Preparation method of degerming algicide for circulating cooling water of power plant

Abstract

The invention relates to the technical field of preparation data processing, in particular to a power plant circulating cooling water degerming algicide preparation method, which comprises the following steps: acquiring Zeta potential regulation and control data, interfacial tension change rules and corrosion inhibitor adsorption kinetic parameters based on an orthogonal test, extracting principal component factors through data cleaning and standardization treatment, and calculating the concentration of the principal component factors; and constructing a dynamic response curved surface model of the ratio of the main and auxiliary agents and the microbial inhibition effect. In a dynamic simulation experiment, convolutional coding is fused with turbidity, a hardness space distribution matrix and a microbial population multi-dimensional vector to generate a water quality feature vector, and the association rule mapping library is combined with system operation parameters to generate a self-adaptive adding mode switching instruction. Cyclic verification data iteratively updates the model weight through a feedback correction module, an optimization parameter set of the compounding proportion, the temperature control gradient and the adding strategy is generated through cross validation, a preparation process closed-loop control system is formed, and the technical problems that multi-source data integration is low in efficiency, dynamic correlation analysis is missing, and a process quantification model is imperfect are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing preparation, and particularly to a preparation method of a bactericidal and algaecidal agent for power plant circulating cooling water. Background Art

[0002] The preparation of a bactericidal and algaecidal agent for power plant circulating cooling water needs to be designed based on the microbial inhibition mechanism and water chemical characteristics. Its core processes include active component screening, synergistic effect optimization, and stability control. First, aiming at the metabolic characteristics of bacteria, algae, and biological slime in the circulating water system, an organic compound with broad-spectrum bactericidal effects (such as isothiazolinone derivatives) is selected as the main agent, and its penetration ability into the microbial cell membrane is enhanced through molecular structure modification to interfere with the activity of the enzyme system and block energy metabolism; at the same time, an oxidizing biocide (such as peroxide) is compounded to rapidly decompose the photosynthetic pigments and cell structure of algae by using its strong oxidizing property. During the preparation process, the ratio of the main and auxiliary agents needs to be determined through orthogonal experiments, and the adsorption efficiency of the agent and suspended particles is optimized by combining the Zeta potential regulation technology. To improve environmental compatibility, a biodegradable surfactant is introduced to reduce the interfacial tension, and a passivation film is formed with the addition of corrosion inhibitor components to balance the requirements of sterilization and system anti-corrosion. In the production stage, a microreactor continuous synthesis process is adopted to achieve the efficient synthesis of active ingredients by precisely controlling the reaction temperature, pH, and mixing intensity. Finally, the chemical stability is maintained by adding stabilizers and adjusting the pH value, and its applicability under different water quality conditions is verified through dynamic simulation experiments. In practical applications, the dosing strategy needs to be adjusted according to the system concentration ratio, microbial population, and water temperature changes. Usually, a combination of continuous dosing and shock dosing is adopted to maintain a long-term antibacterial effect.

[0003] In the prior art, there are still bottleneck problems in data modeling and dynamic matching of process parameters for the multi-factor synergistic optimization of the bactericidal and algaecidal agent formula. The specific manifestations are as follows: When determining the ratio of the main and auxiliary agents through orthogonal experiments, traditional methods are difficult to effectively integrate multi-source heterogeneous data such as Zeta potential regulation data, interfacial tension change rules, and corrosion inhibitor adsorption kinetic parameters, resulting in insufficient modeling accuracy of the synergistic effect of active components; at the same time, there is a lack of a real-time correlation analysis mechanism between the multi-dimensional water quality parameters (such as hardness, conductivity, turbidity) generated by dynamic simulation experiments and the microbial population distribution data, and it is impossible to quickly construct the mapping relationship between the water quality characteristic vector and the dosing strategy of the agent. Especially in the microreactor continuous synthesis process, a quantitative model has not been established for the causal chain between the sequential process data of reaction temperature, pH, and mixing intensity and the synthesis efficiency of active ingredients, which restricts the closed-loop optimization ability of process parameters based on real-time monitoring data. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a preparation method of a bactericidal and algaecidal agent for circulating cooling water in a power plant, which solves the technical bottlenecks existing in the multi-factor collaborative optimization, such as low data integration efficiency in the preparation of the algaecide, lack of dynamic parameter correlation analysis, and imperfect quantitative models of process data and synthesis efficiency.

[0005] To solve the above technical problems, the specific technical solutions of the present invention are as follows: A preparation method of a bactericidal and algaecidal agent for circulating cooling water in a power plant provided by the present invention includes: Obtaining the Zeta potential regulation data, the change law of interfacial tension, and the corrosion inhibitor adsorption kinetic parameters in the orthogonal experiment, cleaning and standardizing the multi-source heterogeneous data through a data preprocessing module, extracting the principal component factors, generating a dynamic response surface model of the main and auxiliary agent ratio and the microbial inhibition effect, and outputting an active component compounding ratio control instruction based on the dynamic response surface model; Real-time collecting the hardness, conductivity, turbidity, and microbial population distribution data in the dynamic simulation experiment, generating a multi-dimensional water quality feature vector through convolutional coding, inputting the water quality feature vector into a pre-trained association rule mapping library, and generating an adaptive dosing mode switching instruction in combination with the system operation parameters, where the system operation parameters include the concentration multiple and the water temperature time series change data; Inputting the reaction temperature, pH value, and mixing intensity time series data in the continuous synthesis process of the microreactor into a piecewise fitting module, identifying the process window of the active ingredient synthesis efficiency, constructing a causal inference network between the dynamic response surface model and the process parameters, and outputting a temperature control gradient adjustment signal and a material mixing rate control parameter of the microreactor; In the latter stage of the active ingredient synthesis process, injecting biodegradable surfactants and corrosion inhibition components in stages according to the active component compounding ratio control instruction, regulating the adsorption direction of the agent on the surface of suspended particles based on the surface tension gradient distribution map, and simultaneously switching the reaction environment of the passivation film formation stage and the stability control stage according to the piecewise pH adjustment instruction; Inputting the verification data of the microbial inactivation rate and the biofouling inhibition effect in the circulating cooling water system simulation platform into a feedback correction module, updating the weight coefficients of the dynamic response surface model, and generating an iteratively optimized set of preparation process control parameters, where the control parameter set includes compounding ratio, temperature control gradient, and dosing strategy instructions.

[0006] Furthermore, in the preparation method of the bactericidal and algaecidal agent for circulating cooling water in a power plant of the present invention, the data cleaning and standardization process includes: Performing noise elimination processing on the Zeta potential data to generate a denoised potential distribution sequence; Filtering the environmental interference of the interfacial tension data and outputting a standardized interfacial tension characteristic value; Perform dynamic rate fitting on the corrosion inhibitor adsorption kinetic parameters to generate an adsorption rate parameter matrix; Input the potential distribution sequence, interfacial tension eigenvalue, and adsorption rate parameter matrix into the principal component analysis module to extract the key factors of the synergistic effect for constructing the dynamic response surface model.

[0007] Furthermore, in the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water of the present invention, the construction of the water quality characteristic vector includes: Based on the collected turbidity and hardness data, generate a spatial distribution matrix through a convolution kernel; Perform encoding conversion on the microbial population distribution data to generate a multi-dimensional vector; Fuse the features of the spatial distribution matrix and the multi-dimensional vector to generate a water quality characteristic vector, input it into the pre-trained association rule mapping library, and generate an adaptive dosing mode switching instruction.

[0008] Furthermore, in the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water of the present invention, the time series data analysis module includes: Perform piecewise linear fitting on the reaction temperature and pH time series data of the microreactor to extract the parameter change slopes of each process stage; Extract the frequency domain features of the mixing intensity time series data to generate a mixing uniformity evaluation index; Based on the parameter change slopes and the mixing uniformity evaluation index, identify the key process window of the active ingredient synthesis efficiency, and input the boundary parameters of the process window into the causal inference model.

[0009] Furthermore, in the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water of the present invention, the construction of the causal inference model includes: Based on the identified key process window, extract the temperature gradient, pH fluctuation range, and mixing intensity threshold as input variables; Input the input variables and the active ingredient synthesis efficiency data into a Bayesian network to establish a causal chain between process parameters and synthesis efficiency; Dynamically update the conditional probability table of the Bayesian network according to the online monitoring feedback data, generate a real-time adjustment instruction, and output it to the microreactor temperature control gradient adjustment module.

[0010] Furthermore, in the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water of the present invention, the staged injection includes: Inject a biodegradable surfactant in the passivation film formation stage, and control the fluid shear force parameter based on the generated water quality characteristic vector to adjust the adsorption layer thickness; Inject a corrosion inhibitor component in the stability control stage, and adjust the pH to the neutral range in combination with the identified key process window; Generate a surface tension gradient distribution map based on the adsorption layer thickness and pH adjustment results, and feedback it to the charge matching degree matrix generation module.

[0011] Further, in the preparation method of the bactericidal and algaecidal agent for circulating cooling water in power plants according to the present invention, the surface tension gradient regulation includes: Based on the generated surface tension gradient distribution map, utilize the charge complementary effect between the hydrophobic groups of surfactant molecules and the surface functional groups of suspended particles to generate a charge matching degree matrix; Input the charge matching degree matrix into the hydrodynamic control module of the microreactor, and dynamically adjust the fluid shear force parameters to control the adsorption layer thickness.

[0012] Further, in the preparation method of the bactericidal and algaecidal agent for circulating cooling water in power plants according to the present invention, the optimization feedback signal generation includes: Perform logarithmic transformation on the active ingredient synthesis efficiency data output by the causal inference model to generate an inactivation efficiency index; Perform binary encoding on the verification data of the biological slime inhibition effect to generate an inhibition status identifier; Input the inactivation efficiency index and the inhibition status identifier into the feedback signal generator to generate a weight coefficient update instruction for the dynamic response surface model.

[0013] Further, in the preparation method of the bactericidal and algaecidal agent for circulating cooling water in power plants according to the present invention, the iterative upgrade mechanism includes: Input the weight coefficient update instruction into the dynamic response surface model to reconstruct the main and auxiliary agent ratio constraint conditions; Update the material mixing rate control parameters of the microreactor based on the reconstructed ratio constraint conditions; Perform cross-validation on the updated control parameters and the adaptive dosing mode switching instruction to generate a final process control parameter set.

[0014] Further, the preparation method of the bactericidal and algaecidal agent for circulating cooling water in power plants according to the present invention further includes: Establish a two-way data transmission channel among the data fusion platform, dynamic simulation experiment and microreactor process parameters; Synchronously adjust the material input ratio of the microreactor based on the compounding ratio control instruction output by the dynamic response surface model; Trigger the execution of the segmented pH adjustment threshold according to the dosing strategy switching instruction mapped by the water quality characteristic vector.

[0015] Advantages of the present invention; Through a multi-dimensional data fusion and dynamic process optimization mechanism, the present invention significantly enhances the synergy effect and process stability in the preparation of bactericidal and algaecidal agents. The dynamic response surface model integrates Zeta potential regulation data, the variation law of interfacial tension, and the adsorption kinetic parameters of corrosion inhibitors, extracts the key factors of the synergy effect through principal component analysis, and optimizes the decision-making efficiency of the main and auxiliary agent ratios; the time series data analysis module combines piecewise linear fitting and frequency domain feature extraction technologies to accurately identify the process window of the synthesis efficiency of active ingredients. The causal inference model constructed by the Bayesian network dynamically correlates process parameters with synthesis efficiency to achieve closed-loop regulation of the temperature control gradient and material mixing rate of the microreactor. The adaptive dosing strategy driven by water quality characteristic vectors and the surface tension gradient regulation technology act synergistically to optimize the adsorption layer thickness and the formation effect of the passivation film through the charge matching degree matrix, and form an adaptive optimization system for the preparation process parameters in combination with the weight iteration mechanism of the feedback correction module, systematically solving the technical bottlenecks of inefficient integration of multi-source data, lack of dynamic correlation analysis, and imperfect process quantification models. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the present invention, the following will briefly introduce the drawings required in the embodiments. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the drawings.

[0017] Figure 1 It is a flowchart of a method for preparing a bactericidal and algaecidal agent for power plant circulating cooling water provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention in conjunction with the specific embodiments and corresponding drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following will detail the technical solutions provided by each embodiment of the present invention with reference to the drawings. To better understand the objectives of the present invention, the following will provide a more detailed description of the present invention.

[0019] Please refer to Figure 1 , a method for preparing a bactericidal and algaecidal agent for power plant circulating cooling water provided by the present invention, includes: Step S101, obtain the Zeta potential regulation data, the variation law of interfacial tension, and the corrosion inhibitor adsorption kinetic parameters in the orthogonal experiment. Clean and standardize the multi-source heterogeneous data through the data preprocessing module, extract the principal component factors, generate a dynamic response surface model of the main and auxiliary agent ratio and the microbial inhibition effect, and output the control instruction for the compounding ratio of the active components based on the dynamic response surface model; Step S102, in the dynamic simulation experiment, collect the data of hardness, conductivity, turbidity, and microbial population distribution in real time. Generate a multi-dimensional water quality feature vector through convolutional coding, input the water quality feature vector into the pre-trained association rule mapping library, and generate an adaptive dosing mode switching instruction in combination with the system operation parameters. The system operation parameters include the concentration ratio and the time series change data of water temperature; Step S103, input the time series data of reaction temperature, pH value, and mixing intensity in the continuous synthesis process of the microreactor into the piecewise fitting module, identify the process window of the synthesis efficiency of the active components, construct a causal inference network between the dynamic response surface model and the process parameters, and output the temperature control gradient adjustment signal and the material mixing rate control parameters of the microreactor; Step S104, in the latter stage of the active component synthesis, inject the biodegradable surfactant and the corrosion inhibition component in stages according to the control instruction of the compounding ratio of the active components, regulate the adsorption direction of the agent on the surface of the suspended particles based on the surface tension gradient distribution map, and at the same time switch the reaction environment of the passivation film formation stage and the stability control stage according to the piecewise pH adjustment instruction; Step S105, input the verification data of the microbial inactivation rate and the biological slime inhibition effect in the circulating cooling water system simulation platform into the feedback correction module, update the weight coefficients of the dynamic response surface model, and generate an iteratively optimized set of preparation process control parameters. The set of control parameters includes compounding ratio, temperature control gradient, and dosing strategy instructions.

[0020] In the orthogonal experiment stage, collect the Zeta potential regulation data, the variation law of interfacial tension, and the corrosion inhibitor adsorption kinetic parameters through the multi-parameter sensor network to form a multi-source heterogeneous data set. The data preprocessing module uses the sliding window filtering algorithm to eliminate the noise of the Zeta potential data and generate a denoised potential distribution sequence; uses the Laplace transform to filter the environmental interference of the interfacial tension data and output the standardized interfacial tension characteristic value; combines the Weibull distribution to perform dynamic rate fitting on the corrosion inhibitor adsorption kinetic parameters to generate an adsorption rate parameter matrix. The processed data set is input into the principal component analysis module, and the key factors of the synergistic effect are extracted through the dimensionality reduction algorithm to construct a dynamic response surface model between the main and auxiliary agent ratio and the microbial inhibition effect. The control instruction for the compounding ratio of the active components output by the model is directly related to the material ratio parameters of the subsequent synthesis process.

[0021] In the dynamic simulation experiment, the multi-parameter water quality sensors deployed collect data on hardness, conductivity, turbidity, and microbial population distribution in real time. The turbidity and hardness data are used to generate a spatial distribution matrix through convolution kernel operation, reflecting the spatial aggregation state of suspended particles in the water quality; the microbial population distribution data are converted into multi-dimensional vectors through one-hot encoding to characterize the distribution density of different bacterial communities. The above features are fused to generate a multi-dimensional water quality feature vector, which is input into a pre-trained association rule mapping library. Combining the data of the concentration multiple and the temporal variation of water temperature, an adaptive dosing mode switching instruction is generated through a pattern matching algorithm to achieve the dynamic switching between continuous dosing and shock dosing modes.

[0022] In the continuous synthesis process of the microreactor, the temporal data of reaction temperature, pH value, and mixing intensity are input into a piecewise fitting module. The piecewise linear fitting algorithm is used to extract the change slopes of temperature and pH parameters to identify the reaction kinetic characteristics of each process stage; the frequency domain characteristics of the mixing intensity data are extracted through Fourier transform to generate an evaluation index of mixing uniformity. Based on the correlation analysis of the parameter change slopes and the mixing uniformity evaluation index, the key process window of the synthesis efficiency of the active ingredient is determined, and the boundary parameters of the window are input into the causal inference network. The Bayesian network model takes the temperature gradient, pH fluctuation range, and mixing intensity threshold as input variables, and dynamically updates the conditional probability table in combination with the online monitoring feedback data to establish a causal chain between process parameters and synthesis efficiency, and outputs the temperature control gradient adjustment signal and the material mixing rate control parameters of the microreactor.

[0023] In the later stage of the synthesis of the active ingredient, according to the compounding ratio control instruction output by the dynamic response surface model, biodegradable surfactants and corrosion inhibitor components are injected in stages. In the passivation film formation stage, based on the water quality feature vector, the fluid shear force parameter is regulated to control the adsorption layer thickness of the surfactant on the surface of suspended particles; in the stability control stage, combined with the identification result of the key process window, the phosphate buffer system is used to adjust the pH to the neutral range. The surface tension gradient distribution map reflects the adsorption layer thickness and pH adjustment result in real time, generates a charge matching degree matrix through the charge complementary effect, and drives the microreactor fluid dynamics control module to adjust the shear force parameter to achieve precise control of the directional adsorption of the medicament.

[0024] In the circulating cooling water system simulation platform, the microbial inactivation rate data are logarithmically transformed to generate an inactivation efficiency index, and the biological slime inhibition effect is converted into an inhibition state identifier through binary encoding. The above verification data are input into the feedback correction module, and backpropagation calculation is performed with the weight coefficients of the dynamic response surface model to generate an iterative optimization instruction. The reconstructed main and auxiliary agent ratio constraint conditions update the material mixing rate control parameters of the microreactor, and cross-verify with the adaptive dosing mode switching instruction. Finally, a process control parameter set including compounding ratio, temperature control gradient, and dosing strategy is generated to form a closed-loop optimization mechanism for the preparation process.

[0025] Specifically, for the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water described in the present invention, the data cleaning and standardization processing includes: Performing noise elimination processing on the Zeta potential data to generate a denoised potential distribution sequence; Filtering out environmental interference from the interfacial tension data and outputting a standardized interfacial tension eigenvalue; Performing dynamic rate fitting on the corrosion inhibitor adsorption kinetic parameters to generate an adsorption rate parameter matrix; Inputting the potential distribution sequence, interfacial tension eigenvalue, and adsorption rate parameter matrix into the principal component analysis module to extract the key factors of the synergistic effect for constructing the dynamic response surface model.

[0026] In the Zeta potential data processing step, the raw data collected by the multi-channel potential sensor is subjected to noise elimination through a sliding window filtering algorithm. The sliding window calculates the local mean of the potential signal at a fixed time step, effectively suppressing high-frequency noise interference and generating a smooth potential distribution sequence. This sequence characterizes the interfacial charge distribution state between the chemical agent and suspended particles, providing a basic data set for subsequent synergistic effect analysis.

[0027] During the collection of interfacial tension data, the interference signals introduced by environmental temperature fluctuations and mechanical vibration factors are filtered out through wavelet transform. The Daubechies basis function is selected to perform multi-scale decomposition on the original interfacial tension curve, separating the effective tension component generated by the action of the active ingredients of the chemical agent. After normalization processing, a standardized interfacial tension eigenvalue is output, which reflects the efficiency level of the chemical agent in reducing the water-gas interfacial tension and directly affects the dispersion stability of suspended particles.

[0028] The processing of the corrosion inhibitor adsorption kinetic parameters adopts a dynamic rate fitting method. Based on the Weibull distribution model, a non-linear regression analysis is performed on the adsorption rate-time curve, and the shape parameter and scale parameter are calculated through the maximum likelihood estimation method to generate an adsorption rate parameter matrix. This matrix contains the adsorption energy distribution data of the corrosion inhibitor on the metal surface under different concentration gradients, which is used to quantitatively evaluate the formation efficiency of the passivation film.

[0029] The processed potential distribution sequence, interfacial tension eigenvalue, and adsorption rate parameter matrix are input into the principal component analysis module. The module calculates the correlation strength between each parameter through the covariance matrix and uses the eigenvalue decomposition algorithm to extract the first three principal component factors with a cumulative contribution rate exceeding 85%. These factors comprehensively characterize the synergistic effect intensity between the chemical agent components and serve as the input variables of the dynamic response surface model to guide the optimization decision of the main and auxiliary agent ratio.

[0030] Specifically, for the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water described in the present invention, the construction of the water quality characteristic vector includes: Based on the collected turbidity and hardness data, a spatial distribution matrix is generated through a convolution kernel; The microbial population distribution data is encoded and transformed to generate a multi-dimensional vector; The spatial distribution matrix and the multi-dimensional vector are subjected to feature fusion to generate a water quality feature vector, which is input into the pre-trained association rule mapping library to generate an adaptive dosing mode switching instruction.

[0031] In the turbidity and hardness data processing stage, the turbidity signals collected by multiple sensors and the hardness data obtained by an ion concentration detector are normalized and then segmented into continuous data segments by a sliding window. Each data segment undergoes a spatial convolution operation through a convolution kernel of a preset size, and the convolution kernel weights are dynamically allocated according to the spatial correlation of water quality parameters to generate a spatial distribution matrix representing the aggregation state of suspended particles. The row and column dimensions of this matrix map the sampling time series and spatial position information, reflecting the co-distribution characteristics of turbidity and hardness.

[0032] After the microbial population distribution data is identified by a classifier, it is discretely encoded according to the genus category. Each microbial population corresponds to an independent encoding channel, and its presence state and density level are marked by a binary vector to generate a multi-dimensional vector containing population diversity and abundance information. Sparse processing is introduced during the encoding process to suppress the interference of low-frequency noise data on feature expression and improve the discrimination of the encoding vector.

[0033] The spatial distribution matrix and the multi-dimensional vector are input into the feature fusion module, and a channel attention mechanism is used to perform weighted splicing on the two types of features. The module calculates the correlation matrix between the channels of the feature map and dynamically adjusts the fusion weights of the convolution channels of the spatial distribution matrix and the encoding channels of the microbial vector to generate a fused water quality feature vector. This vector contains both the spatial characteristics of physicochemical parameters and the structural characteristics of biological populations and is input into the pre-trained association rule mapping library for pattern matching.

[0034] The association rule mapping library is obtained by training with historical water quality data and contains multiple sets of mapping relationships between water quality feature patterns and dosing strategies. When the water quality feature vector is input, the mapping library uses the cosine similarity algorithm to match the closest historical feature pattern, and combines the real-time collected concentration multiple and water temperature time series data. Through a decision tree model, the sensitivity threshold of the current water quality state to the chemical dosing mode is analyzed, and finally an adaptive switching instruction for continuous dosing or shock dosing mode is generated.

[0035] Specifically, for the method for preparing a bactericide and algaecide for power plant circulating cooling water described in the present invention, the time series data analysis module includes: Perform piecewise linear fitting on the reaction temperature and pH time series data of the microreactor to extract the parameter change slopes of each process stage; Extract the frequency-domain features from the time-series data of the mixing intensity to generate an evaluation index for the mixing uniformity; Based on the parameter change slope and the mixing uniformity evaluation index, identify the key process window for the synthesis efficiency of the active ingredient, and input the boundary parameters of the process window into the causal inference model.

[0036] During the operation of the microreactor, the embedded sensor continuously collects the time-series data of the reaction temperature and pH value. The data stream is divided into equal-length time periods by the sliding window segmentation algorithm, and the data within each time window is linearly fitted by the least square method to calculate the temperature change rate and the pH drift slope of each process stage. The slope data reflects the thermodynamic stability and the acid-base balance state of the reaction system, providing a dynamic parameter basis for the subsequent process window division.

[0037] The time-series data of the mixing intensity is transformed into the frequency domain space through the fast Fourier transform, and the proportion of the low-frequency band energy is extracted as the mixing uniformity evaluation index. The concentration of the low-frequency energy characterizes the regularity of the macroscopic flow of the fluid, while the high-frequency component corresponds to the intensity of the local turbulent perturbation. By setting the low-frequency energy threshold, identify the process time periods with qualified mixing uniformity, and generate a quantitative evaluation index.

[0038] The temperature and pH change slope data and the mixing uniformity evaluation index are input into the dynamic clustering analysis module. The module uses the density-based spatial clustering algorithm to classify the continuous time periods with the absolute value of the slope less than the preset threshold and qualified mixing uniformity as the efficient synthesis window. The boundary parameters of the window include the start timestamp, the extreme value of the temperature gradient, and the pH fluctuation range, which are used as the input variables of the causal inference network to drive the subsequent process parameter optimization.

[0039] After receiving the process window boundary parameters, the causal inference network establishes the conditional dependence relationship between the temperature gradient, the pH fluctuation, and the mixing intensity through the directed acyclic graph structure. The weights of the network nodes are dynamically adjusted according to the online monitored synthesis efficiency data of the active ingredient, generating the correction instructions for the reaction temperature regulation range and the material mixing rate, and synchronously updating them to the microreactor actuator.

[0040] Specifically, for the preparation method of the circulating cooling water bactericide and algaecide in the power plant described in the present invention, the construction of the causal inference model includes: Based on the identified key process window, extract the temperature gradient, the pH fluctuation range, and the mixing intensity threshold as input variables; Input the input variables and the synthesis efficiency data of the active ingredient into the Bayesian network to establish the causal chain between the process parameters and the synthesis efficiency; Dynamically update the conditional probability table of the Bayesian network according to the online monitored feedback data, generate the real-time adjustment instructions, and output them to the microreactor temperature control gradient adjustment module.

[0041] In the critical process window identification stage, the temperature gradient parameter is obtained by calculating the difference between the temperature extreme values within adjacent time windows, reflecting the intensity of exothermic / endothermic reactions in the microreactor; the pH fluctuation range is determined by the standard deviation of the pH values within the moving average algorithm calculation window, characterizing the acid-base balance stability of the reaction system; the mixing intensity threshold is set based on the proportion of low-frequency energy extracted by frequency domain analysis, quantifying the level of fluid mixing uniformity. After normalization, the above parameters form a multi-dimensional input vector, serving as the data input source for the causal inference model.

[0042] During the construction of the Bayesian network, the network topology structure is predefined according to the physicochemical interaction relationships among process parameters. The temperature gradient node is connected as the parent node to the pH fluctuation and mixing intensity nodes, and the three jointly point to the active ingredient synthesis efficiency child node. In the network initialization stage, based on historical synthesis experiment data, the conditional probability table is trained to establish the joint probability distribution relationship between process parameter changes and synthesis efficiency fluctuations.

[0043] The online monitoring feedback data collects real-time synthesis efficiency values through a sliding window and inputs them into the Bayesian network for posterior probability calculation. The network uses the Markov chain Monte Carlo method to dynamically update the conditional probability table. When it is detected that the temperature gradient exceeds the process window boundary, the recalibration of the conditional dependence strength between nodes is triggered. The updated probability distribution generates temperature control gradient adjustment amplitude and direction instructions, which are converted into power adjustment signals for the heating jacket of the microreactor through a PID controller, realizing the closed-loop optimization of synthesis process parameters.

[0044] Specifically, for the preparation method of the circulating cooling water bactericide and algaecide in a power plant described in the present invention, the staged injection includes: Injecting a biodegradable surfactant in the passivation film formation stage, and controlling the fluid shear force parameter based on the generated water quality characteristic vector to adjust the adsorption layer thickness; Injecting a corrosion inhibitor component in the stability control stage, and adjusting the pH to the neutral range in combination with the identified critical process window; Generating a surface tension gradient distribution map according to the adsorption layer thickness and pH adjustment results, and feeding it back to the charge matching degree matrix generation module.

[0045] In the passivation film formation stage, the injection timing of the biodegradable surfactant is triggered by the turbidity and hardness correlation index in the water quality characteristic vector. The fluid shear force parameter is dynamically adjusted through the closed-loop control system of the stirring paddle speed of the microreactor. The rotation speed gradient is negatively correlated with the adsorption layer thickness. When the turbidity spatial distribution matrix shows an increase in the aggregation degree of suspended particles, the system automatically increases the stirring rate to reduce the adsorption layer thickness and maintain the uniformity of the particle surfactant coating.

[0046] During the stability control stage, the injection amount of the corrosion inhibitor component is linked to the pH fluctuation range identified in the key process window. When the process window boundary parameters output by the causal inference model indicate that the system tends to be acidic, the metering pump starts the injection program of the phosphate buffer solution, stabilizes the pH value in the range of 6.8 - 7.5 through the dissociation equilibrium of the buffer pair. At the same time, the microreactor jacket temperature control system adjusts the heat exchange rate according to the corrosion inhibitor adsorption rate parameter matrix to inhibit the tendency of local corrosion.

[0047] The surface tension gradient distribution map is generated by collecting data collaboratively with a contact angle measuring instrument and an interfacial tension sensor. The measured value of the adsorption layer thickness is converted into a thickness distribution heat map through gray-scale image analysis, and is superimposed with the ion strength data after pH adjustment to generate a three-dimensional surface tension gradient map. After this map is input into the charge matching degree matrix generation module, it drives the surfactant hydrophobic group orientation optimization algorithm to adjust the fluid shear force parameter threshold in the subsequent injection stage, forming a closed-loop feedback mechanism for adsorption regulation.

[0048] Specifically, in the preparation method of the circulating cooling water bactericide and algaecide for power plants described in the present invention, in the step of staged injection, the surface tension gradient regulation includes: Based on the generated surface tension gradient distribution map, a charge matching degree matrix is generated by using the charge complementary effect between the hydrophobic groups of surfactant molecules and the surface functional groups of suspended particles; The charge matching degree matrix is input into the fluid dynamics control module of the microreactor to dynamically adjust the fluid shear force parameter to control the adsorption layer thickness.

[0049] In the stage of generating the surface tension gradient distribution map, the contact angle measuring instrument collects the wettability data of the suspended particle surface in real time, combines with the change value of the liquid-solid interfacial tension monitored by the interfacial tension sensor, and generates a heat map of the particle surface wetness distribution through gray-scale image analysis. This heat map is subjected to spatial superposition operation with the ion strength data after synchronous pH adjustment to form a three-dimensional surface tension gradient distribution map, characterizing the intensity difference of the drug adsorption effect in different regions.

[0050] During the generation of the charge matching degree matrix, the electron cloud density distribution data of the hydrophobic groups of surfactant molecules are obtained by simulation with quantum chemistry calculation software, and the charge characteristics of the surface functional groups of suspended particles are analyzed by actual measurement with a Zeta potentiometer. The two sets of data are input into the matrix generation algorithm to calculate the electrostatic potential complementary index between the hydrophobic group and the functional group, generating a charge matching degree matrix reflecting the molecular-level interaction intensity. The row and column dimensions of the matrix correspond to the surface coordinate positions of the particles and the types of surfactant molecules respectively, and the numerical value indicates the preference degree of the adsorption effect at a specific position.

[0051] After receiving the charge matching degree matrix, the hydrodynamic control module analyzes the spatial adsorption priority distribution in the matrix through a convolutional neural network. Based on the coordinate clustering results of the high matching degree regions in the matrix, the module dynamically adjusts the radial flow velocity distribution pattern of the stirring paddle in the microreactor. When it is detected that the thickness of the adsorption layer in a local area exceeds the set threshold, the rotation speed of the corresponding azimuth blade is increased to enhance the shear force, forcing the overly thick adsorption layer to peel off and maintaining the uniformity of the particle surface coating.

[0052] Specifically, for the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water described in the present invention, the generation of the optimized feedback signal includes: Performing a logarithmic transformation on the active ingredient synthesis efficiency data output by the causal inference model to generate an inactivation efficiency index; Performing binary encoding on the biofouling inhibition effect verification data to generate an inhibition status identifier; Inputting the inactivation efficiency index and the inhibition status identifier into a feedback signal generator to generate a weight coefficient update instruction for the dynamic response surface model.

[0053] In the active ingredient synthesis efficiency data processing stage, the sequence of efficiency values output by the causal inference model is subjected to dimension normalization through natural logarithmic transformation. The transformed inactivation efficiency index eliminates the absolute numerical differences between different batches of synthesis experiments and retains the relative efficiency fluctuation characteristics, providing a standardized input benchmark for subsequent weight coefficient updates.

[0054] The biofouling inhibition effect verification data is obtained through optical microscope image analysis, and the slime coverage rate is converted into a binary identifier through a threshold segmentation algorithm. When the slime coverage area exceeds the preset threshold, the identifier is set to 1, triggering an alarm for an abnormal inhibition state; when it is lower than the threshold, it is set to 0, indicating that the inhibition effect meets the standard. This identifier and the inactivation efficiency index form a multi-dimensional feedback signal input source.

[0055] After receiving the above data, the feedback signal generator calculates the weight correction amount of the dynamic response surface model using a weighted fusion algorithm. The inactivation efficiency index and the inhibition status identifier are respectively assigned adjustable weight coefficients, and the partial derivatives of the model output error with respect to the weight parameters are calculated through the backpropagation algorithm to generate an update instruction including the weight adjustment direction and amplitude. After verification, the instruction is written into the parameter storage area of the dynamic response surface model to complete the closed-loop optimization iteration of the preparation process.

[0056] Specifically, for the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water described in the present invention, the iterative upgrade mechanism includes: Inputting the weight coefficient update instruction into the dynamic response surface model to reconstruct the main and auxiliary agent ratio constraint conditions; Updating the material mixing rate control parameters of the microreactor based on the reconstructed ratio constraint conditions; Cross-verify the updated control parameters with the adaptive dosing mode switching instruction to generate the final process control parameter set.

[0057] In the weight coefficient update instruction processing stage, after receiving the update instruction, the constraint condition reconstruction module of the dynamic response surface model maps the weight coefficient to the main and auxiliary agent ratio parameter space through the parameter normalization algorithm. The constraint boundary adjustment algorithm dynamically shrinks or expands the allowable concentration range of each active component according to the weight change amplitude, generates a new ratio constraint set adapted to the current water quality characteristics, and synchronously writes it into the formula database of the microreactor.

[0058] During the update process of the material mixing rate control parameter, the ratio constraint condition parsing engine extracts the reconstructed concentration ratio relationship between the main agent and the auxiliary agent, combines the volume of the microreactor and the material viscosity characteristics, and calculates the theoretical flow rate range of each feed port through the fuzzy logic algorithm. The control module adjusts the pulse frequency and valve opening parameters of the peristaltic pump according to the theoretical values to achieve precise proportional mixing of materials with different viscosities, and at the same time triggers the temperature compensation program in the mixing chamber to maintain the stability of the reaction system.

[0059] The updated control parameters and the adaptive dosing mode switching instruction are input into the cross-validation module, and the historical process data set is used for compatibility verification. The module compares the change trend of the control parameters with the time alignment degree of the dosing mode switching node through the time series matching algorithm. When it detects a conflict between the parameter mutation and the mode switching, it starts the logic arbitration mechanism to reallocate the execution priority, and finally generates a process control parameter set containing verification marks and writes it into the central controller.

[0060] Specifically, the preparation method of the bactericidal and algaecidal agent for power plant circulating cooling water described in the present invention, after the causal inference network between the dynamic response surface model and the process parameters is constructed, further includes: Establish a two-way data transmission channel between the data fusion platform, the dynamic simulation experiment and the process parameters of the microreactor; Based on the compounding ratio control instruction output by the dynamic response surface model, synchronously adjust the material input ratio of the microreactor; Trigger the execution of the segmented pH adjustment threshold according to the dosing strategy switching instruction mapped by the water quality characteristic vector.

[0061] In the data fusion platform architecture, the OPC UA protocol constructs a real-time communication link between the process parameters of the microreactor and the dynamic simulation experiment module. The platform is built with a data format conversion engine to align and encode the time stamp sequence of the process parameters and the spatial coordinates of the experimental data, and realizes millisecond-level data synchronization through the edge computing node. This channel supports the reverse writing function of the process parameters. When the dynamic simulation experiment module identifies abnormal fluctuations, it can reverse-correct the actuator control signal of the microreactor.

[0062] After receiving the active ingredient concentration ratio data output by the dynamic response surface model, the compounding ratio control instruction analysis module calls the viscosity-temperature correlation table in the material property database. The module automatically compensates for the influence of viscosity changes on the mixing uniformity according to the current reaction system temperature, generates the adapted material input flow rate parameters, and drives the synchronous adjustment of the stepping motor speed and the proportional valve opening of the peristaltic pump at the feed end of the microreactor to maintain the precise execution of the preset compounding ratio.

[0063] The dosing strategy switching instruction mapped by the water quality characteristic vector triggers the segmented pH adjustment actuator. When the microbial population density index in the vector exceeds the threshold, the switching instruction activates the alkaline adjustment program in the passive film formation stage, and the real-time data of the pH sensor is input into the fuzzy PID controller to regulate the pulse injection frequency of the sodium hydroxide metering pump. In the stability control stage, the injection amount of the phosphate buffer solution is linked with the mixing intensity parameter, and the injection rate is dynamically adjusted through the buffer capacity calculation module to maintain the chemical stability in the target pH range.

[0064] In the orthogonal test stage, the multi-parameter sensor network collects the Zeta potential regulation data, the law of interfacial tension change, and the corrosion inhibitor adsorption kinetic parameters. The noise of the potential data is eliminated by the sliding window filtering algorithm to generate the denoised potential distribution sequence. The interfacial tension data is filtered by wavelet transform to remove environmental interference, and the standardized interfacial tension characteristic value is output; the corrosion inhibitor adsorption kinetic parameters are fitted based on the Weibull distribution to generate the adsorption rate parameter matrix. The processed data is input into the principal component analysis module to extract the first three principal component factors with a cumulative contribution rate exceeding 85%, and a dynamic response surface model of the main and auxiliary agent ratio and the microbial inhibition effect is constructed. The active ingredient compounding ratio control instruction output by the model is directly related to the microreactor material ratio parameter.

[0065] In the dynamic simulation experiment, the turbidity and hardness data generate a spatial distribution matrix through the convolution kernel, and the microbial population distribution data is converted into a multi-dimensional vector through the One-hot encoding. After fusion, a water quality characteristic vector is formed. The pre-trained association rule mapping library combines the data of the concentration multiple and the water temperature time series change, matches the historical characteristic pattern through the cosine similarity algorithm, and generates an adaptive switching instruction for continuous dosing or shock dosing mode. The time series data of the microreactor process parameters are extracted by piecewise linear fitting to obtain the temperature and pH change slopes. Combined with the frequency domain characteristic analysis of the mixing intensity, the key process window of the active ingredient synthesis efficiency is identified. The window boundary parameters are input into the Bayesian network to construct a causal inference model, dynamically associate the quantitative relationship between the temperature gradient, the pH fluctuation range and the synthesis efficiency, and output the temperature control gradient adjustment signal and the material mixing rate control parameter.

[0066] In the later stage of the synthesis of the active ingredient, in the passivation film formation stage, the fluid shear force parameters are regulated according to the water quality characteristic vector, and the thickness of the adsorption layer is adjusted through the closed-loop control system of the stirring paddle speed; in the stability control stage, combined with the recognition result of the key process window, the phosphate buffer system is used to stabilize the pH in the range of 6.8 - 7.5. The surface tension gradient distribution map is generated based on the data of the adsorption layer thickness and ionic strength, and the charge matching degree matrix drives the hydrodynamic module of the microreactor to adjust the shear force parameters. In the cyclic verification data, the microbial inactivation rate is logarithmically transformed to generate the inactivation efficiency index, and the biofouling inhibition effect is converted into an inhibition status identifier through binary encoding. The feedback signal generator fuses the two types of data to generate a weight coefficient update instruction, reconstructs the ratio constraint conditions of the dynamic response surface model, and generates the final process control parameter set after cross-verifying with the adaptive dosing mode switching instruction. The data fusion platform establishes a two-way data transmission channel based on the OPC UA protocol, synchronizes the process parameters of the microreactor and the dynamic simulation experiment data in real time, realizes the linkage execution of the compounding ratio control instruction and the segmented pH adjustment threshold, and forms a closed-loop optimization system.

[0067] The explanations of the technical features in the technical solution of the present invention are as follows: Zeta potential regulation data: It refers to the experimental data obtained by measuring the surface charge characteristics of suspended particles by electrophoresis, which is used to characterize the adsorption efficiency between the reagent and the particles. The potential distribution sequence generated after eliminating high-frequency noise through sliding window filtering reflects the degree of neutralization of the surface charge of particles by the active ingredient of the reagent, providing a quantitative basis for optimizing the adsorption kinetics.

[0068] Interface tension change law: It describes the dynamic data of the ability of the reagent to reduce the water-air or water-solid interface tension. The standardized eigenvalue after filtering out environmental interference by wavelet transform is used to evaluate the improvement effect of the surfactant on the dispersion stability of suspended particles, and directly affects the spreading performance of the reagent on the particle surface.

[0069] Corrosion inhibitor adsorption kinetic parameters: The parameter matrix generated by fitting the adsorption rate-time curve through the Weibull distribution, including key indicators such as adsorption energy barrier and diffusion rate. This matrix quantifies the adsorption process of the corrosion inhibitor on the metal surface and is used to predict the formation rate of the passivation film and the anti-corrosion performance.

[0070] Dynamic response surface model: A multi-variable optimization model constructed based on principal component analysis, which maps the Zeta potential, interface tension and adsorption kinetic data to the main and auxiliary agent ratio space. By extracting the key factors of the synergistic effect, the model establishes a quantitative relationship between the ratio parameters and the microbial inhibition effect, guiding the optimization of the compounding ratio of the active components.

[0071] Convolutional coding generates multi-dimensional water quality feature vectors: convolution kernel operations are performed on the spatial distribution characteristics of turbidity and hardness data to generate a spatial matrix that represents the aggregation state of suspended particles; combined with the one-hot encoding vector of the microbial population, it is fused into a comprehensive water quality feature vector through the channel attention mechanism. This vector integrates physical, chemical and biological characteristics to provide an input benchmark for dosing strategy matching.

[0072] Bayesian network causal reasoning: With temperature gradient, pH fluctuation range and mixing intensity threshold as input variables, the causal chain between process parameters and synthesis efficiency is established through conditional probability tables. The network dynamic update mechanism adjusts the node weights according to online monitoring data to achieve closed-loop regulation of the microreactor temperature control gradient and material mixing rate.

[0073] Segmented pH adjustment technology: sodium hydroxide gradient adjustment (pH 8.5-9.2) is used in the passivation film formation stage to promote the directional adsorption of surfactants; in the stability control stage, it is switched to the phosphate buffer system (pH 6.8-7.5), and the injection rate is dynamically adjusted through the buffer capacity calculation module to maintain the chemical stability of the system.

[0074] Charge matching matrix: A two-dimensional matrix generated based on the electrostatic potential complementarity between the hydrophobic groups of the surfactant and the functional groups of the particles. The matrix values ​​reflect the strength of the adsorption at the molecular level, driving the fluid dynamics control module to adjust the radial velocity distribution of the stirring paddle and optimize the uniformity of the adsorption layer.

[0075] Feedback correction module: The efficiency index generated by the logarithmic transformation of the microbial inactivation rate is integrated with the binary identifier of the bioslime inhibition effect, and the weight correction amount of the dynamic response surface model is calculated through the back propagation algorithm to achieve iterative optimization of the preparation process parameters.

[0076] Bidirectional data transmission channel: A data synchronization link built based on the OPC UA protocol enables millisecond-level interaction between microreactor process parameters, dynamic simulation experiment data and the data fusion platform. The channel supports the reverse writing function of process parameters to ensure real-time correction of control instructions under abnormal conditions.

[0077] The above features form a closed-loop control system through multi-dimensional data fusion: the response surface model constructed by orthogonal test data guides the compounding of main and auxiliary agents; the association rule library driven by water quality feature vector generates the dosing strategy; the microreactor time series data analysis optimizes the synthesis process window; the surface tension gradient regulation and segmented pH adjustment ensure the stability of the agent; the feedback mechanism updates the model parameters through verification data iteration. Each module realizes dynamic collaboration through a two-way data channel, systematically solving the technical defects of inefficient data integration and imperfect process quantification model.

[0078] The present invention systematically solves the problems of data integration and dynamic analysis in multi-factor collaborative optimization by constructing a dynamic response surface model and a causal inference network. First, in the orthogonal experiment stage, data such as Zeta potential regulation data, interfacial tension change laws, and corrosion inhibitor adsorption kinetic parameters are integrated. The data preprocessing module uses sliding window filtering and Laplace transform to eliminate environmental noise, generating a denoised potential distribution sequence and a standardized interfacial tension eigenvalue. The corrosion inhibitor adsorption kinetic parameters are fitted by Weibull distribution to generate an adsorption rate parameter matrix. The three are used to extract the key factors of the synergistic effect through principal component analysis, construct a dynamic response surface model of the main and auxiliary agent ratio and the microbial inhibition effect, and output the control instruction for the compounding ratio of the active components, realizing the efficient integration of multi-source heterogeneous data.

[0079] In the dynamic simulation experiment stage, the turbidity and hardness data are converted into a spatial distribution matrix through a convolution kernel, and the microbial population distribution data are generated into a multi-dimensional vector through One-hot encoding. After feature fusion, a water quality feature vector is formed. The pre-trained association rule mapping library combines the data of the concentration multiple and the time series change of water temperature, and generates an adaptive dosing mode switching instruction through a pattern matching algorithm. The time series data of the microreactor process parameters are identified by a segmented fitting module to obtain the process window of the active ingredient synthesis efficiency. The Bayesian network establishes a causal chain based on the temperature gradient, pH fluctuation range, and mixing intensity threshold, dynamically updates the conditional probability table, and generates a temperature control gradient adjustment signal and a material mixing rate control parameter, forming a quantitative association between the process parameters and the synthesis efficiency.

[0080] In the feedback optimization link, the microbial inactivation rate data are logarithmically transformed to generate an inactivation efficiency index, and the biological slime inhibition effect is converted into an inhibition state identifier through binary encoding. The feedback signal generator fuses the two types of data to generate a weight coefficient update instruction, and reconstructs the ratio constraint conditions of the dynamic response surface model. The reconstructed parameters and the adaptive dosing mode switching instruction are cross-validated to generate a final control parameter set. At the same time, the input ratio of the microreactor material and the segmented pH adjustment threshold are synchronously adjusted through a two-way data transmission channel, forming a closed-loop optimization mechanism, and systematically improving the synergistic effect and process stability of algaecide preparation.

Claims

1. A method for preparing a sterilizing and algaecide agent for circulating cooling water in a power plant, characterized in that: include: Obtain the Zeta potential control data, interfacial tension change law and corrosion inhibitor adsorption kinetic parameters in the orthogonal experiment, clean and standardize the multi-source heterogeneous data through the data preprocessing module, extract the main component factors, generate a dynamic response surface model of the main and auxiliary agent ratio and the microbial inhibition effect, and output the active component compounding ratio control instruction based on the dynamic response surface model; In the dynamic simulation experiment, hardness, conductivity, turbidity and microbial population distribution data are collected in real time, and a multi-dimensional water quality feature vector is generated through convolution coding. The water quality feature vector is input into a pre-trained association rule mapping library, and an adaptive dosing mode switching instruction is generated in combination with system operation parameters. The system operation parameters include concentration multiples and water temperature time series change data; Inputting the reaction temperature, pH value and mixing intensity time series data in the continuous synthesis process of the microreactor into the segmented fitting module, identifying the process window of the synthesis efficiency of the active ingredient, constructing a causal reasoning network between the dynamic response surface model and the process parameters, and outputting the temperature control gradient adjustment signal of the microreactor and the material mixing rate control parameter; In the latter stage of active ingredient synthesis, biodegradable surfactants and corrosion inhibition components are injected in stages according to the active component compounding ratio control instructions, the adsorption direction of the agent on the surface of the suspended particles is regulated based on the surface tension gradient distribution map, and the reaction environment of the passivation film formation stage and the stability control stage is switched according to the segmented pH adjustment instructions; The verification data of the microbial inactivation rate and the biological slime inhibition effect in the circulating cooling water system simulation platform are input into the feedback correction module, the weight coefficient of the dynamic response surface model is updated, and the preparation process control parameter set after iterative optimization is generated, and the control parameter set includes the compounding ratio, the temperature control gradient and the dosing strategy instructions.

2. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 1, characterized in that: The data cleaning and standardization processing includes: Perform noise elimination processing on Zeta potential data to generate a denoised potential distribution sequence; Filter out environmental interference from the interfacial tension data and output standardized interfacial tension characteristic values; Perform dynamic rate fitting on the adsorption kinetic parameters of the corrosion inhibitor to generate an adsorption rate parameter matrix; The potential distribution sequence, interfacial tension characteristic values ​​and adsorption rate parameter matrix are input into a principal component analysis module to extract key factors of synergistic effect for constructing the dynamic response surface model.

3. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 2, characterized in that: The water quality feature vector construction includes: Based on the collected turbidity and hardness data, a spatial distribution matrix is ​​generated through a convolution kernel; Encode and convert the microbial population distribution data to generate a multidimensional vector; The spatial distribution matrix and the multidimensional vector are subjected to feature fusion to generate a water quality feature vector, which is input into the pre-trained association rule mapping library to generate an adaptive dosing mode switching instruction.

4. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 1, characterized in that: The time series data analysis module includes: The reaction temperature and pH time series data of the microreactor were fitted with piecewise linear fitting to extract the slope of parameter changes in each process stage. Extract frequency domain features from the mixing intensity time series data to generate a mixing uniformity evaluation index; Based on the parameter change slope and the mixing uniformity evaluation index, the critical process window for the synthesis efficiency of the active ingredient is identified, and the process window boundary parameters are input into the causal reasoning model.

5. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 4, characterized in that: The construction of the causal inference model includes: Based on the identified critical process windows, the temperature gradient, pH fluctuation range, and mixing intensity threshold are extracted as input variables; Inputting the input variables and active ingredient synthesis efficiency data into a Bayesian network to establish a causal chain between process parameters and synthesis efficiency; The conditional probability table of the Bayesian network is dynamically updated according to the online monitoring feedback data, and real-time adjustment instructions are generated and output to the microreactor temperature control gradient adjustment module.

6. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 1, characterized in that: In the step of injecting in stages, the step of injecting in stages comprises: In the passive film formation stage, a biodegradable surfactant is injected, and the fluid shear force parameter is controlled based on the generated water quality characteristic vector to adjust the adsorption layer thickness; Inject corrosion inhibition components during the stability control stage and adjust the pH to the neutral range in combination with the identified critical process windows; A surface tension gradient distribution map is generated according to the adsorption layer thickness and the pH adjustment result, and fed back to the charge matching matrix generation module.

7. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 6, characterized in that: The surface tension gradient regulation includes: Based on the generated surface tension gradient distribution map, the charge matching matrix is ​​generated by utilizing the charge complementarity effect between the hydrophobic groups of the surfactant molecules and the functional groups on the surface of the suspended particles. The charge matching matrix is ​​input into the fluid dynamics control module of the microreactor, and the fluid shear force parameters are dynamically adjusted to control the thickness of the adsorption layer.

8. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 1, characterized in that: In the feedback correction module, the optimization feedback signal generation includes: The active ingredient synthesis efficiency data output by the causal inference model were logarithmically transformed to generate an inactivation efficiency index; Binary encoding of the bioslime inhibition effect verification data is performed to generate an inhibition status identifier; The inactivation efficiency index and the inhibition state identifier are input into a feedback signal generator to generate a weight coefficient update instruction of a dynamic response surface model.

9. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 8, characterized in that: The iterative upgrade mechanism includes: Inputting the weight coefficient update instruction into the dynamic response surface model to reconstruct the main and auxiliary agent ratio constraint conditions; Update the material mixing rate control parameters of the microreactor based on the reconstructed ratio constraint conditions; The updated control parameters are cross-validated with the adaptive dosing mode switching instructions to generate the final process control parameter set.

10. The method for preparing a sterilizing and algaecide agent for circulating cooling water of a power plant according to claim 1, characterized in that: After the causal inference network is constructed, it also includes: Establish a two-way data transmission channel between the data fusion platform, dynamic simulation experiment and microreactor process parameters; Based on the compounding ratio control instructions output by the dynamic response surface model, the material input ratio of the microreactor is adjusted synchronously; According to the dosing strategy switching instruction mapped by the water quality feature vector, the execution of the segmented pH adjustment threshold is triggered.

Images