An electrocatalytic treatment method for removing uranium-containing wastewater

Abstract

The invention discloses an electrocatalytic treatment method for removing uranium-containing wastewater, and relates to the technical field of electrocatalysis. The invention utilizes an embedded sensor network to collect data such as the concentration, conductivity, and pH value of uranyl ions U(VI) and coexisting metal ions in wastewater in real time, corrects time series data through DTW, and eliminates time dimension inconsistency deviations. Based on preprocessed data, a deep learning model combining a convolutional neural network (CNN) and a recurrent neural network (RNN) is constructed. The CNN is used to extract the spatial characteristics of ion concentration distribution, and the RNN captures the time-dependent characteristics of dynamic changes in ions. An attention mechanism is introduced to focus on key interfering ions, and a characteristic matrix for priority processing is generated. The interfering ion characteristics output by the model are predicted and their dynamic influence on the selective adsorption of U(VI), so as to further optimize electrocatalytic operation parameters to balance adsorption efficiency and energy consumption. At the same time, global optimization is performed in combination with a comprehensive objective function to optimize the adsorption efficiency of uranyl ions.

Description

Technical Field

[0001] The present invention relates to the field of electrocatalysis technology, in particular to an electrocatalytic treatment method for removing uranium-containing wastewater. Background Art

[0002] A large amount of uranium-containing wastewater is discharged during the nuclear industry, uranium mining and smelting. The wastewater contains uranyl ions (UO2 2+ , namely U(VI)), is often accompanied by a variety of coexisting metal ions, such as calcium ions, magnesium ions, iron ions, and other non-metallic anions such as chloride ions and sulfate ions. These wastewater components make it particularly difficult to extract U(VI) from it. Since uranyl ions are chemically similar to some metal ions, these ions will be competitively adsorbed or reduced in traditional electrocatalytic treatment methods, thereby significantly interfering with the selective extraction of uranium.

[0003] In traditional electrocatalytic methods, most use specific electrode materials or optimize reaction conditions to improve the selectivity for U(VI). However, in complex wastewater environments, there will be competitive adsorption of other metal ions, which can easily lead to a decrease in uranium extraction efficiency. In addition, traditional methods have poor adaptability to different water qualities, and the treatment performance can easily be significantly reduced in wastewater with high concentrations of coexisting ions.

[0004] In order to cope with the interference of complex water quality on the selective extraction of U(VI), some traditional schemes introduce functional groups with coordination selectivity (such as amide oxime or phosphonic acid groups) to enhance the specific adsorption of uranium; or use dynamic electric field control technology to increase the migration rate of uranyl ions to the electrode while reducing the adsorption of other ions; however, these methods still face problems of adsorption saturation, high energy consumption and environmental unfriendliness in complex systems, and are difficult to meet actual needs; therefore, how to further improve the selective extraction ability of U(VI) and reduce the interference of coexisting ions remains an important challenge that needs to be solved in electrocatalytic treatment technology. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides an electrocatalytic treatment method for removing uranium-containing wastewater to solve the problems that traditional electrocatalytic methods are difficult to cope with the dynamic interference of metal ions in complex wastewater and the reduction of selective extraction efficiency.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The embodiment of the present invention provides an electrocatalytic treatment method for removing uranium-containing wastewater, which comprises:

[0009] Step S1, deploying a monitoring sensor network to collect real-time wastewater composition data during wastewater treatment and preprocessing the collected data;

[0010] Step S2, based on the wastewater composition data pretreated in step S1, constructing a deep learning model to analyze the characteristics of the multi-ion system in the wastewater components and predict the interference effect during the U(VI) extraction process;

[0011] Step S3, adjusting electrocatalytic operation parameters based on the interference prediction result of step S2;

[0012] Step S4: Continuously monitor the concentration changes of U(VI) and other metal ions in the wastewater under the adjusted operating parameters, input the monitoring data into the deep learning model of step S2, update the interference effect prediction results in real time, and adjust the electrocatalytic operating parameters.

[0013] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the wastewater components include uranyl ion U(VI) concentration, coexisting metal ion concentration, conductivity and pH value;

[0014] The data preprocessing method is: normalizing the data, eliminating abnormal data, and correcting the time series data using the dynamic time alignment algorithm DTW.

[0015] As a preferred solution of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the step of correcting the time series data using the dynamic time alignment algorithm DTW is as follows:

[0016] Construct a dynamic programming matrix and calculate the cumulative alignment distance between the target time series and the actual time series based on the point-to-point difference between the two series. The alignment distance calculation formula is:

[0017]

[0018] Where D(i, j) represents the cumulative alignment distance between time series X and Y at positions i and j, and d(x i ,y j ) represents the Euclidean distance between X and Y at corresponding points i and j, calculated as |x i -y j |,x i Represents the value of the time series X at the i-th time point, where i ranges from 1, 2, ..., n, y jrepresents the value of time series Y at the jth time point, where j ranges from 1, 2, ..., m. D(i-1, j) represents the cumulative alignment distance from the i-1th point of sequence X to the jth point of Y. D(i, j-1) represents the cumulative alignment distance from the i-th point of sequence X to the j-1th point of Y. D(i-1, j-1) represents the cumulative alignment distance from sequences X and Y to both i-1 and j-1.

[0019] By backtracking the dynamic programming matrix, the optimal alignment path between the two time series is obtained. The optimal distance calculation formula is:

[0020] P={(p1, q1), (p2, q2),…, (p k ,q k )},

[0021] Among them, P represents the index set of the optimal alignment path, p i represents the index of the i-th alignment point of the time series X in the path, q i represents the index of the i-th alignment point of time series Y in the path, k represents the total length of the alignment path, and satisfies k ≥ max(n, m), where n and m are the lengths of X and Y respectively;

[0022] According to the optimal alignment path, the collected time series data is nonlinearly resampled to generate correction data consistent with the reference time series, which is expressed as:

[0023]

[0024] Where X′ represents the time series X after correction by the alignment path, Indicates the time series X corresponding to index p in the alignment path P i The value of , k represents the total length of the alignment path.

[0025] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the characteristics of the multi-ion system include the spatial characteristics of ion concentration distribution and the time-dependent characteristics of ion dynamic changes;

[0026] The deep learning model includes:

[0027] Convolutional neural network (CNN) is used to extract the spatial characteristics of ion concentration distribution in wastewater, specifically the concentration correlation between ions;

[0028] A recurrent neural network (RNN) is used to capture the time-dependent characteristics of ion dynamic changes;

[0029] The interference effect prediction output includes: the main characteristics of the interfering ions and their dynamic effects on the selective adsorption of U(VI).

[0030] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the step of extracting the spatial characteristics of the ion concentration distribution in the wastewater using a convolutional neural network (CNN) is as follows:

[0031] The convolutional neural network (CNN) was used to process the corrected time series data, extract the concentration correlation between different ions in the multi-ion system, and analyze the spatial characteristics of the ion concentration distribution. The convolution extraction formula is:

[0032]

[0033] Among them, H(i, j) represents the value of the i-th row and j-th column in the feature matrix generated after the convolution operation, σ(·) represents the ReLU activation function, σ(x) = max(0, x), F(i+p, q) represents the sub-region value starting from position (i, q) in the input data matrix F, W(p, q) represents the weight value of the p-th row and q-th column in the convolution kernel weight matrix, the size of the convolution kernel is k×m, b represents the bias of the convolution operation, k represents the row dimension of the convolution kernel, and m represents the column dimension of the convolution kernel, which is consistent with the number of ion species in the input matrix.

[0034] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the step of using a recurrent neural network (RNN) to capture the time-dependent characteristics of the dynamic changes of ions is as follows:

[0035] Using the recurrent neural network (RNN), the extracted spatial characteristics are analyzed in time series to capture the dynamic change characteristics of the multi-ion system and reflect the time dependence of the ion concentration change. The analysis formula is:

[0036] h t =f(W h h t-1 +W x x t +b h ),

[0037] Among them, h t represents the hidden state vector of the recurrent neural network at time step t, f(·) represents the Tanh function, and W h Represents the weight matrix from the previous time step hidden state to the current hidden state, h t-1 represents the hidden state vector at the previous time step t-1, W x Represents the weight matrix from the input vector to the hidden state at the current time step, x t Represents the input vector of the current time step, which comes from the characteristics of the F matrix at time step t, b h Bias vector representing the hidden state;

[0038] In step S2, the attention mechanism is introduced to determine the interfering ions that have the greatest impact on the selective adsorption of uranyl ions based on the dynamic characteristics of the model output, and generate a priority feature matrix. The attention weight calculation formula is:

[0039]

[0040]

[0041] Among them, α i represents the attention weight of ion i, e i represents the feature score of ion i, v represents the projection vector calculated by attention weight, W h The weight matrix representing the linear transformation of the hidden state, h t represents the hidden state vector at the current time step t, W s The linear transformation weight matrix representing ion characteristics, s i represents the eigenvector of ion i,

[0042] Weighted interference characteristic output formula:

[0043] H f =∑ i α i s i ,

[0044] Among them, H f represents the interference ion feature matrix after weighting by the attention mechanism, α i represents the attention weight of ion i, s i represents the eigenvector of ion i.

[0045] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the operating parameter adjustment method includes:

[0046] Adjust the electric field strength and electrode voltage;

[0047] The reaction time was adjusted to balance the adsorption efficiency and energy consumption.

[0048] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the step of adjusting the electrocatalytic operating parameters based on the interference prediction result of step S2 is as follows:

[0049] According to the influence of interfering ions on the adsorption efficiency of uranyl ions, the electric field strength and electrode voltage are dynamically adjusted. The target optimized electric field strength is set as E opt , the target electrode voltage after optimization is V opt , the optimization formula is:

[0050] Eopt =E init +ΔE,

[0051] V opt =V init +ΔV,

[0052] Among them, E opt represents the optimized electric field strength, E init represents the initial electric field strength, ΔE represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results, V opt Represents the optimized electrode voltage, V init Represents the initial electrode voltage, ΔV represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results.

[0053] Assume the target adsorption efficiency is η t , the current adsorption efficiency is η c , the reaction time adjustment formula is:

[0054]

[0055] Among them, t opt represents the optimized reaction time, t init represents the initial reaction time, η t represents the target adsorption efficiency, η c Indicates the current adsorption efficiency,

[0056] Define the comprehensive optimization objective function J, which integrates adsorption efficiency, residual concentration and energy consumption. The objective function formula is:

[0057] J=ω1·η+ω2·(1-C e )-ω3·E c ,

[0058] Among them, J represents the comprehensive optimization target value, ω1, ω2, ω3 are weight factors, which control the importance of adsorption efficiency, residual concentration and energy consumption respectively, η represents the adsorption efficiency, C e represents the residual concentration, which is equal to the concentration of unadsorbed uranyl ions, E c Represents energy consumption, which is dynamically calculated based on electric field strength and reaction time.

[0059] By gradient descent method, E opt , V opt , t opt Perform iterative updates to maximize J.

[0060] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, in step S4, an embedded online sensor is used for real-time monitoring, and the online sensor collects dynamic concentration data of uranyl ions U(VI) and coexisting metal ions in the wastewater.

[0061] As a preferred embodiment of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention, the step of updating the interference effect prediction result in real time and adjusting the electrocatalytic operation parameters is as follows:

[0062] Using embedded online sensors, the concentration change data of uranyl ions and other metal ions in wastewater are collected in real time. The real-time monitoring data is Z t ={Z t，1 , z t，2 ,…,z t，m}, represents the ion concentration vector at time t, and the monitoring data is input into the deep learning model for prediction.

[0063] The deep learning model outputs the updated interference prediction result H′(t), which is formulated as follows:

[0064] H′(t)=f model (Z t ),

[0065] Among them, H′(t) represents the interference prediction characteristic matrix after the update at time t, f model represents the deep learning model constructed in step S2, Z t represents the monitoring data vector at time t;

[0066] According to the updated interference prediction result H′(t), the operating parameters are adjusted as follows:

[0067] E opt =E init +ΔE t , V opt =V init +ΔV t ,

[0068] Where ΔE t and ΔV t They respectively represent the electric field intensity and electrode voltage change adjusted according to the prediction results at time t.

[0069] The beneficial effects of the present invention are as follows: the present invention utilizes an embedded sensor network to collect data such as the concentration, conductivity, and pH value of uranyl ions U(VI) and coexisting metal ions in wastewater in real time, corrects time series data through DTW, and eliminates inconsistency deviations in the time dimension; based on preprocessed data, a deep learning model combining a convolutional neural network (CNN) and a recurrent neural network (RNN) is constructed, wherein the CNN is used to extract the spatial characteristics of ion concentration distribution, and the RNN captures the time-dependent characteristics of dynamic changes in ions, and an attention mechanism is introduced to focus on key interfering ions, generating a characteristic matrix for priority processing, thereby achieving in-depth analysis of complex multi-ion systems; through the model prediction output of interfering ion characteristics and their dynamic influence on the selective adsorption of U(VI), the electrocatalytic operating parameters are further optimized to balance the adsorption efficiency and energy consumption, and at the same time, global optimization is performed in combination with a comprehensive objective function to optimize the adsorption efficiency of uranyl ions.

[0070] The present invention introduces a real-time monitoring and feedback mechanism, which can dynamically update the prediction results of the deep learning model and adjust the operating parameters according to real-time data. It overcomes the problem that the traditional method has a decreased adsorption efficiency under the interference of high-concentration coexisting metal ions, significantly improves the selective extraction ability of U(VI), enhances the adaptability to complex water quality environments, and reduces energy consumption through dynamic optimization, thereby achieving high efficiency, economy and sustainability of wastewater treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0072] Figure 1 The figure is a schematic flow chart of the electrocatalytic treatment method for removing uranium-containing wastewater according to the present invention. DETAILED DESCRIPTION

[0073] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0074] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0075] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0076] Example 1, with reference to Figure 1 This embodiment provides an electrocatalytic treatment method for removing uranium-containing wastewater, comprising the following steps:

[0077] Step S1, deploying a monitoring sensor network to collect real-time wastewater composition data during wastewater treatment and preprocessing the collected data;

[0078] Wastewater components include uranyl ion U(VI) concentration, coexisting metal ion concentration, conductivity, and pH value;

[0079] The data preprocessing method is as follows: normalize the data, remove abnormal data, and use the dynamic time alignment algorithm DTW to correct the time series data;

[0080] The steps for correcting time series data using the dynamic time alignment algorithm DTW are as follows:

[0081] Construct a dynamic programming matrix and calculate the cumulative alignment distance between the target time series and the actual time series based on the point-to-point difference between the two series. The alignment distance calculation formula is:

[0082]

[0083] Where D(i, j) represents the cumulative alignment distance between time series X and Y at positions i and j, and d(x i ,y j ) represents the Euclidean distance between X and Y at corresponding points i and j, calculated as |x i -y j |,x i Represents the value of the time series X at the i-th time point, where i ranges from 1, 2, ..., n, y j represents the value of time series Y at the jth time point, where j ranges from 1, 2, ..., m. D(i-1, j) represents the cumulative alignment distance from the i-1th point of sequence X to the jth point of Y. D(i, j-1) represents the cumulative alignment distance from the i-th point of sequence X to the j-1th point of Y. D(i-1, j-1) represents the cumulative alignment distance from sequences X and Y to both i-1 and j-1.

[0084] By backtracking the dynamic programming matrix, the optimal alignment path between the two time series is obtained. The optimal distance calculation formula is:

[0085] P={(p1, q1), (p2, q2),…, (p k ,q k )},

[0086] Among them, P represents the index set of the optimal alignment path, p i represents the index of the i-th alignment point of the time series X in the path, q i represents the index of the i-th alignment point of time series Y in the path, k represents the total length of the alignment path, and satisfies k ≥ max(n, m), where n and m are the lengths of X and Y respectively;

[0087] According to the optimal alignment path, the collected time series data is nonlinearly resampled to generate correction data consistent with the reference time series, which is expressed as:

[0088]

[0089] Among them, X ′ represents the time series X after correction by the alignment path, Indicates the time series X corresponding to index p in the alignment path P i The value of , k represents the total length of the alignment path;

[0090] Specifically, the DTW algorithm is used through dynamic programming to calculate the optimal alignment path between time series to solve the data deviation problem caused by inconsistent time dimensions. After path backtracking, the corrected time series is obtained.

[0091] Step S2, based on the wastewater composition data pretreated in step S1, constructing a deep learning model to analyze the characteristics of the multi-ion system in the wastewater components and predict the interference effect during the U(VI) extraction process;

[0092] The characteristics of multi-ion systems include the spatial characteristics of ion concentration distribution and the time-dependent characteristics of ion dynamic changes;

[0093] Deep learning models include:

[0094] Convolutional neural network (CNN) is used to extract the spatial characteristics of ion concentration distribution in wastewater, specifically the concentration correlation between ions;

[0095] A recurrent neural network (RNN) is used to capture the time-dependent characteristics of ion dynamic changes;

[0096] Interference effect prediction outputs include: the main characteristics of interfering ions and their dynamic effects on the selective adsorption of U(VI);

[0097] The steps of using convolutional neural network (CNN) to extract the spatial characteristics of ion concentration distribution in wastewater are as follows:

[0098] The convolutional neural network (CNN) was used to process the corrected time series data, extract the concentration correlation between different ions in the multi-ion system, and analyze the spatial characteristics of the ion concentration distribution. The convolution extraction formula is:

[0099]

[0100] Where H(i,j) represents the value of the i-th row and j-th column in the feature matrix generated after the convolution operation, σ(·) represents the ReLU activation function, σ(x) = max(0,x), F(i+p,q) represents the value of the subregion starting at position (i,q) in the input data matrix F, W(p,q) represents the weight value of the p-th row and q-th column in the convolution kernel weight matrix, the size of the convolution kernel is k×m, b represents the bias of the convolution operation, k represents the row dimension of the convolution kernel, and m represents the column dimension of the convolution kernel, which is consistent with the number of ion species in the input matrix;

[0101] The steps of using recurrent neural network RNN ​​to capture the time-dependent characteristics of ion dynamic changes are as follows:

[0102] Using the recurrent neural network (RNN), the extracted spatial characteristics are analyzed in time series to capture the dynamic change characteristics of the multi-ion system and reflect the time dependence of the ion concentration change. The analysis formula is:

[0103] h t =f(W h h t-1 +W x x t +b h ),

[0104] Among them, h t represents the hidden state vector of the recurrent neural network at time step t, f(·) represents the Tanh function, and W h Represents the weight matrix from the previous time step hidden state to the current hidden state, h t-1 represents the hidden state vector at the previous time step t-1, W x Represents the weight matrix from the input vector to the hidden state at the current time step, x t Represents the input vector of the current time step, which comes from the characteristics of the F matrix at time step t, b h Bias vector representing the hidden state;

[0105] In step S2, the attention mechanism is introduced to determine the interfering ions that have the greatest impact on the selective adsorption of uranyl ions based on the dynamic characteristics of the model output, and generate a priority feature matrix. The attention weight calculation formula is:

[0106]

[0107]

[0108] Among them, α i represents the attention weight of ion i, e i represents the feature score of ion i, v represents the projection vector calculated by attention weight, W h The weight matrix representing the linear transformation of the hidden state, h t represents the hidden state vector at the current time step t, W s The linear transformation weight matrix representing ion characteristics, s i represents the eigenvector of ion i,

[0109] Weighted interference characteristic output formula:

[0110] H f =∑ i α i s i ,

[0111] Among them, H f represents the interference ion feature matrix after weighting by the attention mechanism, α i represents the attention weight of ion i, s i represents the eigenvector of ion i;

[0112] Specifically, CNN is used to extract the spatial distribution characteristics of wastewater ion concentrations, combined with RNN to capture the time-dependent characteristics, and then an attention mechanism is introduced to focus on ions with significant interference effects, thereby achieving in-depth analysis of the dynamic characteristics of multi-ion systems and prioritizing the analysis of the characteristics of key interfering ions.

[0113] Step S3, adjusting electrocatalytic operation parameters based on the interference prediction result of step S2;

[0114] The operating parameter adjustment methods include:

[0115] Adjust the electric field strength and electrode voltage;

[0116] Adjust the reaction time to balance adsorption efficiency and energy consumption;

[0117] Based on the interference prediction result of step S2, the step of adjusting the electrocatalytic operation parameters is as follows:

[0118] According to the influence of interfering ions on the adsorption efficiency of uranyl ions, the electric field strength and electrode voltage are dynamically adjusted. The target optimized electric field strength is set as E opt , the target electrode voltage after optimization is V opt , the optimization formula is:

[0119] E opt =E init +ΔE,

[0120] V opt =V init +ΔV,

[0121] Among them, E opt represents the optimized electric field strength, E init represents the initial electric field strength, ΔE represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results, V opt Represents the optimized electrode voltage, V init Represents the initial electrode voltage, ΔV represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results.

[0122] Assume the target adsorption efficiency is η t , the current adsorption efficiency is η c , the reaction time adjustment formula is:

[0123]

[0124] Among them, t opt represents the optimized reaction time, t init represents the initial reaction time, η t represents the target adsorption efficiency, η c Indicates the current adsorption efficiency,

[0125] Define the comprehensive optimization objective function J, which integrates adsorption efficiency, residual concentration and energy consumption. The objective function formula is:

[0126] J=ω1·η+ω2·(1-C e )-ω3·E c ,

[0127] Among them, J represents the comprehensive optimization target value, ω1, ω2, ω3 are weight factors, which control the importance of adsorption efficiency, residual concentration and energy consumption respectively, η represents the adsorption efficiency, C e represents the residual concentration, which is equal to the concentration of unadsorbed uranyl ions, E c Represents energy consumption, which is dynamically calculated based on electric field strength and reaction time.

[0128] By gradient descent method, E opt , V opt , t opt Perform iterative updates to maximize J;

[0129] Specifically, by predicting the results of the interference effect, the electric field strength, electrode voltage and reaction time are dynamically adjusted to ensure the optimal adsorption efficiency of uranyl ions.

[0130] Step S4, continuously monitoring the concentration changes of U(VI) and other metal ions in the wastewater under the adjusted operating parameters, inputting the monitoring data into the deep learning model of step S2, updating the interference effect prediction results in real time, and adjusting the electrocatalytic operating parameters;

[0131] In step S4, an embedded online sensor is used for real-time monitoring, and the online sensor collects dynamic concentration data of uranyl ions U(VI) and coexisting metal ions in the wastewater;

[0132] The steps of updating the interference effect prediction results in real time and adjusting the electrocatalytic operation parameters are as follows:

[0133] Using embedded online sensors, the concentration change data of uranyl ions and other metal ions in wastewater are collected in real time. The real-time monitoring data is Z t ={z t，1 , z t，2 ,…,z t，m}, represents the ion concentration vector at time t, and the monitoring data is input into the deep learning model for prediction.

[0134] The deep learning model outputs the updated interference prediction result H′(t), which is formulated as follows:

[0135] H′(t)=f model (Z t ),

[0136] Among them, H′(t) represents the interference prediction characteristic matrix after the update at time t, f model represents the deep learning model constructed in step S2, Z t represents the monitoring data vector at time t;

[0137] According to the updated interference prediction result H′(t), the operating parameters are adjusted as follows:

[0138] E opt =E init +ΔE t ,V opt =V init +ΔV t ,

[0139] Where ΔE t and ΔV t They represent the electric field intensity and electrode voltage changes adjusted according to the prediction results at time t respectively;

[0140] Specifically, embedded online sensors are used to monitor the concentration changes of uranyl ions and coexisting metal ions in wastewater in real time, and the monitoring data is dynamically input into the deep learning model to continuously update the interference effect prediction results. Based on the real-time updated prediction results, the operating parameters are dynamically adjusted to improve the adsorption efficiency and system stability during the wastewater treatment process.

[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An electrocatalytic treatment method for removing uranium-containing wastewater, characterized by: include, Step S1, deploying a monitoring sensor network to collect real-time wastewater composition data during wastewater treatment and preprocessing the collected data; Step S2, based on the wastewater composition data pretreated in step S1, constructing a deep learning model to analyze the characteristics of the multi-ion system in the wastewater components and predict the interference effect during the U(VI) extraction process; Step S3, adjusting electrocatalytic operation parameters based on the interference prediction result of step S2; Step S4, continuously monitoring the concentration changes of U(VI) and other metal ions in the wastewater under the adjusted operating parameters, inputting the monitoring data into the deep learning model of step S2, updating the interference effect prediction results in real time, and adjusting the electrocatalytic operating parameters; The characteristics of the multi-ion system include the spatial characteristics of ion concentration distribution and the time-dependent characteristics of ion dynamic changes; The deep learning model includes: Convolutional neural network (CNN) is used to extract the spatial characteristics of ion concentration distribution in wastewater, specifically the concentration correlation between ions; A recurrent neural network (RNN) is used to capture the time-dependent characteristics of ion dynamic changes; Interference effect prediction outputs include: the main characteristics of interfering ions and their dynamic effects on the selective adsorption of U(VI); The steps of using the recurrent neural network RNN ​​to capture the time-dependent characteristics of the dynamic changes of ions are as follows: Using the recurrent neural network (RNN), the extracted spatial characteristics are analyzed in time series to capture the dynamic change characteristics of the multi-ion system and reflect the time dependence of the ion concentration change. The analysis formula is: h t =f(W h h t-1 +W x x t +b h ), Among them, h t represents the hidden state vector of the recurrent neural network at time step t, f(·) represents the Tanh function, and W h Represents the weight matrix from the previous time step hidden state to the current hidden state, h t-1 represents the hidden state vector at the previous time step t-1, W x Represents the weight matrix from the input vector to the hidden state at the current time step, x t Represents the input vector of the current time step, which comes from the characteristics of the F matrix at time step t, b h Bias vector representing the hidden state; In step S2, the attention mechanism is introduced to determine the interfering ions that have the greatest impact on the selective adsorption of uranyl ions based on the dynamic characteristics of the model output, and generate a priority feature matrix. The attention weight calculation formula is: Among them, α i represents the attention weight of ion i, e i represents the feature score of ion i, v represents the projection vector calculated by attention weight, W h The weight matrix representing the linear transformation of the hidden state, h t represents the hidden state vector at the current time step t, W s The linear transformation weight matrix representing ion characteristics, s i represents the eigenvector of ion i, Weighted interference characteristic output formula: H f =∑ i a i s i , Among them, H f represents the interference ion feature matrix after weighting by the attention mechanism, α i represents the attention weight of ion i, s i represents the eigenvector of ion i; The operating parameter adjustment method includes: Adjust the electric field strength and electrode voltage; Adjust the reaction time to balance adsorption efficiency and energy consumption; In step S4, real-time monitoring is performed using an embedded online sensor, which collects dynamic concentration data of uranyl ions U(VI) and coexisting metal ions in the wastewater.

2. The electrocatalytic treatment method for removing uranium-containing wastewater according to claim 1, characterized in that: The wastewater components include uranyl ion U(VI) concentration, coexisting metal ion concentration, conductivity and pH value; The data preprocessing method is: normalizing the data, eliminating abnormal data, and correcting the time series data using the dynamic time alignment algorithm DTW.

3. The electrocatalytic treatment method for removing uranium-containing wastewater according to claim 2, characterized in that: The steps of correcting time series data using the dynamic time alignment algorithm DTW are as follows: Construct a dynamic programming matrix and calculate the cumulative alignment distance between the target time series and the actual time series based on the point-to-point difference between the two series. The alignment distance calculation formula is: Where D(i, j) represents the cumulative alignment distance between time series X and Y at positions i and j, and d(x i ,y j ) represents the Euclidean distance between X and Y at corresponding points i and j, calculated as |x i -y j |,x i Represents the value of the time series X at the i-th time point, where i ranges from 1, 2, ..., n, y j represents the value of time series Y at the jth time point, where j ranges from 1, 2, ..., m. D(i-1, j) represents the cumulative alignment distance from the i-1th point of sequence X to the jth point of Y. D(i, j-1) represents the cumulative alignment distance from the i-th point of sequence X to the j-1th point of Y. D(i-1, j-1) represents the cumulative alignment distance from sequences X and Y to both i-1 and j-1. By backtracking the dynamic programming matrix, the optimal alignment path between the two time series is obtained. The optimal distance calculation formula is: P={(p1,q1),(p2,q2),...,(p k ,q k )}, Among them, P represents the index set of the optimal alignment path, p i represents the index of the i-th alignment point of the time series X in the path, q i represents the index of the i-th alignment point of time series Y in the path, k represents the total length of the alignment path, and satisfies k ≥ max(n, m), where n and m are the lengths of X and Y respectively; According to the optimal alignment path, the collected time series data is nonlinearly resampled to generate correction data consistent with the reference time series, which is expressed as: Where X′ represents the time series X after correction by the alignment path, Indicates the time series X corresponding to index p in the alignment path P i The value of , k represents the total length of the alignment path.

4. The electrocatalytic treatment method for removing uranium-containing wastewater according to claim 3, characterized in that: The steps of using convolutional neural network (CNN) to extract the spatial characteristics of ion concentration distribution in wastewater are as follows: The convolutional neural network (CNN) was used to process the corrected time series data, extract the concentration correlation between different ions in the multi-ion system, and analyze the spatial characteristics of the ion concentration distribution. The convolution extraction formula is: Among them, H(i, j) represents the value of the i-th row and j-th column in the feature matrix generated after the convolution operation, σ(·) represents the ReLU activation function, σ(x) = max(0, x), F(i+p, q) represents the sub-region value starting from position (i, q) in the input data matrix F, W(p, q) represents the weight value of the p-th row and q-th column in the convolution kernel weight matrix, the size of the convolution kernel is k×m, b represents the bias of the convolution operation, k represents the row dimension of the convolution kernel, and m represents the column dimension of the convolution kernel, which is consistent with the number of ion species in the input matrix.

5. The electrocatalytic treatment method for removing uranium-containing wastewater according to claim 4, characterized in that: The step of adjusting the electrocatalytic operation parameters based on the interference prediction result of step S2 is as follows: According to the influence of interfering ions on the adsorption efficiency of uranyl ions, the electric field strength and electrode voltage are dynamically adjusted. The target optimized electric field strength is set as E opt , the target electrode voltage after optimization is V opt , the optimization formula is: E opt =E init +ΔE, V opt =V init +ΔV, Among them, E opt represents the optimized electric field strength, E init represents the initial electric field strength, ΔE represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results, V opt Represents the optimized electrode voltage, V init Represents the initial electrode voltage, ΔV represents the adjustment amount, which is dynamically calculated based on the interference effect prediction results. Assume the target adsorption efficiency is η t , the current adsorption efficiency is η c , the reaction time adjustment formula is: Among them, t opt represents the optimized reaction time, t init represents the initial reaction time, η t represents the target adsorption efficiency, η c Indicates the current adsorption efficiency, Define the comprehensive optimization objective function J, which integrates adsorption efficiency, residual concentration and energy consumption. The objective function formula is: J=ω1·η+ω2·(1-C e )-ω3·E c , Among them, J represents the comprehensive optimization target value, ω1, ω2, ω3 are weight factors, which control the importance of adsorption efficiency, residual concentration and energy consumption respectively, η represents the adsorption efficiency, C e represents the residual concentration, which is equal to the concentration of unadsorbed uranyl ions, E c Represents energy consumption, which is dynamically calculated based on electric field strength and reaction time. By gradient descent method, E opt ,V opt ,t opt Perform iterative updates to maximize J.

6. The electrocatalytic treatment method for removing uranium-containing wastewater according to claim 5, characterized in that: The steps of updating the interference effect prediction results in real time and adjusting the electrocatalytic operation parameters are as follows: Using embedded online sensors, the concentration change data of uranyl ions and other metal ions in wastewater are collected in real time. The real-time monitoring data is Z t ={z t，1 , z t，2 ,...,z t，m }, represents the ion concentration vector at time t, and the monitoring data is input into the deep learning model for prediction. The deep learning model outputs the updated interference prediction result H′(t), which is formulated as follows: H′(t)=f model (Z t ), Among them, H′(t) represents the interference prediction characteristic matrix after the update at time t, f model represents the deep learning model constructed in step S2, Z t represents the monitoring data vector at time t; According to the updated interference prediction result H′(t), the operating parameters are adjusted as follows: E opt =E init +ΔE t ,V opt =V init +ΔV t , Where ΔE t and ΔV t They respectively represent the electric field intensity and electrode voltage change adjusted according to the prediction results at time t.

Images