Intelligent phosphate adding method and system for low-phosphorus boiler water

Abstract

The invention relates to the technical field of intelligent control, in particular to an intelligent phosphate adding method and system for low-phosphorus boiler water. Parameters of conductivity, pH and phosphate radical in boiler water are tested and collected, and a use data sequence of the boiler is collected; according to the use data sequence, boiler corrosion is predicted, and predicted corrosion parameters are obtained; generating a conductivity correction coefficient and a phosphate radical correction coefficient according to the predicted corrosion parameters, and correcting the conductivity and the phosphate radical parameters to obtain corrected conductivity and corrected phosphate radical parameters; extracting newest use data from the use data sequence, performing boiler water characteristic prediction, obtaining predicted conductivity and predicted phosphate radical parameters, analyzing the similarity between the corrected conductivity and the corrected phosphate radical parameters, configuring parameter weights and corrosion weights, performing phosphate addition optimization in combination with the predicted corrosion parameters, obtaining optimized phosphate addition parameters, and performing phosphate addition optimization according to the optimized phosphate addition parameters. And adding phosphate. And intelligent and accurate addition of the phosphate in the low-phosphorus boiler water is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a method and system for intelligent phosphate addition in low-phosphorus boiler water. Background Art

[0002] In industrial production, boilers are crucial thermal energy equipment, and the treatment of their boiler water is of paramount importance. To prevent boiler scaling and corrosion, it is common practice to add chemicals such as phosphates to the boiler water. Traditional phosphate addition methods rely heavily on manual experience or simple automated controls, adding phosphates based on basic boiler water parameters. This makes it difficult to accurately adjust the phosphate dosage according to the actual condition of the boiler, potentially leading to insufficient addition, which fails to effectively prevent boiler corrosion, or excessive addition, resulting in chemical waste and boiler water quality deterioration. These methods suffer from technical problems such as inaccurate measurement and utilization of boiler water parameters and poor timeliness and adaptability in phosphate addition control. Summary of the Invention

[0003] This invention addresses the technical problems of insufficient accuracy in the measurement and utilization of boiler water parameters and poor timeliness and adaptability of phosphate addition in existing technologies by providing a method and system for intelligent phosphate addition in low-phosphorus boiler water.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for intelligent phosphate addition in low-phosphorus boiler water, comprising: testing and collecting conductivity, pH, and phosphate parameters in the boiler water, and collecting a boiler usage data sequence; predicting boiler corrosion based on the usage data sequence to obtain predicted corrosion parameters; generating conductivity correction coefficients and phosphate correction coefficients based on the predicted corrosion parameters, and correcting the conductivity and phosphate parameters to obtain corrected conductivity and corrected phosphate parameters; extracting the latest usage data from the usage data sequence, predicting boiler water characteristics to obtain predicted conductivity and predicted phosphate parameters, analyzing the similarity with the corrected conductivity and corrected phosphate parameters, configuring parameter weights and corrosion weights, optimizing phosphate addition based on the predicted corrosion parameters to obtain optimized phosphate addition parameters, and then adding phosphate.

[0005] Optionally, the conductivity, pH, and phosphate parameters in the boiler water are tested and collected, and the boiler usage data sequence is collected, including: testing and collecting the conductivity, pH, and phosphate parameters in the boiler water; obtaining the usage data sequence after the boiler is put into use, wherein each usage data includes boiler operation data within the usage time period.

[0006] The process of predicting boiler corrosion based on the usage data sequence and obtaining predicted corrosion parameters includes: acquiring a boiler aging predictor; inputting the usage data sequence into the boiler aging predictor; and outputting the predicted corrosion parameters.

[0007] The steps for obtaining the boiler aging predictor include: collecting a set of sample usage data sequences based on boiler operation data over a historical period, and labeling the corrosion size of the boiler under different sample usage data sequences to obtain a set of sample corrosion parameters; constructing a machine learning-based boiler aging predictor using the usage data sequences as input data and the corrosion parameters as output data; and supervising the training of the boiler aging predictor using the set of sample usage data sequences and the set of sample corrosion parameters until the accuracy converges.

[0008] Optionally, based on the predicted corrosion parameters, conductivity correction coefficients and phosphate correction coefficients are generated, and the conductivity and phosphate parameters are corrected to obtain corrected conductivity and corrected phosphate parameters. This includes: inputting the predicted corrosion parameters into a parameter correction classification table, and outputting conductivity correction coefficients and phosphate correction coefficients, wherein the parameter correction classification table includes a conductivity correction classification sub-table and a phosphate correction classification sub-table, constructed based on the mapping relationship between the sample corrosion parameter set and the sample conductivity error coefficient set and the sample phosphate error coefficient set in the boiler operation history data; inputting the predicted corrosion parameters into the parameter correction classification table, mapping and classifying to obtain conductivity correction coefficients and phosphate correction coefficients; and using the conductivity correction coefficients and phosphate correction coefficients to correct and calculate the conductivity and phosphate parameters respectively to obtain corrected conductivity and corrected phosphate parameters.

[0009] Optionally, the latest usage data is extracted from the usage data sequence to predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, and configure parameter weights and corrosion weights. This includes: extracting the latest usage data from the usage data sequence; inputting the latest usage data into a boiler water characteristic predictor and outputting predicted conductivity and predicted phosphate parameters, wherein the boiler water characteristic predictor is trained using a sample usage dataset, a sample predicted conductivity set, and a sample predicted phosphate parameter set; calculating the average similarity between the corrected conductivity and corrected phosphate parameters and the predicted conductivity and predicted phosphate parameters; using the average similarity as parameter weights and calculating corrosion weights.

[0010] Optionally, phosphate addition is optimized by combining the predicted corrosion parameters to obtain optimized phosphate addition parameters, and then phosphate addition is performed, including: randomly generating first phosphate addition parameters; calculating fused conductivity and fused phosphate parameters based on the predicted conductivity, predicted phosphate parameters, corrected conductivity, and corrected phosphate parameters; processing the pH, fused conductivity, and fused phosphate parameters to obtain a base phosphate concentration, and calculating a first added phosphate concentration based on the first phosphate addition parameters, and calculating the similarity with the standard phosphate concentration to obtain a first addition fitness; predicting corrosion protection based on the first added phosphate concentration fitness and the predicted corrosion parameters to obtain a first corrosion protection coefficient; weighting the first addition fitness and the first corrosion protection coefficient based on the parameter weights and corrosion weights to obtain a first phosphate fitness; continuing iterative optimization of the phosphate addition parameters until convergence, obtaining the optimized phosphate addition parameters with the largest phosphate fitness, and then performing phosphate addition.

[0011] The process of predicting corrosion protection based on the first added phosphate concentration fitness and the predicted corrosion parameters to obtain a first corrosion protection coefficient includes: collecting a set of sample corrosion parameters and a set of sample phosphate concentrations based on historical boiler phosphate addition data, testing and obtaining the scale ratio of the protective layer inside the boiler under different sample corrosion parameters and sample phosphate concentrations, and labeling to obtain a set of sample corrosion protection coefficients; constructing a corrosion protection predictor using machine learning, and performing supervised training based on the set of sample corrosion parameters, the set of sample phosphate concentrations, and the set of sample corrosion protection coefficients until the requirements are met; inputting the first added phosphate concentration and the predicted corrosion parameters into the corrosion protection predictor, and outputting the first corrosion protection coefficient.

[0012] Secondly, the present invention provides a smart phosphate addition system for low-phosphorus boiler water, comprising: The key data acquisition module is used to test and collect parameters such as conductivity, pH, and phosphate in the boiler water, and to collect the boiler's operating data sequence. The corrosion parameter acquisition module is used to predict boiler corrosion based on the usage data sequence and obtain the predicted corrosion parameters. The measurement parameter correction module is used to generate conductivity correction coefficients and phosphate correction coefficients based on the predicted corrosion parameters, and to correct the conductivity and phosphate parameters to obtain corrected conductivity and corrected phosphate parameters. The phosphate addition optimization module is used to extract the latest usage data within the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition in conjunction with the predicted corrosion parameters, obtain optimized phosphate addition parameters, and then add phosphate.

[0013] By implementing this invention, it is possible to test and collect parameters such as conductivity, pH, and phosphate in boiler water, and to collect boiler usage data sequences, ensuring accurate and comprehensive data support for subsequent analysis and calculations, laying a data foundation for the entire intelligent phosphate addition process, and ensuring the reliability of subsequent steps. By implementing this invention, it is possible to predict boiler corrosion based on the data sequence, obtain predicted corrosion parameters, know in advance the possible corrosion conditions of the boiler, and make subsequent parameter correction and phosphate addition more targeted, so as to better deal with corrosion problems and extend the service life of the boiler. By implementing this invention, it is possible to generate conductivity correction coefficients and phosphate correction coefficients based on predicted corrosion parameters, correct the conductivity and phosphate parameters, obtain corrected conductivity and corrected phosphate parameters, eliminate parameter measurement errors caused by corrosion and other factors, make the conductivity and phosphate parameters more accurate, and provide accurate data basis for subsequent phosphate addition optimization. By implementing this invention, it is possible to extract the latest usage data within the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition based on the predicted corrosion parameters, obtain optimized phosphate addition parameters, and add phosphate to make the amount of phosphate added more precise, which can meet the treatment requirements of boiler water while avoiding problems caused by adding too much or too little. At the same time, combined with the predicted corrosion parameters, the corrosion protection effect on the boiler is enhanced.

[0014] In summary, by implementing this invention, intelligent and precise addition of phosphate to low-phosphorus boiler water can be achieved, effectively improving the boiler's corrosion protection capabilities and ensuring stable boiler operation. Attached Figure Description

[0015] Figure 1 A schematic flowchart of a method for intelligent phosphate addition in low-phosphorus boiler water provided by the present invention; Figure 2 This is a schematic diagram of a phosphate intelligent addition system for low-phosphorus boiler water provided by the present invention.

[0016] In the attached diagram, the components represented by each number are as follows: Key data acquisition module 11, corrosion parameter prediction acquisition module 12, measurement parameter correction module 13, and phosphate addition optimization module 14. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0019] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0020] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for intelligent phosphate addition in low-phosphorus boiler water, comprising: S100: Tests and collects parameters such as conductivity, pH, and phosphate in boiler water, and collects boiler usage data sequences; S200: Based on the aforementioned data sequence, perform boiler corrosion prediction to obtain predicted corrosion parameters; S300: Generate conductivity correction coefficient and phosphate correction coefficient based on the predicted corrosion parameters, and correct the conductivity and phosphate parameters to obtain the corrected conductivity and corrected phosphate parameters; S400: Extract the latest usage data from the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition by combining the predicted corrosion parameters, obtain optimized phosphate addition parameters, and add phosphate.

[0021] In step S100 of this application embodiment, the conductivity, pH, and phosphate parameters in the boiler water are tested and collected, and the boiler usage data sequence is collected, including: The conductivity, pH, and phosphate parameters of the boiler water were tested and collected. Obtain the usage data sequence after the boiler is put into use, where each usage data includes boiler operation data within the usage time period.

[0022] In this embodiment of the application, the purpose of step S100 is to collect the conductivity, pH, and phosphate parameters in the boiler water and collect the boiler usage data sequence to determine whether the boiler water is within a reasonable operating range. This provides a historical basis for subsequent analysis of the boiler's operating patterns, aging trends, and corrosion risks, ensuring that subsequent boiler corrosion prediction, phosphate parameter correction, and phosphate addition optimization can be combined with the actual use of the boiler, rather than relying solely on the current instantaneous data.

[0023] The conductivity, pH, and phosphate parameters in the boiler water can be obtained directly by periodically testing boiler water samples using conductivity sensors, pH sensors, and an online phosphate analyzer installed in the boiler water circulation system. Specifically, the phosphate parameter is determined using spectrophotometry, which measures the degree of light absorption by phosphate ions at a specific wavelength to determine their concentration in the boiler water.

[0024] The usage data sequence after the boiler is put into use can be obtained through the data recording system during boiler operation. Each usage data sequence includes boiler operation data within the usage time period, such as operating time, load size, furnace temperature, pipeline pressure, etc.

[0025] For example, during the operation of a boiler in a factory, the conductivity of the boiler water is measured every certain time interval, such as every hour. A conductivity meter is used to measure the pH value of the boiler water, a pH meter is used to measure the pH value, and a spectrophotometric online phosphate analyzer is used to measure the phosphate concentration of the boiler water. Simultaneously, the boiler's data recording system records real-time usage data such as boiler operating time, load, furnace temperature, and pipeline pressure every hour, forming a boiler usage data sequence. This data will be used for subsequent boiler corrosion prediction and phosphate addition optimization.

[0026] In step S200 of this application embodiment, boiler corrosion prediction is performed based on the data sequence to obtain predicted corrosion parameters, including: Obtain a boiler aging predictor; The data sequence is input into the boiler aging predictor, and the predicted corrosion parameters are output.

[0027] In step S200 of this application embodiment, the step of obtaining the boiler aging predictor includes: Based on boiler operation data over a historical period, a set of sample data sequences was collected, and the corrosion size of the boiler under different sample data sequences was labeled to obtain a set of sample corrosion parameters. Using data sequences as input data and corrosion parameters as output data, a machine learning-based boiler aging predictor is constructed. The boiler aging predictor is trained under supervision using the sample data sequence set and the sample corrosion parameter set until the accuracy converges.

[0028] In this embodiment of the application, a boiler aging predictor is obtained, which provides a reliable model tool for subsequent boiler corrosion prediction based on the data sequence, ensuring the accuracy of the predicted corrosion parameters, and thus laying the foundation for subsequent parameter correction and phosphate addition optimization.

[0029] First, sample data needs to be collected to train the boiler aging predictor. This involves collecting a set of sample usage data sequences based on historical boiler operation data, and labeling the corrosion size of the boiler under different sample usage data sequences to obtain a set of sample corrosion parameters. The boiler aging predictor is then trained using the set of sample usage data sequences and the set of sample corrosion parameters.

[0030] Specifically, it is necessary to collect boiler operation data over a historical period and organize it into a sample data sequence set. Each sample data sequence contains boiler operation data for the corresponding time period, such as the parameters mentioned above, including operating time, load, furnace temperature, and pipeline pressure.

[0031] For example, if a power plant has 10 boilers of the same model with a 5-year operating history, then the monthly operating data of the 10 boilers of the same model over the past 5 years can be collected, such as the monthly cumulative running time, average load, maximum temperature, average pipeline pressure, etc., forming a set of 10 sets of sample usage data sequences for 60 months. Then, for each sample, using the data sequence, and through actual detection methods such as non-destructive testing and corrosion site measurement, the maximum corrosion depth, corrosion area, and other corrosion dimensions of the corresponding boiler are obtained and labeled as sample corrosion parameters, forming a sample corrosion parameter set. For example, through an annual shutdown inspection, the maximum corrosion depth of each boiler is recorded annually, such as 0.1 mm in the first year, 0.15 mm in the second year, etc. The corresponding monthly data are labeled as sample corrosion parameters, which can be further refined to the estimated monthly corrosion increment using interpolation.

[0032] For the task type of the boiler aging predictor, a random forest can be selected to build the boiler aging predictor.

[0033] The input features of the boiler aging predictor are various boiler operating data in the sample data sequence, including operating time, load, furnace temperature, and pipeline pressure.

[0034] The output target of the boiler aging predictor is the sample corrosion parameters, specifically the maximum corrosion depth of the boiler.

[0035] In the parameter settings of the boiler aging predictor, the number of decision trees is set to 100 to balance prediction accuracy and computational efficiency. The maximum tree depth is 15 to limit the complexity of a single tree and avoid overfitting. The minimum number of split samples is 5, meaning splitting stops when the number of samples at a node is less than 5, ensuring branch stability. The minimum number of leaf node samples is 2 to ensure that leaf nodes have enough samples to support the prediction results. The feature sampling ratio is 0.8, meaning that 80% of the input features are randomly selected for each tree to enhance the model's generalization ability.

[0036] For training the boiler aging predictor, at least 500 sets of sample data need to be selected from the sample usage data sequence set and the sample corrosion parameter set. Each set of samples contains one usage data sequence and one corresponding labeled corrosion parameter. The samples cover boiler operating scenarios with different runtimes, different load fluctuation ranges, and different operating conditions to ensure data diversity. 20% of the sample data from the 500 sets is used as a validation set. The training rounds are set to 50 rounds, with each round performing a complete training run on all 500 sets of samples, i.e., iterating through all samples to update the model parameters.

[0037] When the model's prediction accuracy on the validation set fluctuates by no more than 1% in five consecutive training rounds, and the average absolute value of the difference between the actual corrosion parameters and the predicted corrosion parameters is stable within 0.02 mm, the model is considered to have converged and obtained the boiler aging predictor.

[0038] Finally, the data sequence is input into the boiler aging predictor, and the predicted corrosion parameters are output. For example, if the operating time, load, furnace temperature, and pipeline pressure of a boiler are input into the boiler aging predictor, the predicted corrosion depth value is 0.12 mm.

[0039] In step S300 of this application embodiment, a conductivity correction coefficient and a phosphate correction coefficient are generated based on the predicted corrosion parameters, and the conductivity and phosphate parameters are corrected to obtain corrected conductivity and corrected phosphate parameters, including: The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are output. The parameter correction classification table includes a conductivity correction classification sub-table and a phosphate correction classification sub-table, which are constructed based on the mapping relationship between the sample corrosion parameter set and the sample conductivity error coefficient set and the sample phosphate error coefficient set in the boiler operation history data. The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are obtained by mapping and classification. The conductivity and phosphate parameters are corrected by using the conductivity correction coefficient and phosphate correction coefficient, respectively, to obtain the corrected conductivity and corrected phosphate parameters.

[0040] In actual boiler operation, corrosion can produce iron ions and colored complexes, affecting the accuracy of conductivity measurements and spectrophotometric phosphate determination. Therefore, the purpose of step S300 in this application embodiment is to generate a phosphate correction coefficient based on the predicted corrosion parameters, thereby eliminating measurement errors in conductivity and phosphate parameters caused by boiler corrosion, making these two key boiler water parameters more accurate, providing reliable data for subsequent phosphate addition optimization, and ensuring the targeted and effective nature of boiler water treatment and corrosion protection.

[0041] First, a parameter correction classification table needs to be constructed, which includes a conductivity correction classification sub-table and a phosphate correction classification sub-table. The construction is based on the mapping relationship between the sample corrosion parameter set and the sample conductivity error coefficient set and sample phosphate error coefficient set in the boiler operation history data.

[0042] Specifically, it is necessary to analyze the deviations between the actual conductivity and phosphate parameters and the ideal values ​​in the boiler operation history data for different corrosion levels, i.e., the error coefficients under different sample corrosion parameters, and construct a mapping relationship between the sample corrosion parameter set, sample conductivity error coefficient set, and sample phosphate error coefficient set in the boiler operation history data.

[0043] For example, under certain corrosion parameter conditions, if the actual measured value of conductivity is 2% lower than the true value when there is no corrosion interference, i.e., the error coefficient is -2%, then the corresponding conductivity correction coefficient is 1.02; if the actual measured value of phosphate is 1% higher than the true value when there is no corrosion interference, i.e., the error coefficient is 1%, then the corresponding phosphate correction coefficient is 0.99.

[0044] Furthermore, a correspondence between corrosion parameters and correction coefficients is established, forming a parameter correction classification table. For example, a corrosion depth of 0.1 mm corresponds to a conductivity correction coefficient of 1.02 and a phosphate correction coefficient of 0.98.

[0045] Then, the predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are obtained through mapping and classification. For example, if the predicted corrosion parameter is a corrosion depth of 0.15 mm, referring to the parameter correction classification table, it is found that the conductivity correction coefficient corresponding to this corrosion depth is 1.05, and the phosphate correction coefficient is 0.95. If the original conductivity collected in step S100 is 300 μS / cm, and the original phosphate parameter is 5 mg / L, then the corrected conductivity = 300 × 1.05 = 315 μS / cm; the corrected phosphate parameter = 5 × 0.95 = 4.75 mg / L. The corrected parameters here are closer to the actual state of the boiler water and can be used for subsequent phosphate addition optimization.

[0046] In step S400 of this application embodiment, the latest usage data is extracted from the usage data sequence, boiler water characteristic prediction is performed, predicted conductivity and predicted phosphate parameters are obtained, the similarity with the corrected conductivity and corrected phosphate parameters is analyzed, and parameter weights and corrosion weights are configured, including: Extract the latest usage data within the usage data sequence; The latest usage data is input into the boiler water feature predictor, and the output is the predicted conductivity and predicted phosphate parameters. The boiler water feature predictor is trained using the sample usage dataset, the sample predicted conductivity set, and the sample predicted phosphate parameter set. Calculate the average similarity between the corrected conductivity and corrected phosphate parameters and the predicted conductivity and predicted phosphate parameters; The average similarity is used as a parameter weight, and the corrosion weight is calculated.

[0047] In this embodiment of the application, the purpose of the above steps is to predict the future characteristics of the boiler water by extracting the latest usage data, compare its similarity with the corrected current conductivity and corrected phosphate parameters to determine the parameter weights, and at the same time determine the corrosion weights by combining the corrosion situation, so as to provide a quantitative weight basis for subsequent phosphate addition optimization, so that the optimization results not only fit the current actual state of the boiler water, but also fully consider the corrosion protection requirements, thereby improving the accuracy and targeting of phosphate addition.

[0048] First, the set of usage data that is closest to the current time needs to be selected from the boiler usage data sequence collected in step S100, i.e., the latest usage data, such as the operating data of the most recent hour, including operating time, load, furnace temperature, pipeline pressure, etc., as the input basis to reflect the latest operating status of the boiler.

[0049] Considering the task type of the aforementioned boiler water feature predictor, a gradient boosting regression tree can be used for construction.

[0050] The input features of the boiler water feature predictor are the latest used data, and the output targets are the sample predicted conductivity and sample predicted phosphate parameters. In the parameter settings of the boiler water feature predictor, the number of basic decision trees is 200. The learning rate is 0.05, used to control the contribution of each tree to the model and avoid overfitting. The maximum tree depth is 8, used to limit the complexity of a single tree and prevent the model from overfitting the training data. The minimum number of split samples is 10, that is, splitting stops when the number of node samples is less than 10, ensuring branch stability. The subsampling ratio is 0.9, that is, 90% of the samples are randomly selected for training each tree, enhancing the model's generalization ability.

[0051] The training data for the boiler water feature predictor consists of the sample usage dataset, the sample predicted conductivity set, and the sample predicted phosphate parameter set, with a data volume of no less than 800 sets. 20% of the training data is used as a validation set. The training run is set to 100 rounds, with each round performing a complete training run on all 800 sets of samples, i.e., updating the model parameters by iterating through all samples. The boiler water feature predictor is considered converged when, over eight consecutive training rounds, the prediction error of the boiler water feature predictor on the validation set meets the following conditions: conductivity prediction error ≤ 3 μS / cm, phosphate parameter prediction error ≤ 0.1 mg / L, and error fluctuation does not exceed 5%. The latest usage data is then input into the boiler water feature predictor, and the predicted conductivity and predicted phosphate parameters are output.

[0052] If you input the past hour's running time as 60 minutes, average load as 90%, furnace temperature as 185℃, and pipeline pressure as 1.2MPa, you will get a predicted conductivity of 320μS / cm and a predicted phosphate parameter of 4.8mg / L.

[0053] Furthermore, it is necessary to calculate the average similarity between the corrected conductivity and corrected phosphate parameters and the predicted conductivity and predicted phosphate parameters. First, the similarity between the two parameters needs to be calculated. For example, the similarity between parameters A and B can be calculated as: Similarity = [1 - |parameter A - parameter B| / max(parameter A, parameter B)] × 100%, where parameter A is the corrected value, such as the corrected conductivity, parameter B is the predicted value, such as the predicted conductivity, |parameter A - parameter B| represents the absolute difference between the two, and max(parameter A, parameter B) represents the maximum value of the two.

[0054] Assuming the corrected conductivity obtained in step S300 is 315 μS / cm and the corrected phosphate parameter is 4.75 mg / L, while the predicted conductivity is 320 μS / cm and the predicted phosphate parameter is 4.8 mg / L, then according to the above formula, the similarity between the corrected conductivity and the predicted conductivity is 98%, and the similarity between the corrected phosphate parameter and the predicted phosphate parameter is 97%. The average similarity between the two sets of parameters is (98% + 97%) / 2 = 97.5%. This average similarity is then used as the parameter weight, i.e., the parameter weight is 97.5%, or 0.975.

[0055] Furthermore, corrosion weights need to be calculated based on preset rules, such as a corrosion weight of 0.3 for corrosion depth < 0.05 mm and a corrosion weight of 1.0 for corrosion depth > 0.3 mm, and so on. In step S400 of this application embodiment, phosphate addition is optimized based on the predicted corrosion parameters to obtain optimized phosphate addition parameters, and then phosphate addition is performed, including: Randomly generate parameters for the first phosphate addition; Based on the predicted conductivity, predicted phosphate parameter, corrected conductivity, and corrected phosphate parameter, the fusion conductivity and fusion phosphate parameter are calculated. Based on the pH, fusion conductivity, and fusion phosphate parameters, the baseline phosphate concentration is obtained. Combined with the first phosphate addition parameters, the first added phosphate concentration is calculated, and the similarity to the standard phosphate concentration is calculated to obtain the first addition fitness. Based on the first phosphate concentration adaptability and the predicted corrosion parameters, corrosion protection prediction is performed to obtain the first corrosion protection coefficient. Based on the parameter weights and corrosion weights, the first addition fitness and the first corrosion protection coefficient are weighted and calculated to obtain the first phosphoric acid fitness. Continue iteratively optimizing the phosphate addition parameters until convergence, obtain the optimized phosphate addition parameters with the highest phosphate fitness, and then add phosphate.

[0056] In this embodiment of the application, the purpose of randomly generating the first phosphate addition parameter and calculating its first addition fitness is to calculate the basic phosphate concentration by randomly generating the first phosphate addition parameter and combining it with the fused boiler water parameters, and then determine the first addition fitness by comparing the similarity between the added phosphate concentration, i.e., the first added phosphate concentration and the standard phosphate concentration. This provides an initial evaluation basis for the subsequent optimization of the phosphate addition parameter, and ensures that the finally obtained optimized parameters can meet the boiler water standard requirements and adapt to the corrosion protection requirements corresponding to the predicted corrosion parameters.

[0057] First, the initial phosphate addition parameters need to be randomly generated, for example, phosphate addition amount 0.5 g / L. Next, the fusion conductivity and fusion phosphate parameters need to be calculated based on the predicted conductivity, predicted phosphate parameters, corrected conductivity, and corrected phosphate parameters.

[0058] The fused conductivity is based on parameter weights and corrosion weights, and is a weighted fusion of predicted conductivity and corrected conductivity. For example, fused conductivity = (parameter weight × corrected conductivity + corrosion weight × predicted conductivity) / (parameter weight + corrosion weight). For instance, if the parameter weight is 0.975, the corrosion weight is 0.8, the predicted conductivity is 320 μS / cm, and the corrected conductivity is 315 μS / cm, then the fused conductivity = (0.975 × 315 + 0.8 × 320) / (0.975 + 0.8) ≈ 317 μS / cm.

[0059] Similarly, the fused phosphate parameter is a weighted fusion of the predicted phosphate parameter and the corrected phosphate parameter based on parameter weights and corrosion weights, such as fused phosphate parameter = (parameter weight × corrected phosphate parameter + corrosion weight × predicted phosphate parameter) / (parameter weight + corrosion weight). For example, if the predicted phosphate parameter is 4.8 mg / L and the corrected phosphate parameter is 4.75 mg / L, then the fused phosphate parameter = (0.975 × 4.75 + 0.8 × 4.8) / (0.975 + 0.8) ≈ 4.77 mg / L.

[0060] Furthermore, based on the pH, fusion conductivity, and fusion phosphate parameters, a basic phosphate concentration needs to be obtained. Combined with the first phosphate addition parameters, a first added phosphate concentration is calculated, and the similarity to the standard phosphate concentration is calculated to obtain the first addition fitness.

[0061] The basic phosphate concentration is calculated based on the pH parameter collected in step S100, the aforementioned fused conductivity and fused phosphate parameter, and a boiler water chemical characteristic model such as the phosphate buffer solution equilibrium model. This reflects the basic sulfate concentration of the boiler water when no phosphate is added. The application of the boiler water chemical characteristic model is existing technology and will not be elaborated here. For example, based on pH=9.2, fused conductivity 317 μS / cm, and fused phosphate parameter 4.77 mg / L, the basic phosphate concentration is calculated to be 4.77 mg / L.

[0062] The first added phosphate concentration is obtained by adding the basic phosphate concentration to the first phosphate addition parameter to obtain the predicted phosphate concentration after addition. For example, in the above example, the first added phosphate concentration = 4.77 + 0.5 = 5.27 mg / L.

[0063] The first addition fitness is calculated by comparing the first added phosphate concentration with a preset standard phosphate concentration, where the preset standard phosphate concentration is an ideal concentration for safe boiler operation, such as 5.0 mg / L. The similarity calculation method can refer to the method used to calculate the similarity of parameters A and B mentioned earlier. If the similarity between the first added phosphate concentration and the preset standard phosphate concentration is calculated to be 95%, then the first addition fitness is 95%, or 0.95. A higher first addition fitness indicates that the phosphate addition parameter corresponding to that first addition fitness is closer to the ideal state.

[0064] In step S400 of this application embodiment, the method of predicting corrosion protection based on the first added phosphate concentration adaptability and the predicted corrosion parameters to obtain a first corrosion protection coefficient includes: Based on historical data on phosphate addition to boilers, a set of sample corrosion parameters and a set of sample phosphate concentrations were collected. The scale ratio of the protective layer inside the boiler under different sample corrosion parameters and sample phosphate concentrations was tested and obtained, and a set of sample corrosion protection coefficients was obtained by labeling. A corrosion protection predictor is constructed using machine learning, and supervised training is performed on the sample corrosion parameter set, sample phosphate concentration set, and sample corrosion protection coefficient set until the requirements are met. The first added phosphate concentration and predicted corrosion parameters are input into the corrosion protection predictor, and the first corrosion protection coefficient is output.

[0065] In this embodiment of the application, the above steps involve constructing and using a corrosion protection predictor, combining the adaptability of the first added phosphate concentration with predicted corrosion parameters, to quantitatively evaluate the corrosion protection effect of the boiler under the phosphate addition parameters, i.e., the first corrosion protection coefficient. This provides an evaluation basis for the subsequent optimization of phosphate addition parameters at the corrosion protection level, ensuring that the final optimized phosphate addition parameters can both meet the boiler water concentration standard and effectively improve the corrosion protection capability.

[0066] The boiler inner protective layer mentioned in this step refers to a protective film or covering layer formed on the surface of the boiler's metal inner wall. It is mainly generated by the chemical reaction between phosphates in the boiler water and the metal surface. It can directly isolate the boiler's metal inner wall from the boiler water, prevent corrosive substances in the boiler water from contacting the metal surface, reduce the occurrence of electrochemical corrosion and chemical corrosion from the source, and reduce the risk of boiler tube wall thinning and perforation.

[0067] In this embodiment, the first step is to collect a set of sample corrosion parameters and a set of sample phosphate concentrations based on historical boiler phosphoric acid addition data. Then, the scale ratio of the protective layer inside the boiler under different sample corrosion parameters and sample phosphate concentrations is tested and obtained, and a set of sample corrosion protection coefficients is labeled. For example, in historical samples, when the sample corrosion parameters are a corrosion depth of 0.1 mm and a sample phosphate concentration of 5.0 mg / L, the detected coverage area of ​​the protective layer inside the boiler is 85%, and the labeled sample corrosion protection coefficient is 0.85; when the sample corrosion depth is 0.1 mm and the sample phosphate concentration is 5.5 mg / L, the coverage area of ​​the protective layer inside the boiler is 90%, and the labeled coefficient is 0.9, and so on. The set of sample corrosion parameters, the set of sample phosphate concentrations, and the set of sample corrosion protection coefficients are obtained through the above method and used as the training sample dataset for the corrosion protection predictor. 20% of the data in the training sample dataset is then used as the validation set. The training sample dataset needs to contain no fewer than 600 sets of training sample data, and each set of training sample data must contain one sample corrosion parameter, one sample phosphate concentration, and one corresponding sample corrosion protection coefficient.

[0068] Furthermore, machine learning is needed to construct a corrosion protection predictor, which will be trained under supervision based on the set of sample corrosion parameters, the set of sample phosphate concentrations, and the set of sample corrosion protection coefficients until it meets the requirements. Depending on the task type of the corrosion protection predictor, a multilayer perceptron can be selected to build the corrosion protection predictor.

[0069] Specifically, the input features of the corrosion protection predictor are corrosion parameters and phosphate concentration; the output target is the corrosion protection coefficient.

[0070] The corrosion protection predictor uses a three-layer structure: one input layer, one hidden layer, and one output layer. The hidden layer has 32 neurons. The hidden layer uses the ReLU function, and the output layer uses the Sigmoid function. The learning rate is 0.01 to control the parameter update speed and avoid convergence oscillations. The batch size is 32.

[0071] In training the corrosion protection predictor, the number of training rounds is set to 80. In each round, all 600 sets of samples are trained once, that is, all samples are traversed to update the network parameters. When the average value of the prediction error of the model on the validation set, that is, the absolute difference between the predicted protection coefficient and the actual labeled coefficient, is ≤0.02 and the error fluctuation does not exceed 3% in 6 consecutive training rounds, the model is judged to have converged and the corrosion protection predictor is obtained.

[0072] The first added phosphate concentration and predicted corrosion parameters are input into the corrosion protection predictor, and the first corrosion protection coefficient is output.

[0073] For example, if the first added phosphate concentration is 5.27 mg / L and the predicted corrosion parameter is a corrosion depth of 0.12 mm, after inputting both into the corrosion protection predictor, the corrosion protection predictor outputs a first corrosion protection coefficient of 0.88, indicating that the boiler corrosion protection effect is good under this phosphate addition parameter, but there is still room for optimization.

[0074] Furthermore, the first addition fitness and the first corrosion protection coefficient need to be weighted and calculated based on the parameter weight and corrosion weight to obtain the first phosphoric acid fitness. Specifically, the first addition fitness and the first corrosion protection coefficient need to be weighted based on the parameter weight and corrosion weight. The calculation formula is: First phosphoric acid fitness = (Parameter weight × First addition fitness) + (Corrosion weight × First corrosion protection coefficient).

[0075] For example, if the parameter weight is 0.975, the corrosion weight is 0.8, the first addition fitness is 0.95, and the first corrosion protection coefficient is 0.88, then the first phosphoric acid fitness = (0.975 × 0.95) + (0.8 × 0.88) ≈ 0.926 + 0.704 = 1.63.

[0076] Furthermore, iterative optimization of the phosphate addition parameters needs to be continued until convergence, obtaining the optimized phosphate addition parameters with the highest phosphate fitness, and then phosphate addition can be performed.

[0077] Optionally, the iterative optimization process can be based on the first phosphate addition parameter, by increasing or decreasing the concentration of added phosphate, to generate new phosphate addition parameters, namely the second, third... nth phosphate addition parameters.

[0078] For each new phosphate addition parameter, its corresponding addition fitness, corrosion protection coefficient, and phosphate fitness are repeatedly calculated in the same way as the first phosphate addition parameter.

[0079] Then, the phosphate addition parameters are continuously adjusted iteratively until the difference in phosphate fitness between two adjacent iterations is less than a preset threshold. If the difference is ≤0.5%, convergence is determined.

[0080] The phosphate addition parameter with the highest phosphate fitness during the iteration process is selected as the optimized phosphate addition parameter, and phosphate addition is performed. For example, the optimized phosphate addition parameter can be 0.55 g / L.

[0081] Example 2, as Figure 2 As shown, based on the same inventive concept as the intelligent phosphate addition method for low-phosphorus boiler water provided in Embodiment 1, this embodiment of the invention also provides an intelligent phosphate addition system for low-phosphorus boiler water, comprising: The key data acquisition module 11 is used to test and collect parameters such as conductivity, pH, and phosphate in the boiler water, and to collect the boiler's operating data sequence. The corrosion prediction parameter acquisition module 12 is used to predict boiler corrosion based on the usage data sequence and obtain predicted corrosion parameters. The measurement parameter correction module 13 is used to generate conductivity correction coefficient and phosphate correction coefficient based on the predicted corrosion parameters, and to correct the conductivity and phosphate parameters to obtain the corrected conductivity and corrected phosphate parameters. The phosphate addition optimization module 14 is used to extract the latest usage data within the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition in combination with the predicted corrosion parameters, obtain optimized phosphate addition parameters, and add phosphate.

[0082] Furthermore, the key data acquisition module 11 includes the following execution steps: The conductivity, pH, and phosphate parameters of the boiler water were tested and collected. Obtain the usage data sequence after the boiler is put into use, where each usage data includes boiler operation data within the usage time period.

[0083] Furthermore, the corrosion parameter acquisition module 12 includes the following execution steps: Obtain a boiler aging predictor; The data sequence is input into the boiler aging predictor, and the predicted corrosion parameters are output.

[0084] The steps for obtaining the boiler aging predictor include: Based on boiler operation data over a historical period, a set of sample data sequences was collected, and the corrosion size of the boiler under different sample data sequences was labeled to obtain a set of sample corrosion parameters. Using data sequences as input data and corrosion parameters as output data, a machine learning-based boiler aging predictor is constructed. The boiler aging predictor is trained under supervision using the sample data sequence set and the sample corrosion parameter set until the accuracy converges.

[0085] Furthermore, the measurement parameter correction module 13 includes the following execution steps: The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are output. The parameter correction classification table includes a conductivity correction classification sub-table and a phosphate correction classification sub-table, which are constructed based on the mapping relationship between the sample corrosion parameter set and the sample conductivity error coefficient set and the sample phosphate error coefficient set in the boiler operation history data. The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are obtained by mapping and classification. The conductivity and phosphate parameters are corrected by using the conductivity correction coefficient and phosphate correction coefficient, respectively, to obtain the corrected conductivity and corrected phosphate parameters.

[0086] Furthermore, the phosphate addition optimization module 14 includes the following execution steps: Extract the latest usage data within the usage data sequence; The latest usage data is input into the boiler water feature predictor, and the output is the predicted conductivity and predicted phosphate parameters. The boiler water feature predictor is trained using the sample usage dataset, the sample predicted conductivity set, and the sample predicted phosphate parameter set. Calculate the average similarity between the corrected conductivity and corrected phosphate parameters and the predicted conductivity and predicted phosphate parameters; The average similarity is used as a parameter weight, and the corrosion weight is calculated.

[0087] Randomly generate parameters for the first phosphate addition; Based on the predicted conductivity, predicted phosphate parameter, corrected conductivity, and corrected phosphate parameter, the fusion conductivity and fusion phosphate parameter are calculated. Based on the pH, fusion conductivity, and fusion phosphate parameters, the baseline phosphate concentration is obtained. Combined with the first phosphate addition parameters, the first added phosphate concentration is calculated, and the similarity to the standard phosphate concentration is calculated to obtain the first addition fitness. Based on the first phosphate concentration adaptability and the predicted corrosion parameters, corrosion protection prediction is performed to obtain the first corrosion protection coefficient. Based on the parameter weights and corrosion weights, the first addition fitness and the first corrosion protection coefficient are weighted and calculated to obtain the first phosphoric acid fitness. Continue iteratively optimizing the phosphate addition parameters until convergence, obtain the optimized phosphate addition parameters with the highest phosphate fitness, and then add phosphate.

[0088] Specifically, based on the first phosphate concentration fitness combined with the predicted corrosion parameters, corrosion protection prediction is performed to obtain a first corrosion protection coefficient, including: Based on historical data on phosphate addition to boilers, a set of sample corrosion parameters and a set of sample phosphate concentrations were collected. The scale ratio of the protective layer inside the boiler under different sample corrosion parameters and sample phosphate concentrations was tested and obtained, and a set of sample corrosion protection coefficients was obtained by labeling. A corrosion protection predictor is constructed using machine learning, and supervised training is performed on the sample corrosion parameter set, sample phosphate concentration set, and sample corrosion protection coefficient set until the requirements are met. The first added phosphate concentration and predicted corrosion parameters are input into the corrosion protection predictor, and the first corrosion protection coefficient is output.

[0089] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0090] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0091] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0092] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0094] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0095] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for intelligent phosphate addition in low-phosphorus boiler water, characterized in that, The method includes: The conductivity, pH, and phosphate parameters in the boiler water were tested and collected, and the boiler's operating data sequence was also collected. Based on the aforementioned data sequence, boiler corrosion prediction is performed to obtain predicted corrosion parameters; Based on the predicted corrosion parameters, conductivity correction coefficients and phosphate correction coefficients are generated, and the conductivity and phosphate parameters are corrected to obtain the corrected conductivity and corrected phosphate parameters. Extract the latest usage data from the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition by combining the predicted corrosion parameters, obtain optimized phosphate addition parameters, and add phosphate.

2. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 1, characterized in that, The conductivity, pH, and phosphate parameters of the boiler water were tested and collected, along with boiler operating data sequences, including: The conductivity, pH, and phosphate parameters of the boiler water were tested and collected. Obtain the usage data sequence after the boiler is put into use, where each usage data includes boiler operation data within the usage time period.

3. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 1, characterized in that, Based on the aforementioned data sequence, boiler corrosion prediction is performed to obtain predicted corrosion parameters, including: Obtain a boiler aging predictor; The data sequence is input into the boiler aging predictor, and the predicted corrosion parameters are output.

4. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 3, characterized in that, The steps for obtaining the boiler aging predictor include: Based on boiler operation data over a historical period, a set of sample data sequences was collected, and the corrosion size of the boiler under different sample data sequences was labeled to obtain a set of sample corrosion parameters. Using data sequences as input data and corrosion parameters as output data, a machine learning-based boiler aging predictor is constructed. The boiler aging predictor is trained under supervision using the sample data sequence set and the sample corrosion parameter set until the accuracy converges.

5. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 1, characterized in that, Based on the predicted corrosion parameters, conductivity correction coefficients and phosphate correction coefficients are generated. The conductivity and phosphate parameters are then corrected to obtain corrected conductivity and corrected phosphate parameters, including: The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are output. The parameter correction classification table includes a conductivity correction classification sub-table and a phosphate correction classification sub-table, which are constructed based on the mapping relationship between the sample corrosion parameter set and the sample conductivity error coefficient set and the sample phosphate error coefficient set in the boiler operation history data. The predicted corrosion parameters are input into the parameter correction classification table, and the conductivity correction coefficient and phosphate correction coefficient are obtained by mapping and classification. The conductivity and phosphate parameters are corrected by using the conductivity correction coefficient and phosphate correction coefficient, respectively, to obtain the corrected conductivity and corrected phosphate parameters.

6. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 1, characterized in that, Extract the latest usage data within the usage data sequence, perform boiler water characteristic prediction, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, and configure parameter weights and corrosion weights, including: Extract the latest usage data within the usage data sequence; The latest usage data is input into the boiler water feature predictor, and the output is the predicted conductivity and predicted phosphate parameters. The boiler water feature predictor is trained using the sample usage dataset, the sample predicted conductivity set, and the sample predicted phosphate parameter set. Calculate the average similarity between the corrected conductivity and corrected phosphate parameters and the predicted conductivity and predicted phosphate parameters; The average similarity is used as a parameter weight, and the corrosion weight is calculated.

7. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 1, characterized in that, Based on the predicted corrosion parameters, phosphate addition is optimized to obtain optimized phosphate addition parameters, and phosphate addition is performed, including: Randomly generate parameters for the first phosphate addition; Based on the predicted conductivity, predicted phosphate parameter, corrected conductivity, and corrected phosphate parameter, the fusion conductivity and fusion phosphate parameter are calculated. Based on the pH, fusion conductivity, and fusion phosphate parameters, the baseline phosphate concentration is obtained. Combined with the first phosphate addition parameters, the first added phosphate concentration is calculated, and the similarity to the standard phosphate concentration is calculated to obtain the first addition fitness. Based on the first phosphate concentration adaptability and the predicted corrosion parameters, corrosion protection prediction is performed to obtain the first corrosion protection coefficient. Based on the parameter weights and corrosion weights, the first addition fitness and the first corrosion protection coefficient are weighted and calculated to obtain the first phosphoric acid fitness. Continue iteratively optimizing the phosphate addition parameters until convergence, obtain the optimized phosphate addition parameters with the highest phosphate fitness, and then add phosphate.

8. The intelligent phosphate addition method for low-phosphorus boiler water according to claim 7, characterized in that, Corrosion protection prediction is performed based on the first phosphate concentration fitness combined with the predicted corrosion parameters to obtain a first corrosion protection coefficient, including: Based on historical data on phosphate addition to boilers, a set of sample corrosion parameters and a set of sample phosphate concentrations were collected. The scale ratio of the protective layer inside the boiler under different sample corrosion parameters and sample phosphate concentrations was tested and obtained, and a set of sample corrosion protection coefficients was obtained by labeling. A corrosion protection predictor is constructed using machine learning, and supervised training is performed on the sample corrosion parameter set, sample phosphate concentration set, and sample corrosion protection coefficient set until the requirements are met. The first added phosphate concentration and predicted corrosion parameters are input into the corrosion protection predictor, and the first corrosion protection coefficient is output.

9. A smart phosphate addition system for low-phosphorus boiler water, characterized in that, The system includes: The key data acquisition module is used to test and collect parameters such as conductivity, pH, and phosphate in the boiler water, and to collect the boiler's operating data sequence. The corrosion parameter acquisition module is used to predict boiler corrosion based on the usage data sequence and obtain the predicted corrosion parameters. The measurement parameter correction module is used to generate conductivity correction coefficients and phosphate correction coefficients based on the predicted corrosion parameters, and to correct the conductivity and phosphate parameters to obtain corrected conductivity and corrected phosphate parameters. The phosphate addition optimization module is used to extract the latest usage data within the usage data sequence, predict boiler water characteristics, obtain predicted conductivity and predicted phosphate parameters, analyze the similarity with the corrected conductivity and corrected phosphate parameters, configure parameter weights and corrosion weights, optimize phosphate addition in conjunction with the predicted corrosion parameters, obtain optimized phosphate addition parameters, and then add phosphate.

Images