic sensing unit. Control group 2 had an acoustic sensing unit connected but the drive waveform adaptive unit disconnected and a fixed sampling frequency of 10 Hz was used. The sampling frequency of control group 3 increased with the flow velocity gradient by a factor of 0.1. The experimental groups used a complete system including an acoustic sensing unit and a drive waveform adaptive unit, and the increase factor was set to 0.8 within the preferred working window.

[0036] In the initial measurement conditions, solid suspended matter was injected into the circulating water tank to achieve a turbidity of 800 NTU. The bubble generator was activated to achieve a microbubble volume fraction of 3%, and the water pump was adjusted to stabilize the real-time flow rate at 2.2 m / s. At this point, the original fluorescence response electrical signal captured by the optical sensing unit of control group 1 was physically distorted by diffuse scattering interference, resulting in an initial measurement error of 45.2%. The acoustic sensing unit of the experimental group emitted ultrasonic pulses and received high-frequency backscattered echo envelope signals. The echo feature extraction unit inside the central processing module extracted the energy attenuation rate of the echo envelope signal in the time domain as 12%. The kinetic parameter mapping unit inputs the energy attenuation rate into the kinetic coupling model, compares it with the fluid scattering feature library, and separates the acoustic energy dissipation attenuation component caused by solid suspended matter as 8.2 dB / m and the acoustic energy dissipation attenuation component caused by bubbles as 4.3 dB / m. The central processing module calculates the suspended matter distribution density and bubble content based on the above physical parameters, and establishes an N×N optical transmittance compensation matrix. Since the real-time flow velocity data of 2.2 m / s is greater than the preset flow velocity threshold of 1.5 m / s, the experimental group's drive waveform adaptive unit switches the drive mode to wideband continuous scanning mode and sliding mode. The combined state of the dynamic averaging feature extraction mode utilizes the spectral coverage characteristics of broadband carrier waves to offset the phase deviation caused by transient eddies in the fluid. The sampling frequency is increased from the reference 10Hz to 45Hz with an increment factor of 0.8. Test data shows that when the microbubble volume fraction exceeds 6%, the acoustic energy attenuation rate enters the saturation region, and the slope of the corresponding parametric curve tends to flatten. The experimental group relies on the established optical transmittance compensation matrix to calibrate the fluorescence response electrical signal, suppressing the final optical measurement error to 1.8%. In the control group two, spatial phase misalignment caused by the fixed sampling frequency in the unsteady flow field of 2.2 m / s resulted in measurement errors. The difference reached 16.5%, and the three-factor increment coefficient of the control group was 0.1, which was lower than the theoretical lower limit of 0.5, and was insufficient to cover the spectral shift of transient eddies. The measurement error reached 12.3%. Quantitative experimental data confirmed that single-mode detection or fixed-frequency sampling could not adapt to the physical interference of unsteady high-turbidity multiphase flow. The acoustic feature stripping parameter and the waveform adaptive sampling frequency formed a pre-constraint and dynamic alignment dependency relationship. Multi-source collaborative in-situ physical compensation combined with adaptive dynamic sampling logic eliminated the optical measurement baseline shift caused by multiple scattering of multiphase suspended particles and output the true value of the detection parameter that conforms to the linear boundary of the Lambert-Beer law.

[0037] Example 3: In the in-situ water environment monitoring of a multiphase flow mixing zone in a tidal estuary, the fluctuations in water velocity and the alternating superposition of multiphase scattering medium concentrations cause the fixed velocity threshold setting and the lagging data compensation model to fail to match the transient dynamic evolution of the monitored water area. This results in sampling aliasing when the flow velocity exceeds the preset benchmark, leading to spectral data outputting data that deviates from the Lambert-Beer law. Before the drive waveform adaptive unit switches the drive mode, the central processing module triggers the velocity threshold calibration procedure, controlling the velocity generator to generate a standard velocity field with increasing gradient in the simulated tank. The acoustic sensing unit records the transient eddy current phase interface spatial displacement at each standard velocity node. The central processing module calculates the acoustic phase deviation gradient caused by this spatial displacement, extracts the stationary point where the phase deviation gradient curve deviates from the linear evolution range, and defines the standard velocity corresponding to the stationary point as the preset velocity threshold. In water environment monitoring, the dynamic parameter mapping unit inside the central processing module inputs the acquired time-domain energy attenuation rate into the dynamic coupling model, and uses the inverted acoustic attenuation discretization control equation and acoustic scattering constraint criterion within the model to perform the dynamic parameter mapping. The total energy attenuation rate is separated into a cubic scattering component proportional to the particle size of the suspended solids and a resonant absorption component proportional to the volume of the microbubbles. The dynamic parameter mapping unit solves the discretized control equations based on the known medium acoustic impedance reference in the fluid scattering feature library to calculate the suspended solids distribution density and bubble content in the water body of the detection area. The acoustic sensing unit uses a broadband ultrasonic transducer covering a preset bandwidth. The echo feature extraction unit performs a fast Fourier transform on the received high-frequency backscattered echo envelope signal to extract the total sound wave attenuation coefficients at the first and second center frequencies, respectively. The dynamic parameter mapping unit internally stores a discretized mapping matrix reflecting the fluid scattering characteristics, including a set of calibration coefficients for the fourth power attenuation of suspended particles following Rayleigh scattering characteristic frequencies and a set of calibration coefficients for the attenuation of microbubbles following resonant absorption characteristic frequencies. The dynamic parameter mapping unit substitutes the extracted total sound wave attenuation coefficients at the two center frequencies into a set of two linear attenuation equations consisting of two sets of calibration coefficients, and separates the acoustic energy dissipation cross terms through matrix inversion to solve for the independent suspended solids distribution density and bubble content.

[0038] After completing the basic parameter separation, to further characterize the morphological features of suspended matter in the water, the central processing module re-extracts the original high-frequency backscattered echo envelope signal captured by the acoustic sensing unit, applies a Hanning window to truncate it in time series, and uses Discrete Fast Fourier Transform to map the time-domain feature vector point by point to the frequency domain space, thereby obtaining a broadband echo energy attenuation spectrum curve reflecting the attenuation amplitude fluctuating with continuous frequency bands. The central processing module calculates the second derivative of the broadband echo energy attenuation spectrum with respect to the detection frequency in the frequency domain, locates the cutoff frequency where the second derivative curve crosses zero, and determines the median particle size distribution of the dominant suspended particles in the water body based on the inverse mapping relationship between the acoustic wavelength corresponding to the cutoff frequency and the particle size. Using the median particle size distribution as an index, it consults a preset scattering angle distribution table to determine the corresponding forward scattered light solid angle interval and outputs control... The signal limits the receiving bandwidth of the optical sensing unit to the solid angle range of the forward scattered light. The central processing module delineates the physical space covering the detection optical path of the optical sensing unit based on the extracted distribution density of suspended matter and bubble content. This physical space is discretized into spatial grid nodes. The attenuation coefficient of the photon mean free path at each grid node is calculated based on the local concentration distribution of suspended matter and bubbles. The attenuation coefficients of each grid node are arranged according to the spatial absolute coordinate mapping to generate an optical transmittance compensation matrix of dimension N×N, where N is a positive integer representing the dimension of the matrix. The measurement reference calibration unit obtains the inverse matrix of the optical transmittance compensation matrix, multiplies the fluorescence response electrical signal collected by the optical sensing unit with the inverse matrix, corrects the optical measurement baseline offset caused by diffuse scattering of the excitation beam in the multiphase scattering medium, and outputs the true value of the detection parameter that conforms to the linear boundary of the Lambert-Beer law.

[0039] Example 4: In the offline calibration process prior to the deployment of the multi-source collaborative regional water environment monitoring system, the multiphase flow interference environment leads to a lack of prior data support for the dynamic coupling model. The control unit drives the multiphase flow circulating water tank to start the baseline calibration program, injecting silica particles with a particle size ranging from 10μm to 50μm and microbubble groups with a gradient volume fraction into the tank to construct a multiphase flow sample flow field with turbidity ranging from 0 NTU to 1000 NTU. The acoustic sensing unit emits ultrasonic pulses in this sample flow field and captures the high-frequency backscattered echo envelope signal. The echo feature extraction unit inside the central processing module calculates the time domain dimension of the high-frequency backscattered echo envelope signal at various concentration nodes within the temperature range of 10℃ to 30℃. The energy decay rate is used to generate a discretized mapping matrix with suspended matter distribution density and bubble content as independent variables and energy decay rate in the time domain as dependent variable. Considering that natural water bodies in tidal estuaries are often mixed with irregularly condensed organic flocs and soft biological residues, the system adds a quantitative amount of mixed bottom sediment and algal suspended matter background solution obtained natively from the target service water area to the water tank at the end of the offline calibration procedure. The central processing module calculates the additional energy dissipation deviation of ideal silica spheres and native heterogeneous mixed suspended matter under the same photoelectric and acoustic field excitation conditions, derives the correction hash compensation vector for the heterogeneous material morphology and absorption characteristics specific to the target estuary, and adds it to the discretized mapping matrix in the form of a compensation array structure.

[0040] The central processing module writes the discretized mapping matrix into the dynamic parameter mapping unit as a preset fluid scattering feature library, establishing a numerical benchmark between the physical concentration of the multiphase medium and the contribution weights of bubbles and solid suspended particles to sound energy dissipation. In the in-situ water environment monitoring process, the dynamic parameter mapping unit receives the acquired energy attenuation rate in the time domain, retrieves the preset fluid scattering feature library, applies a surface interpolation algorithm, analyzes the acoustic wave characteristic parameters based on the solidified entity data benchmark, and outputs the distribution density of suspended matter and bubble content in the water body of the detection area under multiphase flow interference conditions.

[0041] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A multi-source collaborative regional water environment monitoring system based on artificial intelligence, characterized in that, include: A multimodal detection module is used to acquire water body detection parameter data; The multimodal detection module includes an acoustic sensing unit and an optical sensing unit; The central processing module, connected to the multimodal detection module, is used to perform source compensation based on water dynamic parameters. Internally, the central processing module includes: an echo feature extraction unit, used to acquire the high-frequency backscattered echo envelope signal received by the acoustic sensing unit and calculate the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain; a dynamic parameter mapping unit, used to input the energy attenuation rate into a preset dynamic coupling model, compare the energy attenuation rate with a preset fluid scattering feature library to remove the contribution weights of bubbles and solid suspended particles to acoustic energy dissipation, calculate the suspended matter distribution density and bubble content in the water body of the detection area, and establish an optical transmittance compensation matrix; and a measurement benchmark calibration unit, used to calibrate the fluorescence response electrical signal acquired by the optical sensing unit using the optical transmittance compensation matrix, correct the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles, and reconstruct the true values ​​of detection parameters conforming to the linear boundary of the Lambert-Beer law.

2. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The central processing module is used to adjust the detection frequency through the drive waveform adaptive unit. The internal logic of the drive waveform adaptive unit includes: acquiring real-time flow velocity data of the detection area and comparing the real-time flow velocity data with a preset flow velocity threshold; when the real-time flow velocity data exceeds the preset flow velocity threshold, switching the driving mode of the multi-modal detection module to a combination of wideband continuous scanning mode and moving average feature extraction mode, using the spectral coverage characteristics of the wideband carrier to offset the phase deviation caused by the transient eddy current of the fluid, and making the sampling frequency increase linearly with the increase of the flow velocity gradient.

3. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The central processing module also includes a window contamination monitoring module, which monitors the attenuation characteristics of the reflected echo from the surface of the detection window received by the acoustic sensing unit, so as to calculate the amount of physical contamination adhering to the surface of the detection window of the optical sensing unit and correct the optical transmittance compensation matrix accordingly.

4. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The central processing module is used to determine the microbubble density in the detection area based on the amplitude fluctuation frequency of the high-frequency backscattered echo envelope signal, and outputs a compensation control signal for the intensity of the excitation light source of the optical sensing unit.

5. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The system includes a first monitoring subsystem and a second monitoring subsystem distributed in a distributed manner. The second monitoring subsystem is located downstream of the first monitoring subsystem. The first monitoring subsystem is used to calculate the gradient change rate of its output characteristic signal and send an early warning control signal to the second monitoring subsystem to trigger the second monitoring subsystem to shorten the transmission interval of its internal acoustic sensing unit.

6. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 5, characterized in that, The second monitoring subsystem is used to increase the sampling point density of the physical detection domain after receiving the early warning control signal, and to allocate the computational weight of its internal detection tasks according to the gradient change rate of the characteristic signal.

7. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The multimodal detection module also includes an electrochemical sensing unit; the central processing module is used to perform correlation calculations between the calibrated fluorescence response electrical signal and the conductivity data collected by the electrochemical sensing unit, and to correct the cross-interference of dissolved ions in the water on the optical detection results.

8. The multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The central processing module is used to determine the particle size distribution of suspended particles in the water by calculating the second derivative of the energy attenuation rate, and adjust the receiving bandwidth of the optical sensing unit according to the particle size distribution.

9. A multi-source collaborative regional water environment monitoring system based on artificial intelligence according to claim 1, characterized in that, The multimodal detection module also includes a self-circulating cleaning module; the central processing module is used to send a high-pressure jet control signal to the self-circulating cleaning module when the calculated energy attenuation rate exceeds a preset blocking threshold.

10. A method for multi-source collaborative regional water environment monitoring based on artificial intelligence, used to implement the multi-source collaborative regional water environment monitoring system based on artificial intelligence as described in claim 1, characterized in that, Includes the following steps: Step S1101: Obtain multimodal water body detection parameter data; Step S1102: The high-frequency backscattered echo envelope signal received by the acoustic sensing unit is obtained through the echo feature extraction unit, and the energy attenuation rate of the high-frequency backscattered echo envelope signal in the time domain dimension is calculated. Step S1103: The energy attenuation rate is input into the preset dynamic coupling model through the dynamic parameter mapping unit to determine the distribution density of suspended matter and bubble content in the water body of the detection area, and to establish an optical transmittance compensation matrix. Step S1104: The fluorescence response electrical signal collected by the optical sensing unit is calibrated by the optical transmittance compensation matrix through the measurement reference calibration unit, the optical measurement baseline offset caused by multiple scattering of multiphase suspended particles is corrected, and the true value of the detection parameter conforming to the linear boundary of the Lambert-Beer law is reconstructed.

Images