Water quality detection method and system

Abstract

The invention discloses a water quality detection method and system, and relates to the technical field of environment detection. The method comprises the following steps: emitting laser to a target water sample by using an ultraviolet-visible multiband excitation light source, and synchronously collecting a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum comprises a matrix spectrum represented by three-dimensional coordinates of excitation wavelength, emission wavelength and fluorescence intensity; constructing a water quality detection model based on a double-task parallel architecture and a scoring function; inputting the three-dimensional fluorescence spectrum into the water quality detection model, and determining a final detection result; the detection result comprises the current water quality grade and the corresponding pollutant type and concentration. The water quality detection efficiency and accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental detection, and particularly to a water quality detection method and system. Background Art

[0002] Water quality detection is an important part of environmental monitoring and public health fields. Traditional detection methods mainly rely on physical and chemical analysis methods and single-parameter sensors. Although physical and chemical analysis methods (such as chemical titration, chromatography, mass spectrometry, etc.) have high accuracy, they have problems such as complex operation, long time consumption, and high cost, and it is difficult to meet the requirements of real-time and rapid water quality monitoring.

[0003] In recent years, optical detection technology has gradually become a research hotspot due to its advantages such as rapidity, non-destructiveness, and high sensitivity. Among them, fluorescence spectroscopy can effectively reflect the water quality situation by analyzing the fluorescence signals generated after organic substances or specific pollutants in water are excited. However, conventional fluorescence spectroscopy techniques are mostly limited to excitation at specific wavelengths (usually relying on a single excitation or emission band), and it is difficult to comprehensively capture the characteristic information of multiple pollutants in complex water bodies, resulting in limited detection accuracy and application scope. Summary of the Invention

[0004] The purpose of the present invention is to provide a water quality detection method and system that can improve the efficiency and accuracy of water quality detection.

[0005] To achieve the above purpose, the present invention provides the following solutions:

[0006] A water quality detection method includes:

[0007] Emitting laser light to a target water sample using an ultraviolet-visible multi-band excitation light source and synchronously collecting a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of excitation wavelength, emission wavelength, and fluorescence intensity;

[0008] Constructing a water quality detection model based on a dual-task parallel architecture and a scoring function;

[0009] Inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine the final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations.

[0010] Optionally, the step of emitting laser light to a target water sample using an ultraviolet-visible multi-band excitation light source and synchronously collecting a three-dimensional fluorescence spectrum specifically includes:

[0011] An ultraviolet-visible multi-band excitation light source is used to emit lasers of different bands to a target water sample, and the fluorescence intensity is projected in the form of contour lines on a plane with the excitation light wavelength and the emission light wavelength as the horizontal and vertical coordinates to construct a three-dimensional fluorescence spectrum; wherein, in the matrix expression of the three-dimensional fluorescence spectrum, the y-axis is the excitation wavelength, the x-axis is the emission wavelength, and the z-axis is the fluorescence intensity.

[0012] Optionally, the construction process of the water quality detection model specifically includes:

[0013] Construct a historical fluorescence fingerprint database;

[0014] Construct a pre-trained network based on a dual-task parallel architecture and a scoring function; the dual-task parallel architecture includes a multi-scale feature fusion module and an attention mechanism; the multi-scale feature fusion module is used to embed dilated convolutions in an optimized shared layer to capture the features of pollutants with different particle sizes; the attention mechanism is to add a CBAM module after the last layer of convolution in the network;

[0015] Use the historical fluorescence fingerprint database to optimize the parameters of the pre-trained network, and determine the network that meets the preset evaluation accuracy as the water quality detection model; the process of parameter optimization includes training with the goal of minimizing the loss between the network output and the corresponding label.

[0016] Optionally, two loss functions are adopted in the parameter optimization process, namely: the Focal Loss function is used for the classification task, and the Huber Loss function is used for the regression task.

[0017] Optionally, the features of pollutants with different particle sizes at least include the features of algal clusters and dissolved organic matter.

[0018] Optionally, the ultraviolet-visible multi-band excitation light source includes at least 5 characteristic wavelengths, namely: 280nm, 340nm, 405nm, 450nm and 500nm.

[0019] Optionally, before inputting the three-dimensional fluorescence spectrum into the water quality detection model, it further includes: performing wavelet denoising and baseline correction on the data of the three-dimensional fluorescence spectrum.

[0020] The present invention also provides a water quality detection system, including:

[0021] A spectrum acquisition unit, which is used to emit lasers to a target water sample by using an ultraviolet-visible multi-band excitation light source and synchronously acquire a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of an excitation wavelength, an emission wavelength and a fluorescence intensity;

[0022] A model construction unit for constructing a water quality detection model based on a dual-task parallel architecture and a scoring function;

[0023] A prediction unit for inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine the final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations.

[0024] According to the specific embodiments provided by the present invention, the following technical effects are disclosed by the present invention:

[0025] The present invention discloses a water quality detection method and system. The method includes using an ultraviolet-visible multi-band excitation light source to emit laser light to a target water sample and synchronously collecting a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of excitation wavelength, emission wavelength, and fluorescence intensity; constructing a water quality detection model based on a dual-task parallel architecture and a scoring function; inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine the final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations. The present invention can improve the efficiency and accuracy of water quality detection. Description of the Drawings

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0027] Figure 1 It is a schematic flow chart of the water quality detection method of the present invention. Detailed Embodiments

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0029] The object of the present invention is to provide a water quality detection method and system that can improve the efficiency and accuracy of water quality detection.

[0030] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0031] As Figure 1 shown, the present invention provides a water quality detection method, including:

[0032] Step 100: Emitting a laser to a target water sample by using an ultraviolet-visible multi-band excitation light source, and synchronously collecting a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of excitation wavelength, emission wavelength, and fluorescence intensity.

[0033] Step 200: Constructing a water quality detection model based on a dual-task parallel architecture and a scoring function.

[0034] Step 300: Inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine a final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations.

[0035] As a specific implementation manner, the emitting a laser to a target water sample by using an ultraviolet-visible multi-band excitation light source, and synchronously collecting a three-dimensional fluorescence spectrum specifically includes:

[0036] Emitting lasers with different bands to a target water sample by using an ultraviolet-visible multi-band excitation light source, and projecting the fluorescence intensity on a plane with the excitation light wavelength and the emission light wavelength as the vertical and horizontal coordinates in a contour line manner to construct a three-dimensional fluorescence spectrum; wherein, in the matrix expression of the three-dimensional fluorescence spectrum, the y-axis is the excitation wavelength, the x-axis is the emission wavelength, and the z-axis is the fluorescence intensity.

[0037] As a specific implementation manner, the construction process of the water quality detection model specifically includes:

[0038] Constructing a historical fluorescence fingerprint database;

[0039] Constructing a pre-trained network based on a dual-task parallel architecture and a scoring function; the dual-task parallel architecture includes a multi-scale feature fusion module and an attention mechanism; the multi-scale feature fusion module is used to embed dilated convolutions in an optimized shared layer to capture the features of pollutants with different particle sizes; the attention mechanism is to add a CBAM module after the last layer of convolution in the network. Among them, the features of the pollutants with different particle sizes at least include the features of algal clusters and dissolved organic matter.

[0040] Optimizing the parameters of the pre-trained network by using the historical fluorescence fingerprint database, and determining the network that meets the preset evaluation accuracy as the water quality detection model; the process of parameter optimization includes training with the goal of minimizing the loss between the network output and the corresponding labels. Two loss functions are adopted in the process of parameter optimization, which are respectively: the FocalLoss loss function for the classification task and the Huber Loss loss function for the regression task.

[0041] As a specific embodiment, the ultraviolet-visible multi-band excitation light source includes at least 5 characteristic wavelengths, namely: 280nm, 340nm, 405nm, 450nm, and 500nm.

[0042] As a specific embodiment, before inputting the three-dimensional fluorescence spectrum into the water quality detection model, it further includes: performing wavelet denoising and baseline correction on the data of the three-dimensional fluorescence spectrum.

[0043] Based on the above technical solutions, the following embodiments are provided.

[0044] Step 100: Data acquisition and three-dimensional fluorescence spectrum construction.

[0045] Use an ultraviolet-visible multi-band excitation light source (including characteristic wavelengths such as 280nm, 340nm, 405nm, 450nm, 500nm, etc.) to emit laser light to the target water sample to excite the fluorescent substances in the water body. Synchronously collect the fluorescence intensities under different excitation-emission wavelength combinations to construct a three-dimensional fluorescence spectrum. The specific steps are as follows:

[0046] Excitation and emission scanning: Sequentially switch different excitation wavelengths (y-axis), and scan the emission wavelengths (x-axis) at each excitation wavelength, and record the corresponding fluorescence intensities (z-axis).

[0047] Spectral matrix expression: Store the fluorescence intensity data in matrix form, where: y-axis: excitation wavelength (such as 280nm to 500nm), x-axis: emission wavelength (such as 300nm to 600nm), z-axis: fluorescence intensity (represented by contour lines or three-dimensional surfaces).

[0048] Data preprocessing: Perform wavelet denoising (removing high-frequency noise) and baseline correction (eliminating background interference) on the original spectral data to improve the signal-to-noise ratio.

[0049] Step 200: Water quality detection model construction.

[0050] Construct a deep learning model based on a dual-task parallel architecture (classification + regression), combined with multi-scale feature fusion and attention mechanism to optimize the water quality detection performance. The specific process is as follows:

[0051] (1) Construct a historical fluorescence fingerprint database:

[0052] Collect three-dimensional fluorescence spectrum data of different water qualities (such as Class I to Class V water), and label the corresponding water quality grades, pollutant types (such as algae, dissolved organic matter, heavy metals, etc.) and concentrations.

[0053] Data augmentation: Use methods such as rotation, translation, and adding noise to expand the dataset and improve the generalization ability of the model.

[0054] (2) Design a dual-task parallel network architecture:

[0055] Shared feature extraction layer: Use CNN (Convolutional Neural Network) to extract spectral features, and embed Dilated Convolution to capture multi-scale features of pollutants with different particle sizes (such as algal clusters, dissolved organic matter).

[0056] Dual-task branches: Classification task (water quality level, pollutant type): Use Focal Loss (to solve the problem of class imbalance). Regression task (pollutant concentration prediction): Use Huber Loss (to reduce the impact of outliers).

[0057] Attention mechanism: Introduce CBAM (Convolutional Block Attention Module) after the last layer of convolution to enhance the feature weights in the key wavelength region.

[0058] (3) Model training and optimization:

[0059] Use the Adam optimizer and train with the goal of minimizing the combined loss of the classification and regression tasks. The early stopping strategy prevents overfitting, and the model with the highest accuracy on the validation set is selected as the final water quality detection model.

[0060] Step 300: Water quality detection and result output.

[0061] Input the preprocessed three-dimensional fluorescence spectrum into the trained model, and output the detection results: water quality level (such as Class II water, Class IV water, etc.), pollutant type (such as chlorophyll a, humic acid, benzene series, etc.), pollutant concentration (such as COD = 15.2 mg / L, Chl-a = 8.5 μg / L, etc.).

[0062] In summary, this embodiment has the following beneficial effects:

[0063] The present invention uses multi-band excitation, covering the ultraviolet-visible light range, which can improve the detection sensitivity to different pollutants. The present invention uses a dual-task parallel model to simultaneously achieve water quality classification and pollutant quantitative analysis, which can improve the detection efficiency. The present invention uses multi-scale feature fusion and attention mechanism, which can enhance the recognition ability for complex water bodies (such as those containing algae and dissolved organic matter). The model adopted by the present invention has strong robustness and can reduce noise interference through wavelet denoising and Huber Loss, improving the model stability. Therefore, this method is applicable to scenarios such as rivers, lakes, and industrial wastewater, and can achieve fast and high-precision water quality monitoring.

[0064] In addition, the present invention also provides a water quality detection system, including:

[0065] A spectral acquisition unit, configured to emit a laser to a target water sample by using an ultraviolet-visible multi-band excitation light source and synchronously acquire a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of an excitation wavelength, an emission wavelength, and a fluorescence intensity.

[0066] A model construction unit, configured to construct a water quality detection model based on a dual-task parallel architecture and a scoring function.

[0067] A prediction unit, configured to input the three-dimensional fluorescence spectrum into the water quality detection model to determine a final detection result; the detection result includes a current water quality level and corresponding pollutant types and concentrations.

[0068] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same and similar parts among the various embodiments, reference may be made to each other.

[0069] Specific examples are used in this article to elaborate on the principles and implementation manners of the present invention. The descriptions of the above embodiments are only used to help understand the core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A water quality detection method, characterized in that, Including: Emitting laser light from an ultraviolet-visible multi-band excitation light source to a target water sample, and synchronously collecting a three-dimensional fluorescence spectrum; the three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of excitation wavelength, emission wavelength, and fluorescence intensity; Constructing a water quality detection model based on a dual-task parallel architecture and a scoring function; Inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine the final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations.

2. The water quality detection method according to claim 1, characterized in that, The step of emitting laser light from an ultraviolet-visible multi-band excitation light source to a target water sample and synchronously collecting a three-dimensional fluorescence spectrum specifically includes: Emitting laser light of different bands from an ultraviolet-visible multi-band excitation light source to a target water sample, and projecting the fluorescence intensity in the form of contour lines on a plane with the excitation light wavelength and the emission light wavelength as the horizontal and vertical coordinates to construct a three-dimensional fluorescence spectrum; wherein, in the matrix expression of the three-dimensional fluorescence spectrum, the y-axis is the excitation wavelength, the x-axis is the emission wavelength, and the z-axis is the fluorescence intensity.

3. The water quality detection method according to claim 1, characterized in that, The construction process of the water quality detection model specifically includes: Constructing a historical fluorescence fingerprint database; Constructing a pre-trained network based on a dual-task parallel architecture and a scoring function; the dual-task parallel architecture includes a multi-scale feature fusion module and an attention mechanism; the multi-scale feature fusion module is used to embed dilated convolutions in an optimized shared layer to capture the features of pollutants with different particle sizes; the attention mechanism is to add a CBAM module after the last convolutional layer of the network; Optimizing the parameters of the pre-trained network using the historical fluorescence fingerprint database, and determining the network that meets the preset evaluation accuracy as the water quality detection model; the process of parameter optimization includes training with the goal of minimizing the loss between the network output and the corresponding labels.

4. The water quality detection method according to claim 3, characterized in that Two loss functions are adopted in the parameter optimization process, namely: the FocalLoss loss function is used for the classification task, and the Huber Loss loss function is used for the regression task.

5. The water quality detection method according to claim 3, characterized in that, The features of pollutants with different particle sizes at least include the features of algal clusters and dissolved organic matter.

6. The water quality detection method according to claim 1, characterized in that, The ultraviolet-visible multi-band excitation light source includes at least 5 characteristic wavelengths, namely: 280nm, 340nm, 405nm, 450nm, and 500nm.

7. The water quality detection method according to claim 1, characterized in that, Before inputting the three-dimensional fluorescence spectrum into the water quality detection model, it also includes: performing wavelet denoising and baseline correction on the data of the three-dimensional fluorescence spectrum.

8. A water quality detection system, characterized in that, Including: A spectrum acquisition unit for emitting laser light from an ultraviolet-visible multi-band excitation light source to a target water sample and synchronously collecting a three-dimensional fluorescence spectrum; The three-dimensional fluorescence spectrum includes a matrix spectrum characterized by three-dimensional coordinates of excitation wavelength, emission wavelength, and fluorescence intensity; A model construction unit for constructing a water quality detection model based on a dual-task parallel architecture and a scoring function; A prediction unit for inputting the three-dimensional fluorescence spectrum into the water quality detection model to determine the final detection result; the detection result includes the current water quality level and the corresponding pollutant types and concentrations.

Images