A device for real-time monitoring of water quality using hyperspectral and data fitting method

Abstract

The present invention discloses a device and data fitting method for real-time monitoring of water quality using a hyperspectral spectrum, which relates to the technical field of carbon emission detection, and solves the problem that the hyperspectral spectrum cannot monitor some indicators of turbidity and special water quality, cannot quantify carbon emissions, and the data accuracy of existing water quality monitoring devices is low or inconvenient to carry and the monitoring is delayed. The monitoring device of the present invention is provided with a hyperspectral water quality analyzer and a halogen lamp in the main body, and a buoyancy adjustment device is provided outside the main body of the monitoring device; a heat dissipation system is provided in the main body of the monitoring device; the main body of the monitoring device includes a light shield, a water inlet hole is provided on the light shield, a counterweight structure is provided at the lower end of the light shield, a reflector is provided at the bottom end of the light shield, a water level balance device is provided on the reflector, and the hydraulic balance device is connected to the two water outlets provided on the light shield. After filtering the water body, the hyperspectral water quality analyzer monitors the water quality of the water body entering the device; through machine deep learning, a model is constructed to accurately monitor the water quality.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission detection, and in particular to a device for real-time monitoring of water quality using a hyperspectral spectrum and a data fitting method. Background Art

[0002] The water quality monitoring of sewage treatment plants mainly monitors external factors such as pH value, dissolved oxygen, ammonia nitrogen, total nitrogen, total phosphorus, COD, nitrite, nitrate, etc. in the water quality. Traditional water quality monitoring devices are mainly divided into two categories: one is simple handheld monitoring instruments, monitoring pens, etc., which are inexpensive, easy to use, and can monitor water quality at any time; but the disadvantage is that there are fewer monitoring items and the data accuracy is not high. The second is large-scale water quality monitoring equipment, most of which are fixed in the laboratory and inconvenient to carry. Water samples need to be taken and sent to the laboratory for testing. Although more detailed and accurate water quality data can be obtained, it is not flexible enough to use. Because sampling and monitoring are intermittent, it is difficult to capture instantaneous changes in water quality parameters. This monitoring method has hysteresis and cannot take timely measures to deal with emergencies. Summary of the invention

[0003] In order to solve the above-mentioned problems of low data accuracy, inconvenience in carrying, and delayed monitoring of existing water quality monitoring devices, the present invention hereby proposes a device for real-time monitoring of water quality using hyperspectral and a data fitting method. The present invention floats the water quality monitoring device on the water surface, filters the water body through the filter material, and then monitors the water quality of the water entering the device through a hyperspectral water quality analyzer; the data is fitted through the random forest importance ranking and Pearson correlation coefficient method, the feature importance is comprehensively evaluated, the accuracy and stability of feature selection are improved, the data dimension is reduced, the calculation efficiency is improved, and the interpretability of the model is enhanced.

[0004] The present invention proposes a device for real-time monitoring of water quality by using a hyperspectral spectrum, which specifically comprises a hyperspectral water quality analyzer, a plurality of halogen lamps, a heat dissipation system, a buoyancy adjustment device and a monitoring device body, wherein the hyperspectral water quality analyzer and a plurality of halogen lamps are arranged at the top of the internal top of the monitoring device body, the plurality of halogen lamps are arranged around the hyperspectral water quality analyzer, and the buoyancy adjustment device is arranged outside the monitoring device body; the heat dissipation system is arranged inside the monitoring device body; the monitoring device body comprises a light shield, a plurality of water inlet holes are arranged on the light shield, a counterweight structure is arranged at the lower end of the light shield, a reflecting plate is arranged at the bottom end of the light shield, a hydraulic balance device is arranged on the reflecting plate, and the hydraulic balance device is connected to two water outlet holes arranged on the light shield; the reaction force of the hydraulic balance device to discharge water outwards pushes the monitoring device body to move on the water surface, and the vertical movement of the monitoring device body is realized by controlling the height of the liquid inside the monitoring device body.

[0005] Furthermore, the two water outlet holes on the light shield are symmetrically arranged.

[0006] Furthermore, the sunshade outer shell is provided with a shell, and a flange and a filter material are provided between the shell and the sunshade; the flange is provided at the lower end of the shell, and supports the filter material; the filter material blocks a number of water inlet holes, and the external water body flows into the interior of the sunshade through a number of water inlet holes after being filtered by the filter material; a buoyancy adjustment device is provided outside the shell.

[0007] Furthermore, the device for real-time monitoring of water quality using a hyperspectral spectrum also includes a lighting system, which includes a lighting adjustment element. The lighting adjustment element is arranged on the light shield and connected to a plurality of halogen lamps to adjust the light intensity of the halogen lamps.

[0008] Furthermore, the counterweight structure includes a counterweight base plate and a counterweight, the counterweight base plate is a conical structure, and the counterweight is arranged inside the counterweight base plate.

[0009] Furthermore, the generatrix inclination angle of the conical structure of the counterweight base plate is 30 degrees.

[0010] Furthermore, the heat dissipation system includes a circulating water cooling system, and the circulating water cooling system is spirally arranged on the inner wall of the light shield.

[0011] Furthermore, the heat dissipation system also includes an exhaust device, the light shield is provided with an exhaust port, the exhaust device is arranged on the exhaust port, and the exhaust device cools the circulating water cooling system.

[0012] Furthermore, a hook is provided on the top of the sunshade, and the device for real-time monitoring of water quality using the hyperspectrum is hung on an external mobile auxiliary device through the hook. The device for real-time monitoring of water quality using the hyperspectrum is driven to move by the mobile auxiliary device, and the moving direction is perpendicular to the direction of the line connecting the two water outlets.

[0013] A data fitting method for the above-mentioned device for real-time monitoring of water quality using a hyperspectral spectrum automatically selects a sensitive wavelength, automatically matches through machine learning, and ensures model accuracy; comprising the following steps:

[0014] Step 1: Obtain hyperspectral data and corresponding label data through a hyperspectral water quality analyzer;

[0015] Step 2: Use hyperspectral data to train a random forest model to obtain the characteristic importance score of the wavelength of each water quality monitoring parameter, and select the top-ranked wavelengths for each water quality monitoring parameter based on the score ranking;

[0016] Step 3: Calculate the Pearson correlation coefficient between each water quality monitoring parameter wavelength and the target variable; sort by the absolute value of the correlation coefficient and select the top-ranked wavelengths for each water quality monitoring parameter;

[0017] Step 4: Find the wavelengths that appear in both the random forest importance ranking and the Pearson correlation coefficient method as the final feature wavelength set;

[0018] Step 5: Use the final selected characteristic wavelength set to train and verify the model.

[0019] The beneficial effects of the device for real-time monitoring of water quality using hyperspectral and the data fitting method described in the present invention are:

[0020] (1) The device and data fitting method for real-time monitoring of water quality using a hyperspectral sensor described in the present invention overcome the problems of low data accuracy, inconvenience in carrying, and delayed monitoring of existing water quality monitoring devices. The buoyancy adjustment device is provided to enable the monitoring device to float on the water surface. The water quality of the water body after impurities are filtered out by the filter material is monitored by a hyperspectral water quality analyzer. The system integrates advanced sensing technology, data processing and analysis, and automatic control to monitor water quality in real time, ensure the accuracy and comprehensiveness of data, and eliminate the lag of traditional sampling. The data analysis unit uses a big data analysis algorithm to process and analyze sensor data in real time, which can predict the trend of water quality changes and detect abnormal conditions in a timely manner. The entire system can more effectively manage environmental risks, respond quickly to abnormal events, and reduce potential environmental pollution risks through real-time monitoring and intelligent analysis.

[0021] (2) The device and data fitting method for real-time monitoring of water quality using a hyperspectral spectrum described in the present invention, wherein the water in the monitoring device is discharged through a hydraulic balance device and a water outlet, and the reaction force of the drainage is used as a driving force to push the monitoring device to move on the water surface, and the water volume in the monitoring device is controlled by the hydraulic balance device to control the depth of the monitoring device submerged under the water surface;

[0022] (3) The device and data fitting method for real-time monitoring of water quality using a hyperspectral spectrum described in the present invention ensures that the illumination is 100-300 lux by arranging multiple halogen lamps in a ring shape, ensuring uniform illumination and avoiding the light spot effect; the inner facade of the sunshade is sprayed with a non-reflective black material, and a reflective plate is provided at the bottom. The reflective plate is made of a high-reflective material with a reflectivity of 80% to improve the light efficiency;

[0023] (4) The present invention discloses a device for real-time monitoring of water quality using hyperspectral data and a data fitting method. The data fitting method comprehensively considers nonlinear and linear relationships, captures complex nonlinear relationships through random forests, and measures linear relationships through the Pearson correlation coefficient. The combination of the two can more comprehensively evaluate the importance of features. The wavelengths of water quality monitoring parameters screened out by the two methods are important in the random forest model and have a strong linear correlation with the target variable, thereby improving the accuracy of feature selection. By finely screening features, the dimensions of hyperspectral data are reduced, thereby reducing the complexity and computational cost of the model and improving analysis efficiency. Combining the characteristic wavelengths selected by the two methods makes it easier to explain the importance and role of these wavelengths in practical applications, thereby enhancing the interpretability of the model. By combining the random forest importance ranking and the Pearson correlation coefficient method, wavelengths with high contributions in both methods can be selected, thereby improving the accuracy and stability of feature selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0025] In the attached picture:

[0026] Figure 1 It is a cross-sectional view of a device for real-time monitoring of water quality using a hyperspectral spectrum according to the present invention;

[0027] Figure 2 It is a structural schematic diagram of a flange of a device for real-time monitoring of water quality using a hyperspectral spectrum according to the present invention;

[0028] Figure 3 It is a partial enlarged view of the lower end of the light shield of the device for real-time monitoring of water quality using a hyperspectral spectrum according to the present invention;

[0029] Figure 4 It is a flow chart of a device for real-time monitoring of water quality using a hyperspectral spectrum according to the present invention for predicting water quality using the collected data;

[0030] Among them: 1-hook, 2-hyperspectral water quality analyzer, 3-halogen lamp, 4-circulating water cooling system, 5-exhaust device, 6-light shield, 7-filter material, 8-water inlet, 9-buoyancy adjustment device, 10-flange, 11-hydraulic balance device, 12-counterweight bottom plate, 13-counterweight, 14-reflector, 15-water outlet, 16-shell. DETAILED DESCRIPTION

[0031] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0033] In the description of the invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the invention can be understood according to specific circumstances.

[0034] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0035] Specific implementation method 1: See Figure 1-Figure 4The present embodiment is described in detail. The device for real-time monitoring of water quality using a hyperspectral spectrum described in the present embodiment specifically comprises a hyperspectral water quality analyzer 2, a plurality of halogen lamps 3, a heat dissipation system, a buoyancy adjustment device 9 and a monitoring device body. The top of the internal top of the monitoring device body is provided with a hyperspectral water quality analyzer 2 and a plurality of halogen lamps 3. The plurality of halogen lamps 3 are arranged around the hyperspectral water quality analyzer 2. The light emitted by the plurality of halogen lamps 3 is irradiated into the water in the monitoring device. The hyperspectral water quality analyzer 2 analyzes various water quality monitoring parameters in the water body according to the reflected light to obtain the wavelength corresponding to each parameter. The buoyancy adjustment device 9 is arranged outside the monitoring device body. The entire device floats on the water through the buoyancy adjustment device 9. On the surface; a heat dissipation system is arranged inside the main body of the monitoring device to dissipate heat inside the device; the main body of the monitoring device comprises a light shield 6, on which a plurality of water inlet holes 8 are arranged, which are evenly distributed around the light shield 6 and have a diameter of 2-3 cm. The water inlet holes 8 are below the liquid surface, and sewage enters the interior of the device through the water inlet holes 8; a counterweight structure is arranged at the lower end of the light shield 6 to ensure that the entire device floats in the water in a vertical form; a reflective plate 14 is arranged at the bottom end of the light shield 6, and the reflective plate 14 reflects the light emitted by the halogen lamp 3; the reflective plate 14 is circular, and has a reflectivity of 80%, which can better reflect the light, improve the lighting efficiency, and ensure the accuracy of water quality monitoring. A hydraulic balancing device 11 is provided on the reflective plate 14, and the hydraulic balancing device 11 is connected to two water outlets 15 provided on the sunshade 6. The sewage is discharged from the device through the hydraulic balancing device 11. At the same time, the reaction force of the sewage discharged by the hydraulic balancing device 11 can push the monitoring device as a whole to move on the water surface. By switching different water outlets 15, the moving direction can be changed.

[0036] The light shield 6 is made of lightweight material with a density of 1.5-2 g / cm 3 , tensile strength is about 60MPa, bending strength is between 70-80MPa. It provides the framework of the entire water quality stabilization system, ensures the stability of the structure, and serves as a water sample pretreatment area. The cavity diameter of the device ranges from 30 cm to 1.5 meters, and the height ranges from 50 cm to 2 meters. It can be selected according to the actual application scenario and flexibly adapted to different environments.

[0037] The two water outlet holes 15 on the light shield 6 are symmetrically arranged.

[0038] The sunshade 6 is provided with a shell 16 on the outer shell, and a flange 10 and a filter material 7 are provided between the shell 16 and the sunshade 6; the flange 10 is provided at the lower end of the shell 16, and supports the filter material 7; the filter material 7 blocks a plurality of water inlet holes 8, and the external water body flows into the sunshade 6 through a plurality of water inlet holes 8 after filtering large pieces of suspended matter through the filter material 7; a buoyancy regulating device 9 is provided on the outside of the shell 16, and the buoyancy regulating device 9 controls the stored buoyancy through air pressure regulation. The position of the buoyancy regulating device is not affected by wind and waves, and buoyancy is always provided for the device to ensure that the center of gravity does not shift, thereby ensuring the balance of the device and ensuring the stability of the device in various environments.

[0039] The device for real-time monitoring of water quality using a hyperspectral spectrum also includes a lighting system, which includes a lighting adjustment element. The lighting adjustment element is arranged on a light shield 6 and connected to a plurality of halogen lamps 3 to adjust the light intensity of the halogen lamps 3. Halogen lamps 3 with different light intensities can be used to monitor the water quality of different water bodies.

[0040] The flange 10 is provided with a plurality of through holes. The flange 10 is made of light material, with an inner diameter of 30cm-1.5m, an outer diameter of 50cm-1.7m, and circular holes with a diameter of 2-3cm distributed in a ring shape. The filter material 7 is fixed to prevent the filter material 7 from shifting, and the impurities in the fluid are initially intercepted to ensure the filtering effect.

[0041] The counterweight structure includes a counterweight bottom plate 12 and a counterweight 13. The counterweight bottom plate 12 is a conical structure, and the counterweight 13 is arranged inside the counterweight bottom plate 12. The hydraulic balance device 11 arranged on the reflector 14 cooperates with the counterweight 13 and the counterweight bottom plate 12 to move the center of gravity of the monitoring device as a whole downward, increase the stability of the device, and enable it to maintain a vertical state in the water, thereby ensuring the continuity and stability of the filtering process.

[0042] The generatrix inclination angle of the conical structure of the counterweight bottom plate 12 is 30 degrees.

[0043] The heat dissipation system includes a circulating water cooling system 4, which is spirally arranged on the inner wall of the light shield 6. Both ends of the circulating water cooling system 4 are connected to an external cooling water circulation device, and the heat inside the monitoring device is transferred out through cooling water. Through water circulation, the temperature inside and outside the device is ensured to be consistent. The temperature control sensor monitors the water temperature in real time, and the water circulation flow rate is adjusted through the intelligent control system to keep the system shell temperature consistent with the internal water temperature, prevent the temperature difference from generating fog, and ensure the lighting stability and long-term stable operation of the equipment.

[0044] The heat dissipation system further includes an exhaust device 5. The light shield 6 is provided with an exhaust port. The exhaust device 5 is arranged on the exhaust port. The exhaust device 5 cools the circulating water cooling system 4. The exhaust device 5 has a waterproof design, and prevents positive and negative winds from affecting the internal environment, ensuring stable operation of the device. The exhaust device 5 is an exhaust fan.

[0045] A hook 1 is provided on the top of the light shield 6, and the monitoring device is suspended into the pool corridor to be monitored by the hook 1, with the pool corridor as the X-axis. The hook 1 is hung on the mobile auxiliary device, and the monitoring device realizes the X-axis movement of the monitoring device through the mobile auxiliary device, with a moving speed of 0.2-0.8m / s; the recoil force of the hydraulic balance device 11 during drainage controls the detection device to move in the Y-axis direction; by controlling the water flow rate of the hydraulic balance device 11, the liquid level height in the light shield 6 is controlled, so as to realize the movement of the detection device in the Z-axis direction. At the same time, combined with GPS and inertial navigation systems, high-precision positioning is achieved. A sonar system is used to monitor the underwater position to ensure accurate positioning in the biochemical pool, thereby ensuring the accuracy of water quality monitoring.

[0046] The halogen lamp 3 has a lighting range of 100-300 lux and a voltage of 220V. It is located at the top of the sunshade 6 and is distributed in a ring shape to provide the required light source and appropriate wavelength for the hyperspectral water quality analyzer 2 to ensure the accuracy and effectiveness of monitoring. When testing the water quality, the sunshade 6 is used to isolate external natural light, and the inner facade of the sunshade 6 is sprayed with non-reflective black material to avoid interference with the hyperspectral water quality analyzer and ensure that the monitoring results are not affected by changes in external light.

[0047] Temperature, liquid level, light, and electrochemical sensors are installed inside the light shield 6, and liquid flow rate sensors are installed at the water inlet 8 and the water outlet 15 of the device to monitor various data in real time to ensure the accuracy of water quality monitoring. A waterproof data storage module is used to transmit the data to the processing module on the water outlet through wired and wireless methods to ensure the stability of data transmission. The data is then transmitted to the remote monitoring center through a wireless module to support real-time monitoring and historical data query to ensure data traceability.

[0048] A data fitting method for the above-mentioned device for real-time monitoring of water quality using a hyperspectral spectrum automatically selects a sensitive wavelength, automatically matches through machine learning, and ensures model accuracy; comprising the following steps:

[0049] Step 1: Obtain hyperspectral data and corresponding label data through the hyperspectral water quality analyzer 2;

[0050] Step 2: Random forest model training and feature sorting: Use hyperspectral data to train the random forest model, obtain the feature importance score of the wavelength of each water quality monitoring parameter, and sort according to the score, and select several wavelengths with the highest ranking for each water quality monitoring parameter;

[0051] Step 3: Calculate the Pearson correlation coefficient: Calculate the Pearson correlation coefficient between each water quality monitoring parameter wavelength and the target variable; sort by the absolute value of the correlation coefficient and select the top-ranked wavelengths for each water quality monitoring parameter;

[0052] Step 4: Select common wavelengths: find the wavelengths that appear in both the random forest importance ranking and the Pearson correlation coefficient method as the final feature wavelength set;

[0053] Step 5: Model verification and application: Use the final selected characteristic wavelength set to train and verify the model.

[0054] Random forests provide a nonlinear feature importance assessment method by combining the results of multiple decision trees, which can capture complex feature interactions. The Pearson correlation coefficient method provides a linear feature selection method that can directly measure the linear relationship between features and target variables. By combining the two methods, the nonlinear and linear contributions of features can be considered at the same time, and more representative wavelengths can be selected.

[0055] Process and analyze the collected data, and use random forest sorting, Pearson correlation coefficient and other methods to analyze sensitive characteristic bands to ensure the efficiency of data processing. Establish a random forest model to invert water quality parameters to ensure the accuracy and reliability of the model. Train the model with a large amount of sample data, verify and optimize the model accuracy, and ensure its applicability in various environments. Provide a model applicability analysis report to ensure the accuracy of the model in different environments, such as Figure 4 shown.

[0056] The specific working process of the device for real-time monitoring of water quality using a hyperspectral spectrum described in the present invention is as follows:

[0057] The monitoring device is put into the water in an inclined posture. When the water enters from the bottom and is half filled, the monitoring device is lifted up. Due to the combined effect of its own gravity and the water, the monitoring device sinks. During the sinking process, the water passes through the filter material 7 to the water inlet hole 8. Then it enters the internal cavity of the monitoring device through the water inlet hole 8, and then is guided into the biochemical pool through the water outlet hole 15 by the hydraulic balance device 11, ensuring the continuous flow of sewage and making the monitoring device operate continuously and stably.

[0058] The light emitted by the halogen lamp 3 passes through the filtered sewage and shines on the reflective plate 14 and the inner surface of the light shield 6. The inner wall of the light shield 6 is painted black, and the bottom is composed of the reflective plate 14, which can effectively reflect the light back to the water quality analyzer 2, thereby improving the efficiency of the hyperspectral water quality analyzer 2 in receiving the optical signal. The hyperspectral water quality analyzer 2 is one of the core components, responsible for receiving the optical signal emitted by the filtered sewage, and providing a stable light field and external environment by importing the algorithm, so that it can perform high-precision analysis and data processing. It provides reliable data support for water quality monitoring. Thereby enhancing the accuracy and reliability of water quality analysis.

[0059] This device provides a monitoring environment without external interference, a stable data source, and optimized models and algorithms for the hyperspectral water quality monitor 2, so that it can effectively collect hyperspectral data in various water bodies such as sewage plant biochemical pools, environmental water bodies, and marine water quality. Through the optimal wavelength feature extraction and the construction of a deep learning model, the accuracy can be guaranteed to reach more than 90%, thereby realizing real-time monitoring of water quality characteristics of water bodies. The data is transmitted to the remote monitoring center through a wireless module, supporting real-time monitoring and historical data query to ensure data traceability.

[0060] To summarize the above implementation cases, the device and data fitting method for real-time monitoring of water quality using hyperspectral data described in the present invention overcome the problems of low data accuracy, inconvenience in carrying, and delayed monitoring of existing water quality monitoring devices. The buoyancy adjustment device 9 is set to allow the monitoring device to float on the water surface, and the water quality of the water body after impurities are filtered out by the filter material 7 is monitored by the hyperspectral water quality analyzer 2. The system integrates advanced sensing technology, data processing and analysis, and automatic control to monitor water quality in real time, ensure the accuracy and comprehensiveness of data, and eliminate the lag of traditional sampling. The data analysis unit uses big data analysis algorithms to process and analyze sensor data in real time, which can predict water quality change trends and detect abnormal situations in a timely manner. The entire system can manage environmental risks more effectively through real-time monitoring and intelligent analysis. , quickly respond to abnormal events and reduce potential environmental pollution risks; the device and data fitting method for real-time monitoring of water quality using a hyperspectral spectrum described in the present invention are connected to the water outlet 15 through a hydraulic balance device 11 to discharge the water in the monitoring device, and use the reaction force of the drainage as a driving force to push the monitoring device to move on the water surface, and control the amount of water in the monitoring device through the hydraulic balance device 11 to control the depth of the monitoring device sinking below the water surface; the device and data fitting method for real-time monitoring of water quality using a hyperspectral spectrum described in the present invention, through a plurality of halogen lamps 3 arranged in a ring shape, to ensure that the illumination is 100-300 lux, to ensure uniform illumination and avoid spot effect; the inner facade of the sunshade 6 is sprayed with non-reflective black material, and a reflective plate 14 is provided at the bottom, and the reflective plate 14 adopts a high-reflective material with a reflectivity of 80% to improve the light efficiency. The present invention discloses a device for real-time monitoring of water quality using a hyperspectral spectrum and a data fitting method. The data fitting method comprehensively considers nonlinear and linear relationships, captures complex nonlinear relationships through random forests, and measures linear relationships through the Pearson correlation coefficient. The combination of the two can more comprehensively evaluate the importance of features. The wavelengths of water quality monitoring parameters screened out by the two methods are important in the random forest model and have a strong linear correlation with the target variable, thereby improving the accuracy of feature selection. By finely screening features, the dimensions of hyperspectral data are reduced, thereby reducing the complexity and computational cost of the model and improving analysis efficiency. Combining the characteristic wavelengths selected by the two methods makes it easier to explain the importance and role of these wavelengths in practical applications, thereby enhancing the interpretability of the model. By combining the random forest importance ranking and the Pearson correlation coefficient method, wavelengths with high contributions in both methods can be selected, thereby improving the accuracy and stability of feature selection.

[0061] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the invention. It can also be a reasonable combination of the features recorded in the above implementation methods. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A device for real-time monitoring of water quality using hyperspectral, characterized in that: The invention comprises a hyperspectral water quality analyzer (2), a plurality of halogen lamps (3), a heat dissipation system, a buoyancy adjustment device (9) and a monitoring device body, wherein the hyperspectral water quality analyzer (2) and the plurality of halogen lamps (3) are arranged at the top of the monitoring device body, the plurality of halogen lamps (3) are arranged around the hyperspectral water quality analyzer (2), and the buoyancy adjustment device (9) is arranged outside the monitoring device body; the heat dissipation system is arranged inside the monitoring device body; the monitoring device body comprises a light shield (6), the light shield (6) is provided with a plurality of water inlet holes (8), a counterweight structure is arranged at the lower end of the light shield (6), a reflector (14) is arranged at the bottom end of the light shield (6), a hydraulic balance device (11) is arranged on the reflector (14), and the hydraulic balance device (11) is connected to two water outlet holes (15) arranged on the light shield (6); the reaction force of the hydraulic balance device (11) to discharge water outwards pushes the monitoring device body to move on the water surface, and the height of the liquid inside the monitoring device body is controlled to realize the movement of the monitoring device body in the vertical direction; The outer shell of the light shield (6) is provided with a shell (16), and a flange (10) and a filter material (7) are provided between the shell (16) and the light shield (6); the flange (10) is provided at the lower end of the shell (16) and supports the filter material (7); the filter material (7) blocks the plurality of water inlet holes (8), and external water flows into the interior of the light shield (6) through the plurality of water inlet holes (8) after being filtered by the filter material (7); and a buoyancy adjustment device (9) is provided outside the shell (16).

2. The device for real-time monitoring of water quality using a hyperspectral spectrum according to claim 1, characterized in that: The two water outlet holes (15) on the light shield (6) are symmetrically arranged.

3. The device for real-time monitoring of water quality using a hyperspectral spectrum according to claim 1, characterized in that: The device for real-time monitoring of water quality using a hyperspectral spectrum also includes a lighting system, the lighting system including a lighting adjustment element, the lighting adjustment element being arranged on the light shield (6) and connected to a plurality of halogen lamps (3) to adjust the light intensity of the halogen lamps (3).

4. The device for real-time monitoring of water quality using a hyperspectral spectrum according to claim 1, characterized in that: The counterweight structure comprises a counterweight base plate (12) and a counterweight (13); the counterweight base plate (12) is a conical structure, and the counterweight (13) is arranged inside the counterweight base plate (12).

5. The device for real-time monitoring of water quality using a hyperspectral spectrum according to claim 4, characterized in that: The generatrix inclination angle of the conical structure of the counterweight bottom plate (12) is 30 degrees.

6. The device for real-time monitoring of water quality using hyperspectral spectrum according to claim 1, characterized in that: The heat dissipation system comprises a circulating water cooling system (4), and the circulating water cooling system (4) is spirally arranged on the inner wall of the light shield (6).

7. The device for real-time monitoring of water quality using hyperspectral spectrum according to claim 6, characterized in that: The heat dissipation system further comprises an exhaust device (5), the light shield (6) is provided with an exhaust port, the exhaust device (5) is arranged on the exhaust port, and the exhaust device (5) cools the circulating water cooling system (4).

8. The device for real-time monitoring of water quality using a hyperspectral spectrum according to claim 2, characterized in that: A hook (1) is provided on the top of the light shield (6), and the device for real-time monitoring of water quality using a hyperspectral spectrum is hung on an external mobile auxiliary device via the hook (1). The mobile auxiliary device drives the device for real-time monitoring of water quality using a hyperspectral spectrum to move, and the moving direction is perpendicular to the direction of the line connecting the two water outlet holes (15).

9. A data fitting method for the device for real-time monitoring of water quality using a hyperspectral spectrum according to any one of claims 1 to 8, characterized in that: Automatically select wavelengths and automatically match ratios through machine learning to ensure model accuracy; The following steps are involved: Step 1: Obtain hyperspectral data and corresponding label data through a hyperspectral water quality analyzer (2); Step 2: Use hyperspectral data to train a random forest model to obtain the characteristic importance score of the wavelength of each water quality monitoring parameter, and select the top-ranked wavelengths for each water quality monitoring parameter based on the score ranking; Step 3: Calculate the Pearson correlation coefficient between each water quality monitoring parameter wavelength and the target variable; sort by the absolute value of the correlation coefficient and select the top-ranked wavelengths for each water quality monitoring parameter; Step 4: Find the wavelengths that appear in the top several wavelengths selected in step 2 and the top several wavelengths selected in step 3, as the final characteristic wavelength set; Step 5: Use the final selected characteristic wavelength set to train and verify the model.

Images