Intelligent phosphate radical monitoring method based on ion selective electrochromic electrode and image recognition

Abstract

The invention discloses an intelligent phosphate radical monitoring method based on an ion selective electrochromic electrode and image recognition, and belongs to the technical field of environment monitoring. In order to solve the problems that traditional phosphate radical detection depends on chemical reagents, steps are tedious and online monitoring is difficult to realize, the invention provides an intelligent monitoring strategy integrating electrochemical excitation, color response and image recognition. According to the method, an electrochromic sensing electrode with a phosphate ion selective response characteristic is constructed, phosphate radicals are driven to be captured and released under the action of set voltage, and the color of the electrode is induced to be reversibly changed; by combining image acquisition and a color recognition algorithm, extracting color features and training a concentration prediction model based on a standard concentration data set, real-time monitoring of phosphate radicals in different concentration ranges can be realized, and meanwhile, the system has the capabilities of model switching, electrode recovery judgment and automatic cycle monitoring. The method does not need chemical reagents, is sensitive in response, simple and convenient to operate and accurate in identification, and is suitable for on-site visual monitoring of phosphate radicals in various water bodies such as domestic sewage, aquaculture wastewater and surface water.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring, specifically to an intelligent monitoring method for phosphate ions based on ion-selective electrochromic electrodes and image recognition, belonging to the category of visual and automated monitoring technology for phosphate ions in water bodies. Background Technology

[0002] Phosphate is a typical nutrient ion in water bodies. Improper discharge of phosphate can lead to eutrophication, algal blooms, and ecosystem imbalance, seriously threatening aquatic environmental safety and human health. Therefore, it has become a crucial target for water environment management and non-point source pollution control. Existing phosphate detection technologies, such as the ammonium molybdate colorimetric method and ion chromatography, can achieve high sensitivity and accuracy under laboratory conditions. However, they generally suffer from drawbacks such as complex operation, reliance on chemical reagents, long detection cycles, or difficulty in achieving in-situ real-time monitoring, limiting their widespread application in continuous on-site water quality monitoring and automated response.

[0003] Electrochromism refers to the phenomenon where a material's color or optical properties change reversibly under an applied voltage. It boasts advantages such as low energy consumption, fast response, rich colors, and strong controllability. These materials are widely used in smart windows, flexible display devices, and visual energy storage systems. In these applications, electrochromic electrodes not only perform traditional electron conversion functions but also adsorb or embed target ions at specific potentials, inducing regular color changes and providing intuitive optical indicators of ion concentration. If electrochromic materials selective for phosphate ions can be introduced into environmental monitoring systems, combined with electrochemical regulation and image recognition techniques, and utilizing machine learning modeling, a novel, reagent-free, recyclable, and real-time monitoring platform can be constructed. Therefore, this technology, integrating electrochemical response, color change, and intelligent recognition, holds promise for achieving visual response and intelligent recognition of target ions, providing a new pathway for reagent-free, on-site, automated, and portable water quality monitoring. Summary of the Invention

[0004] This invention addresses the problems of existing phosphate detection technologies, such as reliance on reagents, cumbersome procedures, difficulty in in-situ real-time monitoring, and limited response range. It proposes an intelligent phosphate monitoring method based on ion-selective electrochromic electrodes and image recognition. By combining electrochromic reactions, electrochemical responses, image recognition, and machine learning model prediction technologies, it achieves rapid reagent-free detection, intelligent identification, and system linkage of phosphate ion concentration in water.

[0005] The purpose of this invention is to construct an intelligent phosphate detection method that requires no chemical indicators, has a sensitive response, is easy to operate, and is suitable for real-time on-site monitoring. It can also be extended to automatic analysis and remote early warning of water quality parameters. This method employs a specially designed electrochromic electrode material with ion selectivity for phosphate, such as a ferrocene-polyaniline composite material. Under a set voltage, the electrode reversibly binds to and releases phosphate, accompanied by a noticeable color change. By setting the voltage and duration to regulate the redox state of the electrode material, and with the accompanying image acquisition and analysis module, the color change of the electrode is accurately acquired and analyzed, thereby achieving highly sensitive quantitative identification of phosphate concentration. Simultaneously, the system incorporates multiple color-concentration mapping models and supports dynamic switching or adjustment of model parameters according to the monitoring environment. The concentration recognition range and judgment accuracy can be flexibly set according to water quality conditions to adapt to the concentration range and recognition requirements of different application scenarios. Furthermore, this invention uses a reverse voltage excitation and color comparison mechanism to determine whether the electrode response has fully recovered, supporting status judgment and system self-checking during continuous monitoring. It can also be extended to combine concentration threshold settings to achieve remote intelligent early warning and system linkage control, further improving monitoring efficiency and response capabilities. Compared with traditional detection methods, this invention does not require complex reagents and precision equipment, which not only simplifies the operation process and improves monitoring efficiency, but also enhances system stability and adaptability to field applications, providing innovative technical support for building a green, efficient and intelligent water phosphorus pollution monitoring system.

[0006] The present invention provides a method for intelligent monitoring of phosphate ions based on ion-selective electrochromic electrodes and image recognition, which is carried out according to the following steps:

[0007] Step 1: Constructing the sensing electrode: Constructing an electrochromic material film that is selective for phosphate ions on a transparent conductive substrate. The film can undergo a reversible color change when a specific voltage is applied.

[0008] Step 2, Electrochemical Excitation and Image Acquisition: a) Immerse the constructed sensing electrode and inert counter electrode together in the water body to be tested, and connect the voltage control module; b) Apply and maintain a set positive voltage to drive the electrode to capture phosphate ions and induce a color change in the electrode. Then apply and maintain a set reverse voltage to drive the release of phosphate ions and restore the electrode color; c) Under a closed lighting environment, use a fixed-view image acquisition device to acquire electrode images after the application of positive and reverse voltages, respectively. The acquired electrode image with the applied positive voltage is used for concentration monitoring, and the acquired electrode image with the applied reverse voltage is used to determine whether the electrode has recovered to the state corresponding to the initial reference image.

[0009] Step 3, Image Processing and Color Feature Extraction: Image processing algorithms are used to identify the sensing electrode region in the image acquired in Step 2, and color space conversion, normalization, and noise filtering are performed. Color feature values ​​within the electrode region are extracted as input variables for subsequent concentration prediction.

[0010] Step 4: Establish a concentration prediction model library: a) Prepare multiple standard solutions with known phosphate concentrations, excite the constructed sensing electrode according to the voltage and duration set in Step 2, acquire images, and extract color feature values ​​according to Step 3; b) Construct a dataset of color features and corresponding concentrations, and divide it into training and testing sets. Use machine learning algorithms to train prediction models and use cross-validation to establish a color feature-concentration mapping relationship; c) Divide different concentration ranges into sub-intervals according to usage requirements, set different forward and reverse voltages and their durations for each sub-interval, construct multiple prediction models according to steps a) and b) above, and form a backup model library;

[0011] Step 5, Intelligent Monitoring and Model Adaptive Adjustment: The prediction model constructed in Step 4 is used to identify the phosphate concentration in the water body to be tested in real time. When the predicted value exceeds the applicable concentration range of the current model, the system automatically switches to the appropriate prediction model in the model library and adjusts the excitation voltage and duration of the model accordingly to continue monitoring.

[0012] Step Six: Cyclic Monitoring and Electrode Recovery Judgment: After each round of monitoring, a reverse voltage is applied to acquire electrode images and compare them with the initial state images of the electrodes. The results are then calculated using CIELAB to determine the relationship between the images and the initial reference images. Color difference is assessed to determine if the color has been restored. If the preset color deviation threshold is met, the next round of monitoring is initiated. Otherwise, the reverse voltage time is extended and the assessment is repeated. If the color is still not restored after multiple assessments, an electrode replacement prompt is triggered, and the current cycle is terminated.

[0013] The preset color deviation threshold is 1 to 3, preferably 2.

[0014] The image acquisition is performed immediately after the voltage is applied, with a delay of no more than 1 second.

[0015] Furthermore, the electrochromic response material is a material with phosphate selective binding properties, and the electrochromic response material is a ferrocene-polyaniline composite material with a phosphate ion selectivity coefficient greater than 5.0. The molar ratio of ferrocene to polyaniline in the ferrocene-polyaniline composite material is 1:5 to 1:15.

[0016] A forward voltage drives the sensing electrode to capture phosphate ions, and the electrode color gradually changes from yellow to blue or brown. Then, a reverse voltage drives the phosphate ions to be released into the solution, and the electrode color returns to normal.

[0017] Furthermore, during the forward and reverse voltage operations, the sensing electrodes serve as the anode and cathode, respectively, with a voltage range of 0.5~1.5 V and a duration of 1~60 s.

[0018] Furthermore, the applicable concentration range in step five refers to the concentration range in which the prediction model, under the corresponding electrochemical parameter conditions, satisfies the cross-validation requirement of a prediction accuracy ≥ 0.98 and a mean absolute error ≤ 5%.

[0019] Furthermore, the image processing and color feature extraction include: identifying the sensing electrode region in the image using an edge detection algorithm; converting the RGB image to the HSV color space and normalizing it; removing image noise using a Gaussian filtering algorithm; and extracting color feature values, wherein the color feature values ​​are the average of the HSV three channels and the RGB three channels, and the value compared to the initial image. Color difference value.

[0020] Furthermore, the concentration prediction model is trained based on 0.05~100 mM phosphate standard solution, and multiple prediction models are constructed by dividing different concentration ranges and action parameters.

[0021] Furthermore, the aforementioned training of the prediction model using machine learning algorithms employs support vector regression, random forest, and neural network algorithms, achieving a prediction accuracy R0. 2 If the value is ≥0.98, use K-fold cross-validation with K value of 5 or 10.

[0022] Furthermore, the intelligent monitoring method is configured in a cyclic monitoring mode to perform electrode self-cleaning and periodic reverse voltage control; the cyclic monitoring supports automatic electrode recovery judgment, comparing the reverse voltage image with the initial state image after each round of monitoring, and if there is a color difference... If the color deviation is below the preset color deviation threshold, the electrode is determined to have recovered and the next round of monitoring is started. Otherwise, the reverse voltage duration is extended and the image judgment is repeated. If the image is still not recovered after repeated judgments, the electrode is prompted to be replaced and the current monitoring cycle is terminated. The preset color deviation threshold is 1 to 3.

[0023] Furthermore, the aforementioned intelligent phosphate monitoring method supports remote data synchronization, image uploading, and cloud-based model calculation, and features functions such as recording concentration exceedances, generating historical trend charts, and sending early warnings to mobile devices.

[0024] Furthermore, the aforementioned intelligent phosphate monitoring method is suitable for on-site monitoring of phosphate ions in domestic sewage, aquaculture wastewater, and surface water.

[0025] The present invention has the following beneficial effects:

[0026] This invention provides an intelligent phosphate monitoring method based on ion-selective electrochromic electrodes and image recognition. It integrates the high selectivity of electrochemical reactions with the visual readability of electrochromic processes, achieving highly sensitive, selective, reversible, and automated identification of phosphate ions in water without the need for external chemical reagents. This method constructs a mapping relationship between phosphate concentration and color changes through image acquisition, color feature extraction, and machine learning modeling techniques. It possesses rapid response, intelligent analysis, and remote early warning capabilities, making it suitable for online phosphorus pollution monitoring and early warning in various scenarios. Its beneficial effects are mainly reflected in the following aspects:

[0027] 1. Rapid and reversible colorimetric response, avoiding reagent interference and external disturbances: This invention uses an electrochromic material with selective binding ability to phosphate ions to construct the sensing electrode. It utilizes the reversible and stable color change generated under voltage drive to directly respond to changes in phosphate concentration, avoiding the reaction interference and reagent consumption problems common in traditional chemical colorimetric methods, and improving system stability and ease of operation.

[0028] 2. High accuracy in image recognition and modeling, and sensitive quantitative concentration identification: This invention integrates an image acquisition module and an image recognition algorithm, combining machine learning algorithms such as support vector regression, random forest, and neural networks to construct a high-precision prediction model between color features and phosphate concentration, achieving high fitting accuracy (R²) within the concentration range of 0.05~100 mM. 2 (≥0.98) Quantitative identification, high detection accuracy, fast response speed, suitable for online monitoring needs.

[0029] 3. Supports cyclic monitoring and dynamic model updates, with strong adaptability: This invention features reversible response and electrode state recognition functions, supporting continuous cyclic monitoring and automatic reverse voltage recovery. Simultaneously, it can dynamically call upon the prediction model and voltage parameters according to actual water conditions, achieving adaptive model updates and ensuring monitoring accuracy under different water quality backgrounds.

[0030] 4. The system boasts a high degree of intelligence and versatile scalability: This invention supports setting concentration thresholds and linking them with audible and visual alarms or remote communication modules to achieve automatic early warning, enabling automatic early warning, data uploading, and trend analysis. The constructed electrochromic-image recognition framework possesses excellent scalability; after adjusting the electrode response mechanism and modeling parameters, it is expected to be extended for intelligent monitoring of other inorganic anions (such as nitrate and fluoride ions). Attached Figure Description

[0031] Figure 1 This is a flowchart of a smart phosphate monitoring method based on ion-selective electrochromic electrodes and image recognition.

[0032] Figure 2Images of the ferrocene-polyaniline thin film electrode after being subjected to a voltage of 1.2 V for 30 s in phosphate solutions of different concentrations (1~20 mM);

[0033] Figure 3 Flowchart for electrode color feature extraction and modeling;

[0034] Figure 4 The figure shows the fitting effect of the phosphate concentration prediction model for testing. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the spirit of the contents disclosed in the present invention will be described in detail below. After understanding the embodiments of the present invention, any person skilled in the art can make changes and modifications based on the technology taught in the present invention without departing from the spirit and scope of the present invention.

[0036] The illustrative embodiments and descriptions of the present invention are used to explain the present invention, but are not intended to limit the present invention.

[0037] Example 1:

[0038] This embodiment discloses a smart phosphate monitoring method based on an ion-selective electrochromic electrode and image recognition. It combines electrochemical excitation with image recognition to achieve visualized and quantitative detection of phosphate ions in water. The method flow is as follows: Figure 1 As shown, the main steps include electrode response excitation, electrode color change, image acquisition and processing, color feature extraction, prediction model invocation, concentration output and threshold judgment, and electrode regeneration. It has the advantages of simple operation, reusability, and high recognition accuracy.

[0039] The specific implementation steps of the embodiment are as follows:

[0040] 1. Constructing a phosphate-responsive sensing electrode

[0041] Ferrocene-polyaniline (Fc-PANI) thin-film electrode was selected as the sensing electrode. On one hand, the ferrocene in this electrode structure can form hydrogen bonds with phosphate ions, achieving highly selective recognition; on the other hand, it can capture or release phosphate ions under an applied voltage, exhibiting significant electrochromic properties and good electrochemical reversibility and response stability. The specific preparation method is as follows: 1,1'-ferrocene dianiline was prepared by amidation reaction, dissolved in 1 M HCl solution with aniline at a molar ratio of 1:8, and ultrasonically dispersed in an ice bath for 30 min. Ammonium persulfate (1.2 times the molar amount of aniline) was added as an oxidant to initiate chemical oxidative polymerization. The reaction was carried out at room temperature for 24 h, followed by centrifugation, washing, and vacuum drying to obtain Fc-PANI powder. The obtained powder was dispersed in an ethanol solution containing Nafion and ultrasonically dissolved for 30 min to obtain a homogeneous slurry. The slurry was coated onto the surface of a pre-cleaned ITO conductive glass (sheet resistance <10 Ω / □) using a spin coating method (2000 rpm, 30 s), and then vacuum dried at 80 ℃ to form an Fc-PANI thin film electrode.

[0042] 2. Electrochemical Excitation and Color Image Acquisition

[0043] The prepared Fc-PANI sensing electrode and platinum wire counter electrode were immersed together in the test solution, and a positive voltage of 1.2 V (with the sensing electrode acting as the anode) was applied for 30 s to drive the sensing electrode to capture phosphate ions, initiating Fc→Fc + The oxidation process of (Fe(II)→Fe(III)) causes a change in the color of the sensing electrode. Figure 2 The color response of the Fc-PANI sensing electrode in phosphate solutions with different concentrations ranging from 1 to 20 mM was demonstrated. The color gradually transitioned from light yellow to brown with increasing concentration, exhibiting good visual distinguishability. After the above process, a reverse voltage of -1.2 V (with the sensing electrode acting as the cathode) was applied for 30 seconds to drive the release of phosphate ions and restore the initial color of the sensing electrode. Image acquisition was conducted in a closed lighting environment using a fixed-angle camera (1920×1080 resolution) to immediately capture images of the sensing electrode after both the forward and reverse voltage applications. A uniform blue LED light source (λ=450 nm) was used to ensure consistent image brightness and color contrast, improving subsequent recognition accuracy. The image of the sensing electrode with the applied forward voltage was used for phosphate concentration monitoring, while the image with the applied reverse voltage was used to determine whether the sensing electrode had recovered to the state corresponding to the initial reference image.

[0044] 3. Image processing and color feature extraction

[0045] The acquired electrode images are processed as follows: First, an edge detection algorithm is used to automatically identify the electrode regions in the image; second, Gaussian filtering is applied to the image of these regions to smooth the image and suppress noise; then, a grayscale world algorithm is applied to perform white balance correction to eliminate the color cast caused by differences in illumination; next, the image is converted from the RGB color space to the HSV color space and its brightness is normalized; finally, the HSV channel hue values ​​of the electrode regions are extracted as color feature values, which serve as input variables for subsequent concentration prediction.

[0046] 4. Establishment and Real-time Monitoring of Phosphate Concentration Prediction Model

[0047] like Figure 3 As shown, model building includes three processes: image processing, data preparation, and model training. Multiple standard solutions with known phosphate concentrations were prepared, and images were acquired from the Fc-PANI sensing electrode after applying a preset forward voltage for a set duration. The images were processed to extract color features, and a dataset was constructed using the corresponding known concentration data. This dataset was divided into training and testing sets in an 8:2 ratio, and 10-fold cross-validation was employed to improve the model's generalization ability. During the model training phase, a machine learning algorithm, such as random forest, was used to establish the mapping relationship between color features and phosphate concentration. Considering that the color response is related to voltage, duration, and target ion concentration, multiple high-precision sub-models can be constructed using different voltages and durations in different concentration ranges. Therefore, sub-ranges with different phosphate concentration ranges were defined according to usage requirements, and different forward and reverse voltages and durations were set for each sub-range. Multiple phosphate concentration prediction models were constructed based on the above steps and stored in a backup model library.

[0048] Figure 4 The fitting performance of the prediction models constructed in three concentration range sub-intervals is demonstrated. Model A is applicable to a concentration range of 1-20 mM, with electrochemical conditions of a forward voltage of 1.2 V for 30 s and a reverse voltage of -1.2 V for 30 s. It is trained using a random forest algorithm, and the fitting accuracy on the test set is R0. 2 =0.998, RMSE = 0.273mM; Model B is applicable to the concentration range of 0.05-2 mM, with electrochemical conditions of forward voltage 1.5 V for 30 s and reverse voltage -1.5 V for 30 s. It is trained based on an artificial neural network regression algorithm, and the fitting accuracy on the test set is R0. 2 =0.980, RMSE = 0.0814 mM; Model C is applicable to the concentration range of 20-100 mM, with electrochemical conditions of forward voltage 1.2 V for 10 s and reverse voltage -1.2 V for 10 s. It is trained based on the random forest algorithm, and the fitting accuracy on the test set is R0.2 =0.991, RMSE = 2.277 mM. This shows that the fitting accuracy of all three models is relatively good, Rm = 0.991. 2 All are ≥0.98. These three models cover different concentration ranges under their respective electrochemical parameters. In actual monitoring, the system can automatically switch and adjust the corresponding forward and reverse voltages and durations based on the real-time monitoring results and the applicable range of the model, so as to achieve intelligent prediction of concentration range adaptability.

[0049] 5. Electrode regeneration and circulation monitoring mechanism

[0050] After completing one round of monitoring and outputting the concentration results, the system immediately applies a preset reverse voltage and maintains it for a certain duration (e.g., −1.2 V, 30 s), causing the Fc in the electrode to... ＋ The electrode is reduced to Fc, driving the release of phosphate ions, and the electrode color returns to its initial light yellow. This process is called electrode regeneration. The system will automatically acquire images after regeneration and calculate their regeneration ratio with the initial reference image. Color difference The value can be calculated using the CIELAB standard formula. If <2, indicating the electrode has fully recovered, the next monitoring cycle can begin; if If the value exceeds the limit, the electrode regeneration is automatically repeated by applying a reverse voltage, up to three times. If recovery is still not achieved after three times, an electrode abnormality is indicated, and the cyclic monitoring is terminated. This intelligent monitoring method has complete cyclic detection capabilities, with each round of detection consisting of four stages: ion capture, color recognition, ion release, and electrode recovery. These four stages form a complete closed loop, supporting multiple rounds of continuous monitoring. It exhibits high repeatability and stability, making it suitable for complex scenarios such as online environmental monitoring.

Claims

1. A smart monitoring method for phosphate ions based on ion-selective electrochromic electrodes and image recognition, characterized in that, It is done in the following steps: Step 1: Constructing the sensing electrode: Construct an electrochromic material film layer that is selective for phosphate ions on a transparent conductive substrate as a sensing electrode. The sensing electrode can undergo a reversible color change when a voltage is applied. Step 2, Electrochemical Excitation and Image Acquisition: a) Immerse the constructed sensing electrode and inert counter electrode together in the water to be tested, and connect the voltage control module; b) Apply the set positive voltage and duration to drive the electrode to capture phosphate ions and induce a color change in the sensing electrode. Then apply the set reverse voltage and duration to drive the release of phosphate ions and restore the initial color of the sensing electrode; c) Under a closed lighting environment, use a fixed-view image acquisition device to acquire images of the sensing electrode after applying the positive and reverse voltages, respectively. The acquired images of the sensing electrode with the positive voltage applied are used for phosphate concentration monitoring, and the acquired images of the sensing electrode with the reverse voltage applied are used to determine whether the sensing electrode has recovered to the state corresponding to the initial reference image. Step 3, Image Processing and Color Feature Extraction: Image processing algorithms are used to identify the sensor electrode region in the sensor electrode images acquired in Step 2 after applying positive and reverse voltages. Color space conversion, normalization, and noise filtering are performed, and color feature values ​​within the electrode region are extracted as input variables for subsequent concentration prediction. Step 4: Establish a phosphate concentration prediction model library: a) Prepare multiple standard solutions with known phosphate concentrations, excite the constructed sensing electrode according to the voltage and duration set in Step 2, acquire images, and extract color feature values ​​according to Step 3. b) Construct a dataset of color features and corresponding phosphate concentrations, and divide it into training and testing sets. Use machine learning algorithms to train prediction models and use cross-validation to establish a color feature-concentration mapping relationship; c) Divide the data into sub-intervals with different phosphate concentration ranges according to usage requirements. Set different forward and reverse voltages and durations for each sub-interval. Construct multiple phosphate concentration prediction models based on steps a) and b) above, and form a backup model library. Step 5: Intelligent monitoring and adaptive model adjustment: The phosphate concentration prediction model constructed in Step 4 is used to identify the phosphate concentration in the water body under test in real time. When the predicted value exceeds the applicable concentration range of the current model, the system automatically switches to the appropriate phosphate concentration prediction model in the model library and adjusts the excitation voltage and duration of the phosphate concentration prediction model accordingly to continue monitoring. Step Six: Cyclic Monitoring and Electrode Recovery Judgment: After each round of monitoring, a reverse voltage is applied, and the sensor electrode image is acquired and compared with the initial state image of the sensor electrode. The result is calculated using CIELAB based on the initial reference image. The system measures color difference and sets a preset color deviation threshold to determine whether the sensor electrode color has recovered to its initial color. If the preset color deviation threshold is met, the system proceeds to the next monitoring round. Otherwise, the reverse voltage duration is extended and the judgment is repeated. If the color still fails to recover after multiple judgments, an electrode replacement prompt is triggered, and the current cycle is terminated.

2. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The electrochromic response material is a material with phosphate selective binding properties. The electrochromic response material is a ferrocene-polyaniline composite material with a phosphate ion selectivity coefficient greater than 5.

0. The molar ratio of ferrocene to polyaniline in the ferrocene-polyaniline composite material is 1:5 to 1:

15.

3. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, During the forward and reverse voltage operations, the sensing electrodes serve as the anode and cathode, respectively, with a voltage range of 0.5~1.5 V and a duration of 1~60 s.

4. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The applicable concentration range in step five refers to the concentration range in which the prediction model, under the corresponding electrochemical parameter conditions, meets the requirements of a prediction accuracy ≥ 0.98 and a mean absolute error ≤ 5% after cross-validation.

5. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The image processing and color feature extraction include: identifying the sensing electrode region in the image using an edge detection algorithm; converting the RGB image to the HSV color space and normalizing it; removing image noise using a Gaussian filtering algorithm; and extracting color feature values, which are the average values ​​of the HSV three channels, the RGB three channels, and the values ​​compared to the initial image. Color difference value.

6. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The concentration prediction model is trained based on 0.05~100 mM phosphate standard solution, and multiple prediction models are constructed by dividing different concentration ranges and action parameters.

7. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The aforementioned training of the prediction model using machine learning algorithms refers to the use of support vector regression, random forest, or neural network algorithms, with a prediction accuracy R. 2 If the value is ≥0.98, use K-fold cross-validation with K value of 5 or 10.

8. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The intelligent monitoring method is configured in a cyclic monitoring mode to perform electrode self-cleaning and periodic reverse voltage control. The cyclic monitoring supports automatic electrode recovery judgment; after each round of monitoring, the reverse voltage image is compared with the initial state image. If there is a color difference... If the color deviation is below the preset color deviation threshold, the electrode is determined to have recovered and the next round of monitoring is started. Otherwise, the reverse voltage duration is extended and the image judgment is repeated. If the image is still not recovered after repeated judgments, the electrode is prompted to be replaced and the current monitoring cycle is terminated. The preset color deviation threshold is 1 to 3.

9. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The aforementioned intelligent phosphate monitoring method supports remote data synchronization, image uploading, and cloud-based model calculation, and features functions such as recording concentration exceedances, generating historical trend charts, and sending early warning push notifications to mobile devices.

10. The intelligent phosphate monitoring method based on ion-selective electrochromic electrode and image recognition according to claim 1, characterized in that, The aforementioned intelligent phosphate monitoring method is suitable for on-site monitoring of phosphate ions in domestic sewage, aquaculture wastewater, and surface water.

Images