A boiler four-tube corrosion risk intelligent evaluation system based on dissolved hydrogen monitoring and an evaluation method thereof
By combining a baseline prediction model and a graph attention mechanism with a spatial topology graph model, the problem of background hydrogen interference in the corrosion monitoring of four boiler tubes was solved, enabling accurate location and dynamic early warning of corrosion risks, and reducing the risk of tube rupture and maintenance costs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUODIAN ZHENENG NINGDONG POWER GENERATION CO LTD
- Filing Date
- 2026-03-25
- Publication Date
- 2026-07-24
Smart Images
Figure CN122448722A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial boiler safety monitoring technology, and in particular to an intelligent assessment system and assessment method for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring. Background Technology
[0002] The four tubes (economizer, water wall, superheater, and reheater) of boilers in thermal power plants operate under high temperature, high pressure, and complex combustion conditions for extended periods, making them highly susceptible to metal corrosion, primarily due to the "iron-water oxidation reaction." Severe tube wall corrosion can lead to tube thinning and embrittlement, ultimately causing tube rupture and shutdown accidents, seriously threatening the safe and economical operation of the power plant. Currently, industry monitoring of corrosion in boiler tubes mainly relies on random inspections during shutdowns or on testing macroscopic water and steam physicochemical indicators. These traditional methods suffer from significant time lags, and indiscriminate sampling is extremely costly.
[0003] In recent years, online dissolved hydrogen monitoring technology has begun to be applied to reflect the metal corrosion situation in steam-water systems (hydrogen is a byproduct of the oxidation reaction of molten iron). However, existing methods for dissolved hydrogen monitoring and corrosion assessment have the following significant drawbacks: First, not all dissolved hydrogen in steam-water systems comes from pipe wall corrosion. Normal chemical dosing (such as hydrazine and other chemical deoxidizers) generates a large amount of background hydrogen at high temperatures through thermal decomposition. Existing technologies struggle to remove this interference noise, easily leading to false alarms or missed alarms. Second, existing dissolved hydrogen measuring points are usually located at the outlet of the main steam pipe, and the measured data is a macroscopic average after mixing and dilution of the entire boiler tube system. Faced with hundreds or thousands of physical boiler tubes, existing technologies completely lack spatial positioning capabilities, making it impossible to trace back to the source and pinpoint which specific pipe section has a high corrosion risk. Consequently, the monitoring data cannot be effectively transformed into precise maintenance or operational guidance strategies. Summary of the Invention
[0004] This invention provides an intelligent assessment system and method for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring. By removing background hydrogen interference to extract the true hydrogen production, and by utilizing graph attention mechanism and spatial topology graph model, it successfully achieves accurate spatial mapping of the macroscopic corrosion rate of boiler four tubes to microscopic local tube sections and dynamic early warning of wear and explosion prevention.
[0005] This invention provides an intelligent assessment method for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring, including:
[0006] S1. Obtain historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system.
[0007] S2. Based on the thermal parameters and dosage characteristics, a pre-constructed baseline prediction model is used to predict the theoretical background hydrogen concentration under the current operating conditions. The total hydrogen production concentration of the unit is calculated based on the real-time dissolved hydrogen concentration. The difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration is calculated to obtain the actual corrosion hydrogen production concentration.
[0008] S3. Input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and combine it with the steam-water circulation flow rate and steam volume data to calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment.
[0009] S4. The four tubes of the boiler are discretized into multiple physical tube segments according to their physical structure, and a spatial topology model representing the connection relationship between the steam and water flow between the physical tube segments and including multiple tube segment nodes is constructed. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments;
[0010] S5. Calculate the spatial allocation weight based on the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight. Obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold to generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.
[0011] Furthermore, S1 specifically includes:
[0012] S101. Obtain the historical operating parameters within a set time period through the unit data acquisition network, and perform data cleaning and timestamp alignment on the historical operating parameters; wherein, the thermal parameters and chemical dosage characteristics are obtained through the unit control system, and the local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics are obtained through a sensor array arranged inside the boiler and on the surface of the four tubes.
[0013] S102. Extract the unit structure data from the unit's three-dimensional digital model or design drawings and manufacturing ledger; wherein, the steam-water circulation flow rate includes the total feedwater flow rate and the total main steam flow rate under the corresponding load command, and the steam volume data includes the fluid channel volume inside the economizer, water-cooled wall, superheater and reheater at each stage of the heating surface;
[0014] S103. The real-time dissolved hydrogen concentrations of the boiler unit feedwater system and main steam system are obtained by online dissolved hydrogen monitoring instruments installed at the boiler steam-water circulation inlet and outlet nodes, respectively; wherein, the real-time dissolved hydrogen concentration of the feedwater system represents the initial background hydrogen content before entering the heating surface, and the real-time dissolved hydrogen concentration of the main steam system represents the outlet mixed hydrogen content after flowing through the complete four-tube heating surface of the boiler.
[0015] Furthermore, S2 specifically includes:
[0016] S201. Obtain historical operating data of the boiler unit during a period when no abnormal corrosion occurred and the oxide film on the pipe wall was stable, and construct a training sample set; use the thermodynamic parameters and chemical dosage characteristics in the training sample set as model input features, and use the background hydrogen concentration under the corresponding state of no new metal corrosion as model output features, train a support vector regression model or a long short-term memory network model to obtain the baseline prediction model; input the obtained thermodynamic parameters and chemical dosage characteristics at the current moment into the baseline prediction model, and output the theoretical background hydrogen concentration under the current operating condition;
[0017] S202. The total hydrogen production concentration of the computer group is calculated based on the difference between the real-time dissolved hydrogen concentration of the main steam system and the real-time dissolved hydrogen concentration of the feedwater system of the boiler unit.
[0018] S203. Calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration.
[0019] Furthermore, in S202 and S203, the formula for calculating the total hydrogen production concentration of the unit is as follows:
[0020]
[0021] in, This refers to the total hydrogen production concentration of the unit. This refers to the real-time dissolved hydrogen concentration of the main steam system. This refers to the real-time dissolved hydrogen concentration of the water supply system.
[0022] The formula for calculating the actual hydrogen production concentration due to corrosion is as follows:
[0023]
[0024] in, This represents the actual hydrogen production concentration due to corrosion. The theoretical background is the hydrogen concentration.
[0025] Furthermore, S3 specifically includes:
[0026] S301. Multiply the actual corrosion hydrogen production concentration by the steam-water circulation flow rate to calculate the total mass of hydrogen released by the actual corrosion reaction per unit time, which is used as the corrosion hydrogen production mass flow rate.
[0027] S302. Based on the chemical kinetic equation of the iron-water oxidation reaction, the mass transfer conversion inversion model is constructed. According to the molar mass conversion coefficient of hydrogen and iron in the chemical kinetic equation, the corrosion hydrogen production mass flow rate is multiplied by a fixed mass conversion ratio, and the equivalent metal consumption rate of the entire boiler four tubes is calculated by inversion.
[0028] S303. Calculate the average residence time of water vapor in the four boiler tubes based on the steam volume data, use the average residence time to perform a backward translation compensation on the time axis for the equivalent metal consumption rate, divide the compensated equivalent metal consumption rate by the total inner surface area of the four boiler tubes for normalization, and output the overall corrosion rate of the four boiler tubes at the current moment.
[0029] Furthermore, in step S303, the formula for calculating the average dwell time is:
[0030]
[0031] in, V is the average residence time of water vapor in the four tubes of the boiler; V is the total internal volume of the four tubes and connecting pipes of the boiler represented by the steam volume data; M is the water vapor mass flow rate represented by the steam-water circulation flow rate. The average water vapor density is determined based on the thermodynamic parameters in the historical operating parameters.
[0032] Furthermore, S4 specifically includes:
[0033] S401. The four tubes of the boiler are discretized into multiple physical tube segments according to their physical structure and thermal field distribution. When dividing the physical tube segments, the boiler four tubes are adaptively divided along the internal fluid flow direction based on the three-dimensional design drawings of the boiler unit and the location of the manifold, according to the heat flux density gradient of different heating areas inside the furnace. Among them, the area with the larger heat flux density gradient is divided into shorter physical tube segments.
[0034] S402. Abstract each physical pipe segment into a pipe segment node, establish the connection relationship between the pipe segment nodes according to the actual flow direction of feedwater and steam in the boiler between each physical pipe segment, and construct the spatial topology model.
[0035] S403. Extract the local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to each of the physical pipe segments from the historical operating parameters, perform numerical normalization processing, and splice them into a multi-dimensional feature vector, which is then mapped to the corresponding pipe segment node as the node attribute of each pipe segment node.
[0036] Furthermore, S5 specifically includes:
[0037] S501. Combining the node attributes of each pipe segment node, the graph attention mechanism is used to calculate the relative corrosion activity score of each pipe segment node under the current thermal environment, and the relative corrosion activity scores of all pipe segment nodes are globally normalized to obtain the spatial allocation weight of each pipe segment node.
[0038] S502. Multiply the overall corrosion rate with the corresponding spatial allocation weight to obtain the local corrosion risk index of each pipe segment node.
[0039] S503. Based on the material and thickness parameters of the physical pipe segment corresponding to each pipe segment node, pre-configure the corresponding preset node aging damage threshold; compare each of the real-time calculated local corrosion risk indices with the corresponding preset node aging damage thresholds. If the local corrosion risk index exceeds the corresponding threshold, generate a high-brightness corrosion risk assessment report containing the three-dimensional spatial coordinates of the corresponding physical pipe segment and trigger a dynamic early warning strategy; wherein, the dynamic early warning strategy includes operation adjustment instructions, chemical intervention instructions, and precise maintenance guidance.
[0040] Furthermore, in step S501, the calculation process for the spatial allocation weight specifically includes:
[0041] For any pipe segment node i in the spatial topology graph model, its concatenated multidimensional feature vector is used as the node attribute. Calculate the local attention coefficient between node i and its neighboring node j. The calculation formula is:
[0042]
[0043] in, Let i be the set of neighboring nodes of node i; The shared feature is the linear transformation weight matrix; ‖ represents the weight vector of the feedforward neural network with attention mechanism; ‖ represents the concatenation operation of feature vectors; It is a non-linear activation function;
[0044] Based on the local attention coefficient The updated feature vector of node i is obtained by weighted summation of the node attributes of adjacent nodes. :
[0045]
[0046] in, It is a non-linear activation function; the updated feature vector is generated using a linear mapping layer. Mapped to a one-dimensional scalar value, serving as the relative corrosion activity score for node i. :
[0047]
[0048] in, b is the mapping weight vector; b is the bias term;
[0049] The Softmax function is used to normalize the relative corrosion activity scores of all N pipe segment nodes globally, and the spatial allocation weight of each pipe segment node is calculated. : .
[0050] This invention also provides an intelligent assessment system for boiler four-tube corrosion risk based on dissolved hydrogen monitoring. Based on the intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring described above, the system includes:
[0051] The acquisition module is used to acquire historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system.
[0052] The first calculation module is used to predict the theoretical background hydrogen concentration under the current operating conditions based on the thermal parameters and dosage characteristics using a pre-built baseline prediction model, calculate the total hydrogen production concentration of the unit based on the real-time dissolved hydrogen concentration, and calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration.
[0053] The second calculation module is used to input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and combine the steam-water circulation flow rate and steam volume data to calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment.
[0054] The construction module is used to discretize the four tubes of the boiler into multiple physical tube segments according to their physical structure, and to construct a spatial topology model representing the connection relationship between the steam and water flow directions of the physical tube segments and including multiple tube segment nodes. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments;
[0055] The evaluation module is used to calculate the spatial allocation weight by combining the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight, so as to obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold, and generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.
[0056] The beneficial effects of this invention are as follows:
[0057] This invention monitors dissolved hydrogen at both ends of the steam-water circulation inlet and outlet and introduces a baseline prediction model to remove background hydrogen interference caused by system chemical decomposition, accurately extracting the true actual corrosion hydrogen concentration, significantly improving the signal-to-noise ratio and accuracy of corrosion monitoring. Simultaneously, it discretizes the boiler's four tubes into a spatial topology model based on their physical structure, integrating real physical field characteristics such as local heat load, flue gas temperature deviation, and wall temperature as tube segment node attributes. Using a graph attention mechanism to calculate spatial allocation weights, it accurately distributes the overall macroscopic corrosion rate to each microscopic tube segment node. This not only overcomes the technical bottleneck of traditional steam-water monitoring's inability to perform spatial source tracing, achieving quantitative risk positioning from the overall pipeline network to local microscopic tube segments, but also provides scientific guidance for unit operation adjustments and precise maintenance by generating dynamic early warning strategies, significantly reducing the risk of tube rupture and maintenance costs. Attached Figure Description
[0058] Figure 1 This is a flowchart illustrating the intelligent assessment method for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring, as described in this invention.
[0059] Figure 2 This is a schematic diagram of the intelligent assessment system for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring, as per the present invention.
[0060] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0061] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0062] like Figure 1 As shown, this invention also provides an intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring, including:
[0063] S1. Obtain historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system.
[0064] In a preferred embodiment of the present invention, step S1 specifically includes the following sub-steps:
[0065] S101. The historical operating status data within a set time period is continuously acquired through the unit's data acquisition network as historical operating parameters, and the data is cleaned and timestamped.
[0066] For thermal parameters and dosage characteristics, thermal parameters such as unit operating load, main steam pressure, main steam temperature, and feedwater temperature are collected through the unit control system. At the same time, the injection flow rate and concentration of chemical deoxygenating agents (such as hydrazine) are collected as dosage characteristics.
[0067] The characteristics of local heat load, flue gas temperature deviation, and wall temperature are collected by an array of temperature sensors and heat flow meters arranged inside the boiler furnace and on the surface of the four tubes, so as to accurately characterize the actual combustion conditions and the severity of metal service in different heating surface areas inside the boiler.
[0068] The data cleaning and timestamp alignment process specifically involves: identifying and removing sensor failure anomalies and missing values from the historical operating parameters; and, since there are differences in the original sampling frequencies of thermal parameters and water vapor chemical parameters, using a preset evaluation and analysis period as the time reference, employing a polynomial interpolation algorithm to resample the historical operating parameters at different sampling frequencies, thereby aligning all feature data to the same time resolution.
[0069] S102. Extract static physical structural parameters and fluid dynamic boundary conditions from the boiler unit's design drawings, manufacturing records, or three-dimensional digital model as unit structural data.
[0070] For the steam-water circulation flow rate, obtain the total feedwater flow rate and the total main steam flow rate of the unit under different load commands.
[0071] For steam volume data, extract the fluid channel volume inside each stage of the heating surface, such as the economizer, water-cooled wall, superheater, and reheater.
[0072] S103. Highly sensitive online dissolved hydrogen monitoring instruments are installed at the inlet and outlet nodes of the boiler unit's steam-water circulation system to obtain the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system.
[0073] For the feedwater system, sampling points are set up in the economizer inlet pipe or the high-pressure heater outlet pipe to collect the dissolved hydrogen concentration in the feedwater in real time. This concentration represents the initial background hydrogen content before entering the boiler heating surface.
[0074] For the main steam system, sampling points are set at the superheater outlet header or the main steam main pipe to collect the dissolved hydrogen concentration of the main steam in real time. This concentration represents the mixed hydrogen content at the outlet after the fluid flows through the entire boiler's four-tube heating surface.
[0075] S2. Based on the thermal parameters and dosage characteristics, a pre-constructed baseline prediction model is used to predict the theoretical background hydrogen concentration under the current operating conditions. The total hydrogen production concentration of the unit is calculated based on the real-time dissolved hydrogen concentration. The difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration is calculated to obtain the actual corrosion hydrogen production concentration.
[0076] In a preferred embodiment of the present invention, step S2 specifically includes the following sub-steps:
[0077] S201. The theoretical background hydrogen concentration under current operating conditions is predicted using a pre-built baseline prediction model. Specifically,
[0078] The hydrogen in the water-steam system does not come entirely from the corrosion reaction of the pipe wall metal. Under normal operating conditions, trace amounts of residual hydrogen that are not completely removed by the deaerator, as well as the thermal decomposition of chemical deaerators (such as hydrazine, acetone oxime, etc.) added to the water-steam system under high temperature and high pressure (for example, hydrazine decomposes at high temperature to produce ammonia, nitrogen and hydrogen), will all produce non-corrosive "background hydrogen".
[0079] In this embodiment, a baseline prediction model is pre-built (e.g., using Support Vector Regression (SVR) or Long Short-Term Memory (LSTM) networks). The construction and prediction process of the baseline prediction model specifically includes:
[0080] Historical operating data of the boiler unit during a period when no abnormal corrosion occurred and the oxide film on the pipe wall was stable were acquired to construct a training sample set. The thermodynamic parameters (unit load, feedwater temperature, main steam pressure, etc.) and chemical dosing characteristics (deaerator dosing pump stroke or instantaneous dosing flow rate) from the training sample set were used as model input features, and the background hydrogen concentration under the corresponding state of no new metal corrosion was used as the model output feature. A support vector regression model or a long short-term memory network model was trained to obtain the baseline prediction model. The acquired thermodynamic parameters and chemical dosing characteristics at the current moment were input into the baseline prediction model to output the predicted theoretical background hydrogen concentration under the current operating condition. .
[0081] S202. The total hydrogen concentration is calculated based on the real-time dissolved hydrogen concentration of the main steam system and feedwater system of the boiler unit. Specifically, the real-time dissolved hydrogen concentration measured at the sampling point of the feedwater system represents the initial hydrogen concentration of the fluid before it enters the boiler heating surface. The real-time dissolved hydrogen concentration measured at the sampling point of the main steam system represents the outlet mixed hydrogen concentration of the fluid after it has been heated by the boiler's four-tube heating surfaces. .
[0082] Because hydrogen has extremely high volatility and stability in high-temperature steam, and the boiler's four tubes form a relatively closed heating and pressurization channel, the formula... By performing calculations, the increase in the total hydrogen production concentration of the unit during the flow of fluid through all four tubes of the boiler can be obtained. .
[0083] S203. Calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration. Specifically, due to the total hydrogen production concentration of the unit... It includes the superposition of hydrogen production from metal corrosion and hydrogen production from chemical decomposition, therefore, using the formula... Difference calculations were performed to remove background noise. The calculated actual hydrogen production concentration from corrosion was then obtained. In a physicochemical sense, it is strictly defined as the characteristic hydrogen concentration released purely from the "iron-water reaction" (i.e., the oxidation reaction between high-temperature water vapor and carbon steel or alloy steel in the pipe wall to generate magnetite and hydrogen: 3Fe + 4H2O → Fe3O4 + 4H2), which improves the signal-to-noise ratio and sensitivity of corrosion monitoring and avoids false alarms of corrosion caused by changes in unit load or fluctuations in chemical dosage.
[0084] S3. Input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and combine it with the steam-water circulation flow rate and steam volume data to calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment.
[0085] In a preferred embodiment of the present invention, step S3 specifically includes the following sub-steps:
[0086] S301. Based on the steam-water circulation flow rate, calculate the actual corrosion hydrogen production mass flow rate at the current moment. Specifically, the actual corrosion hydrogen production concentration output in step S2 is only a relative ratio of volume or mass (e.g., micrograms per liter, μg / L). To convert it into an absolute product mass, dynamic fluid parameters must be introduced.
[0087] In this embodiment, the steam-water circulation flow rate (i.e., the total feedwater flow rate or the total main steam flow rate flowing through the four boiler tubes under the current load) is extracted from the unit structure data, and multiplied by the actual corrosion hydrogen production concentration to calculate the total mass of hydrogen released by the actual corrosion reaction per unit time, i.e., the corrosion hydrogen production mass flow rate. This flow rate represents the intensity of instantaneous corrosion hydrogen production on the macroscopic scale of the entire four boiler tubes.
[0088] S302. Calculate the equivalent metal consumption rate using the stoichiometric equations in the mass transfer and conversion inversion model. Specifically, the main corrosion mechanism of the four tubes of a thermal power plant boiler under high temperature and high pressure water-steam environment is the "iron-water oxidation reaction" of carbon steel or alloy steel. Its basic chemical kinetic equation is: 3Fe + 4H₂O → Fe₃O₄ + 4H₂. The mass transfer and conversion inversion model incorporates the stoichiometric conversion logic for the above oxidation reaction. According to this equation, for every 4 moles (approximately 8 grams) of hydrogen produced, 3 moles (approximately 168 grams) of iron are consumed. Therefore, by multiplying the corrosion hydrogen production mass flow rate calculated in S301 by a fixed molar mass conversion coefficient (i.e., 168 / 8 = 21), the mass transfer and conversion inversion model can calculate the equivalent metal consumption rate (e.g., grams per hour, g / h) of the entire four tubes of the boiler at the current moment, achieving a cross-dimensional conversion from chemical gas quantity to metal tube wall loss.
[0089] S303. Combine the steam volume data with time delay correction and normalization to output the overall corrosion rate of the four boiler tubes. Specifically, due to the large internal space of the four boiler tubes, there is a physical transport time delay (i.e., mass transfer delay) in the hydrogen gas generated by metal corrosion from the generation location to the main steam sampling point along with the water vapor.
[0090] In this embodiment, the average residence time of steam and water in the four boiler tubes is calculated using the steam volume data and steam-water circulation flow rate from the unit structure data. The specific calculation formula is as follows:
[0091]
[0092] In the formula, V is the average residence time of water vapor in the four tubes of the boiler; V is the total internal volume of the four tubes and connecting pipes of the boiler represented by the steam volume data; M is the water vapor mass flow rate represented by the steam-water circulation flow rate. The average water vapor density is obtained by looking up a table or calculating based on the thermodynamic parameters (i.e., the current average operating pressure and average operating temperature) in the historical operating parameters.
[0093] Calculate the average stay time Then, using this time, the equivalent metal consumption rate output by S302 is shifted along the time axis and dynamically compensated (i.e., the hydrogen production measured at the current moment actually corresponds to...). The corrosion state prior to time is determined to eliminate measurement lag caused by fluid transport. Subsequently, the compensated consumption rate is normalized by dividing the total inner surface area of the four boiler tubes, resulting in a final output with standardized engineering dimensions (e.g., ...). The overall corrosion rate is described. This overall corrosion rate accurately and in real time reflects the macroscopic corrosion degradation level of the entire boiler piping system after eliminating fluid hysteresis interference.
[0094] S4. The four tubes of the boiler are discretized into multiple physical tube segments according to their physical structure, and a spatial topology model representing the connection relationship between the steam and water flow between the physical tube segments and including multiple tube segment nodes is constructed. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments.
[0095] In a preferred embodiment of the present invention, step S4 specifically includes the following sub-steps:
[0096] S401. The four boiler tubes are discretized into multiple physical tube segments according to their physical structure and thermal field distribution. Specifically, the four boiler tubes (economizer, water wall, superheater, and reheater) are physically composed of hundreds or even thousands of metal tube bundles arranged in an interlaced manner, with a huge overall length and a complex thermal environment. In order to achieve precise corrosion localization at the microscale, the entire tube panel cannot be treated as a whole.
[0097] In this embodiment, combining the unit's three-dimensional design drawings and the location of the manifold, and based on the differences in thermal field distribution in different heating areas inside the furnace (e.g., burner elevation area, flame deflector area, roof area, and tail shaft flue area), the continuous four boiler tubes are divided into a finite number of discrete physical tube segments along the flow direction of the internal fluid (feedwater or steam). The length and diameter of each physical tube segment are adaptively divided according to the heat flux density gradient of the area (segments in areas with a large heat flux density gradient are shorter), thereby ensuring that each independently divided physical tube segment can be approximated as a tube wall micro-element with a uniform thermodynamic and hydrodynamic environment inside.
[0098] S402. Construct a spatial topology graph model representing the connection relationship between the steam and water flow directions of physical pipe segments and including multiple pipe segment nodes. Specifically, each physical pipe segment obtained in S401 above is abstracted as a pipe segment node in a graph network algorithm, ensuring that the pipe segment node and the physical pipe segment present a unique one-to-one correspondence in spatial location.
[0099] Subsequently, based on the actual design flow direction of feedwater and steam within the boiler and between various pipe panels and headers, the connection topology between nodes is established. If the internal working fluid outlet of physical pipe segment A is directly fluidly connected to the internal working fluid inlet of physical pipe segment B, a directed edge representing the fluid mass transfer relationship is established between the corresponding pipe segment node A and pipe segment node B. Based on the set of all pipe segment nodes and directed edges, a spatial topology graph model with global physical topology significance is constructed and stored and represented in the form of an adjacency matrix, enabling the connectivity of the macroscopic pipe network to be displayed in digital form.
[0100] S403. Integrate multidimensional thermodynamic characteristics into the node attributes of each of the aforementioned pipe sections. Specifically, under the premise of a systematically consistent water vapor chemical environment (dissolved hydrogen, pH value, etc.), the direct physical cause of severe local corrosion (i.e., corrosion deviation) in different pipe sections of the four boiler tubes lies in the severity of the local thermal environment on the outer wall of the pipe.
[0101] In this embodiment, three types of core thermodynamic characteristics (local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics) that match the spatial location of each physical pipe section are extracted from the historical operating parameters obtained in step S1. Specifically, the local heat load characteristics reflect the flame radiation intensity or convective heat transfer intensity at the elevation of the pipe section or the furnace cross section. The flue gas temperature deviation characteristics reflect the degree of local flue gas overheating caused by the uneven aerodynamic field of the furnace (such as residual rotation caused by tangential combustion at the four corners). The wall temperature characteristics are obtained by actual measurement of the wall temperature thermocouple or by calculation based on fluid-structure interaction heat transfer, and represent the solid metal wall temperature of the physical pipe section. This temperature directly and exponentially affects the chemical kinetic hydrogen production rate of the "iron-water reaction".
[0102] After numerical normalization of these three types of features, they are concatenated into a multidimensional feature vector, which is then mapped to the corresponding pipe segment node as the node attribute of that pipe segment node (i.e., the node feature matrix in the graph neural network).
[0103] S5. Calculate the spatial allocation weight based on the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight. Obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold to generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.
[0104] In a preferred embodiment of the present invention, step S5 specifically includes the following sub-steps:
[0105] S501. Combining the node attributes of each pipe segment node, the spatial allocation weight is calculated using a graph attention mechanism. Specifically, since the corrosion of a local pipe segment depends not only on its own thermal state but also on the topological neighborhood of heat conduction between upstream and downstream fluids, this embodiment uses a graph attention mechanism (GAT) for feature aggregation and weight calculation. For any pipe segment node i in the spatial topological graph model, let its node attributes (i.e., the concatenated multidimensional feature vector) be... First, calculate the local attention coefficients between node i and its neighboring node j. The calculation formula is:
[0106]
[0107] in, Let i be the set of neighboring nodes of node i (determined by the directed edges of the spatial topology graph model). The shared feature is the linear transformation weight matrix; ‖ represents the weight vector of the feedforward neural network with attention mechanism; ‖ represents the concatenation operation of feature vectors; It is a non-linear activation function.
[0108] Based on the local attention coefficient We perform a weighted summation of the features of adjacent nodes to obtain the updated feature vector of node i that incorporates the thermodynamic information of the spatial topological neighborhood. :
[0109]
[0110] in, The activation function is a non-linear activation function (such as ReLU). Subsequently, a linear mapping layer is used to update the feature vector. Mapped to a one-dimensional scalar value, serving as the relative corrosion activity score of node i under the current thermal environment. :
[0111]
[0112] in, is the mapping weight vector; b is the bias term. Finally, in order to convert the score into a proportion that can be used to allocate the total amount, the Softmax function is used to normalize the relative corrosion activity scores of all N pipe segment nodes globally, and the spatial allocation weight of each pipe segment node is calculated. :
[0113]
[0114] Calculated weights The global sum is 1, which in a physical sense accurately represents the percentage of the corrosion severity of the i-th pipe segment node at the current moment relative to the overall corrosion share of the unit.
[0115] S502. The overall corrosion rate is allocated to each of the pipe segment nodes according to the spatial allocation weight to obtain the local corrosion risk index of each of the pipe segment nodes. Specifically,
[0116] The overall corrosion rate calculated in step S3 is a macroscopic average of the entire system and cannot guide specific corrosion prevention and maintenance. In this embodiment, the overall corrosion rate of the four boiler tubes output in step S3 is used... The spatial allocation weights of each pipe segment node are directly compared with those calculated in step S501. Performing a product operation, that is Calculated This refers to the local corrosion risk index of the i-th pipe segment node (e.g., characterized by the instantaneous corrosion depth increase rate of that local pipe segment, in μm / h). Through this weighting mechanism based on physical field characteristics, accurate reverse tracing and localization mapping of dissolved hydrogen main pipe data to microscopic heated surface pipe segments is achieved.
[0117] S503. The localized corrosion risk index is compared with the preset node aging damage threshold to generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node. Specifically,
[0118] Furnace tubes in different physical locations have different materials (e.g., carbon steel 15CrMoG and alloy steel T91) and different service years, resulting in varying corrosion resistance. In this embodiment, a node aging damage threshold corresponding to the material and thickness is pre-configured for each tube segment node. The locally calculated corrosion risk index is compared with this threshold in real time. If the locally corroded risk index of a tube segment node exceeds the set threshold, the physical three-dimensional spatial coordinates of the tube segment node (e.g., at an elevation of 25 meters on the right side of the water-cooled wall) are highlighted in the corrosion risk assessment report, and a dynamic early warning strategy is triggered. The dynamic early warning strategy includes, but is not limited to, operation adjustment commands, chemical intervention commands, and precise maintenance guidance. Specifically:
[0119] Operational adjustment commands: Automatically send operational optimization suggestions to the unit's DCS system to adjust the burner tilt angle, reduce local heat load, or adjust the air-coal ratio to alleviate local overheating;
[0120] Chemical intervention instructions: Instruct chemical personnel to adjust the stroke of the dosing pump for oxygen absorbers or pH adjusters;
[0121] Precise maintenance guidance: High-risk physical pipe sections are automatically included in the mandatory inspection list for the next unit shutdown overhaul, guiding maintenance personnel to erect scaffolding at specific elevations and perform precise pipe cutting or ultrasonic thickness measurement, thereby replacing the traditional blind and random sampling inspection method.
[0122] like Figure 2 As shown, the present invention also provides an intelligent assessment system for boiler four-tube corrosion risk based on dissolved hydrogen monitoring. Based on the intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring as described above, the system includes:
[0123] The acquisition module 1 is used to acquire historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system.
[0124] The first calculation module 2 is used to predict the theoretical background hydrogen concentration under the current operating conditions based on the thermal parameters and dosage characteristics using a pre-built baseline prediction model, calculate the total hydrogen production concentration of the unit based on the real-time dissolved hydrogen concentration, and calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration.
[0125] The second calculation module 3 is used to input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment by combining the steam-water circulation flow rate and steam volume data.
[0126] Module 4 is used to discretize the four tubes of the boiler into multiple physical tube segments according to their physical structure, and to construct a spatial topology model representing the connection relationship between the steam and water flow directions of the physical tube segments and including multiple tube segment nodes. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments.
[0127] Evaluation module 5 is used to calculate the spatial allocation weight by combining the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight, to obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold, and generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.
[0128] Each of the above modules is used to perform the corresponding steps in the intelligent assessment method for corrosion risk of boiler four tubes based on dissolved hydrogen monitoring. The specific implementation method is as described in the above method embodiment, and will not be repeated here.
[0129] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.
[0130] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A smart assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring, characterized in that, include: S1. Obtain historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system. S2. Based on the thermal parameters and dosage characteristics, a pre-constructed baseline prediction model is used to predict the theoretical background hydrogen concentration under the current operating conditions. The total hydrogen production concentration of the unit is calculated based on the real-time dissolved hydrogen concentration. The difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration is calculated to obtain the actual corrosion hydrogen production concentration. S3. Input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and combine it with the steam-water circulation flow rate and steam volume data to calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment. S4. The four tubes of the boiler are discretized into multiple physical tube segments according to their physical structure, and a spatial topology model representing the connection relationship between the steam and water flow between the physical tube segments and including multiple tube segment nodes is constructed. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments; S5. Calculate the spatial allocation weight based on the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight. Obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold to generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.
2. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 1, characterized in that, S1 specifically includes: S101. Obtain the historical operating parameters within a set time period through the unit data acquisition network, and perform data cleaning and timestamp alignment on the historical operating parameters; wherein, the thermal parameters and chemical dosage characteristics are obtained through the unit control system, and the local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics are obtained through a sensor array arranged inside the boiler and on the surface of the four tubes. S102. Extract the unit structure data from the unit's three-dimensional digital model or design drawings and manufacturing ledger; wherein, the steam-water circulation flow rate includes the total feedwater flow rate and the total main steam flow rate under the corresponding load command, and the steam volume data includes the fluid channel volume inside the economizer, water-cooled wall, superheater and reheater at each stage of the heating surface; S103. The real-time dissolved hydrogen concentrations of the boiler unit feedwater system and main steam system are obtained by online dissolved hydrogen monitoring instruments installed at the boiler steam-water circulation inlet and outlet nodes, respectively; wherein, the real-time dissolved hydrogen concentration of the feedwater system represents the initial background hydrogen content before entering the heating surface, and the real-time dissolved hydrogen concentration of the main steam system represents the outlet mixed hydrogen content after flowing through the complete four-tube heating surface of the boiler.
3. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 1, characterized in that, S2 specifically includes: S201. Obtain historical operating data of the boiler unit during a period when no abnormal corrosion occurred and the oxide film on the pipe wall was stable, and construct a training sample set; use the thermodynamic parameters and chemical dosage characteristics in the training sample set as model input features, and use the background hydrogen concentration under the corresponding state of no new metal corrosion as model output features, train a support vector regression model or a long short-term memory network model to obtain the baseline prediction model; input the obtained thermodynamic parameters and chemical dosage characteristics at the current moment into the baseline prediction model, and output the theoretical background hydrogen concentration under the current operating condition; S202. The total hydrogen production concentration of the computer group is calculated based on the difference between the real-time dissolved hydrogen concentration of the main steam system and the real-time dissolved hydrogen concentration of the feedwater system of the boiler unit. S203. Calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration.
4. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 3, characterized in that, In S202 and S203, the formula for calculating the total hydrogen production concentration of the unit is as follows: in, This refers to the total hydrogen production concentration of the unit. This refers to the real-time dissolved hydrogen concentration of the main steam system. This refers to the real-time dissolved hydrogen concentration of the water supply system. The formula for calculating the actual hydrogen production concentration due to corrosion is as follows: in, This represents the actual hydrogen production concentration due to corrosion. The theoretical background is the hydrogen concentration.
5. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 1, characterized in that, S3 specifically includes: S301. Multiply the actual corrosion hydrogen production concentration by the steam-water circulation flow rate to calculate the total mass of hydrogen released by the actual corrosion reaction per unit time, which is used as the corrosion hydrogen production mass flow rate. S302. Based on the chemical kinetic equation of the iron-water oxidation reaction, the mass transfer conversion inversion model is constructed. According to the molar mass conversion coefficient of hydrogen and iron in the chemical kinetic equation, the corrosion hydrogen production mass flow rate is multiplied by a fixed mass conversion ratio, and the equivalent metal consumption rate of the entire boiler four tubes is calculated by inversion. S303. Calculate the average residence time of water vapor in the four boiler tubes based on the steam volume data, use the average residence time to perform a backward translation compensation on the time axis for the equivalent metal consumption rate, divide the compensated equivalent metal consumption rate by the total inner surface area of the four boiler tubes for normalization, and output the overall corrosion rate of the four boiler tubes at the current moment.
6. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 5, characterized in that, In step S303, the formula for calculating the average dwell time is: in, V is the average residence time of water vapor in the four tubes of the boiler; V is the total internal volume of the four tubes and connecting pipes of the boiler represented by the steam volume data; M is the water vapor mass flow rate represented by the steam-water circulation flow rate. The average water vapor density is determined based on the thermodynamic parameters in the historical operating parameters.
7. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 1, characterized in that, S4 specifically includes: S401. The four tubes of the boiler are discretized into multiple physical tube segments according to their physical structure and thermal field distribution. When dividing the physical tube segments, the boiler four tubes are adaptively divided along the internal fluid flow direction based on the three-dimensional design drawings of the boiler unit and the location of the manifold, according to the heat flux density gradient of different heating areas inside the furnace. Among them, the larger the heat flux density gradient, the shorter the length of the divided physical tube segment. S402. Abstract each physical pipe segment into a pipe segment node, establish the connection relationship between the pipe segment nodes according to the actual flow direction of feedwater and steam in the boiler between each physical pipe segment, and construct the spatial topology model. S403. Extract the local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to each of the physical pipe segments from the historical operating parameters, perform numerical normalization processing, and splice them into a multi-dimensional feature vector, which is then mapped to the corresponding pipe segment node as the node attribute of each pipe segment node.
8. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 1, characterized in that, S5 specifically includes: S501. Combining the node attributes of each pipe segment node, the graph attention mechanism is used to calculate the relative corrosion activity score of each pipe segment node under the current thermal environment, and the relative corrosion activity scores of all pipe segment nodes are globally normalized to obtain the spatial allocation weight of each pipe segment node. S502. Multiply the overall corrosion rate with the corresponding spatial allocation weight to obtain the local corrosion risk index of each pipe segment node. S503. Based on the material and thickness parameters of the physical pipe segment corresponding to each pipe segment node, pre-configure the corresponding preset node aging damage threshold; compare each of the real-time calculated local corrosion risk indices with the corresponding preset node aging damage thresholds. If the local corrosion risk index exceeds the corresponding threshold, generate a high-brightness corrosion risk assessment report containing the three-dimensional spatial coordinates of the corresponding physical pipe segment and trigger a dynamic early warning strategy; wherein, the dynamic early warning strategy includes operation adjustment instructions, chemical intervention instructions, and precise maintenance guidance.
9. The intelligent assessment method for boiler four-tube corrosion risk based on dissolved hydrogen monitoring according to claim 8, characterized in that, In step S501, the calculation process of the spatial allocation weight specifically includes: For any pipe segment node i in the spatial topology graph model, its concatenated multidimensional feature vector is used as the node attribute. Calculate the local attention coefficient between node i and its neighboring node j. The calculation formula is: in, Let i be the set of neighboring nodes of node i; The weight matrix is a shared feature linear transformation. ‖ represents the weight vector of the feedforward neural network with attention mechanism; ‖ represents the concatenation operation of feature vectors; It is a non-linear activation function; Based on the local attention coefficient The updated feature vector of node i is obtained by weighted summation of the node attributes of adjacent nodes. : in, It is a non-linear activation function; the updated feature vector is generated using a linear mapping layer. Mapped to a one-dimensional scalar value, serving as the relative corrosion activity score for node i. : in, b is the mapping weight vector; b is the bias term; The Softmax function is used to normalize the relative corrosion activity scores of all N pipe segment nodes globally, and the spatial allocation weight of each pipe segment node is calculated. : .
10. A boiler four-tube corrosion risk intelligent assessment system based on dissolved hydrogen monitoring, based on the boiler four-tube corrosion risk intelligent assessment method based on dissolved hydrogen monitoring according to any one of claims 1 to 9, characterized in that, The system includes: The acquisition module is used to acquire historical operating parameters, unit structure data, and real-time dissolved hydrogen concentration of the boiler unit; wherein, the historical operating parameters include thermal parameters, chemical dosage characteristics, local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics; the unit structure data includes steam-water circulation flow rate and steam volume data; and the real-time dissolved hydrogen concentration is the real-time dissolved hydrogen concentration of the boiler unit's feedwater system and main steam system. The first calculation module is used to predict the theoretical background hydrogen concentration under the current operating conditions based on the thermal parameters and dosage characteristics using a pre-built baseline prediction model, calculate the total hydrogen production concentration of the unit based on the real-time dissolved hydrogen concentration, and calculate the difference between the total hydrogen production concentration of the unit and the theoretical background hydrogen concentration to obtain the actual corrosion hydrogen production concentration. The second calculation module is used to input the actual corrosion hydrogen production concentration into the pre-constructed mass transfer and transformation inversion model, and combine the steam-water circulation flow rate and steam volume data to calculate the overall corrosion rate of the four boiler tubes in the boiler unit at the current moment. The construction module is used to discretize the four tubes of the boiler into multiple physical tube segments according to their physical structure, and to construct a spatial topology model representing the connection relationship between the steam and water flow directions of the physical tube segments and including multiple tube segment nodes. The local heat load characteristics, flue gas temperature deviation characteristics, and wall temperature characteristics corresponding to the physical tube segments in the historical operating parameters are used as the node attributes of each tube segment node; wherein, the tube segment nodes in the spatial topology model correspond one-to-one with the physical tube segments; The evaluation module is used to calculate the spatial allocation weight by combining the node attributes of each pipe segment node, and allocate the overall corrosion rate to each pipe segment node according to the spatial allocation weight, so as to obtain the local corrosion risk index of each pipe segment node and compare it with the preset node aging damage threshold, and generate a corrosion risk assessment report and dynamic early warning strategy for each pipe segment node of the four boiler tubes.