Evaluation method, system and device based on boiler cleaning monitoring data

Abstract

The invention relates to the technical field of boiler cleaning management, in particular to an evaluation method, system and device based on boiler cleaning monitoring data, and the method comprises the steps: comprehensively and accurately capturing boiler operation complex information through multi-source data acquisition and multi-physical field feature fusion; a feature evolution track is constructed to visually present a scaling dynamic state, so that a scaling trend can be pre-judged in advance; the phase change stage is accurately identified through the dynamic convolution kernel group, and the sensitivity to the scaling state change is improved; through thermal stress time-varying differential integral calculation, the regional vulnerability is scientifically evaluated and then is fused with phase change stage information to generate a vector, and the boiler scaling state is accurately reflected; through dynamic weighting directed network and key area judgment, resources can be concentrated to process high-risk areas, and the risk prevention and control efficiency is improved; key parameters are extracted by analyzing scaling risk conduction network output, a cleaning strategy is dynamically optimized, it is ensured that the cleaning operation always fits the actual situation of boiler scaling, and the operation cost is effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of boiler cleaning management technology, and in particular to evaluation methods, systems and devices based on boiler cleaning monitoring data. Background Technology

[0002] Current boiler cleaning management faces multiple challenges: Incomplete monitoring of the scaling process; traditional boiler cleaning management often relies on single or limited data sources for scaling monitoring, such as relying solely on changes in tube wall temperature or pressure, while ignoring the complex interactions between multiple physical fields. This leads to misjudgments or omissions in the scaling process, affecting cleaning effectiveness and boiler operational safety. Inaccurate identification of phase transition stages; phase transition stages in the scaling process (such as scale formation and thickening) are crucial for formulating cleaning strategies. Traditional methods may fail to accurately identify these phase transition stages, leading to inappropriate timing of cleaning. Unclear risk transmission mechanisms; scaling risks inside the boiler may be transmitted between different areas through mechanisms such as fluid flow and heat conduction. Traditional management methods often overlook this risk transmission mechanism, resulting in the inability to predict and control potential high-risk areas in a timely manner, increasing the uncertainty of boiler operation. Therefore, this invention proposes an evaluation method, system, and device based on boiler cleaning monitoring data. Summary of the Invention

[0003] The purpose of this invention is to solve the problems in the background art by proposing an evaluation method, system and device based on boiler cleaning monitoring data.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] The first aspect of this invention provides an evaluation method based on boiler cleaning monitoring data, comprising:

[0006] S1. Obtain the boiler cleaning assessment dataset, extract the multi-physics coupled feature tensor, and construct the feature evolution trajectory of the scaling process;

[0007] S2. Based on the characteristic evolution trajectory of the scaling process, a dynamic convolution kernel group is used to identify the phase transition stage. At the same time, the vulnerability of the region is calculated by the time-varying differential integral of thermal stress, and finally the scaling phase transition state vector is generated by fusion.

[0008] S3. Based on the topological dependency relationship of the scaling phase change state vector, construct the scaling risk transmission network and output the network structure data including node-edge relationships and risk transmission strength coefficients;

[0009] S4. Dynamically optimize the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

[0010] Furthermore, the process of obtaining boiler cleaning assessment datasets, extracting multiphysics coupled feature tensors, and constructing the feature evolution trajectory of the scaling process includes:

[0011] Boiler cleaning assessment datasets were obtained through multiple sources, including pipe wall temperature distribution data, pipe wall strain data, scale adhesion strength data, fluid velocity data, turbulent kinetic energy data, ion deposition related data, scale thickness data, and time information.

[0012] From the boiler cleaning assessment dataset, multi-physics interaction features are simultaneously analyzed and fused: by calculating the spatial mapping relationship between pipe wall thermal stress distribution and real-time scale adhesion intensity, thermo-mechanical coupling features, i.e., thermo-mechanical coupling covariance coefficients, are obtained; based on the dynamic inhibition effect of quantified fluid shear stress field and ion deposition rate, fluidization coupling features, i.e., fluidization coupling inverse correlation factors, are obtained; by capturing the spatial transmission delay effect of scale growth rate in different boiler areas, spatiotemporal correlation features, i.e., spatiotemporal correlation time lag mutual information, are obtained; the extracted multi-physics interaction features are integrated according to the spatiotemporal dimension to form a multi-physics coupling feature tensor.

[0013] The feature evolution trajectory of the scaling process is constructed based on the continuous change of the multi-physics field coupling feature tensor during the cleaning cycle.

[0014] Furthermore, the process of identifying phase transition stages using dynamic convolutional kernel groups, calculating regional vulnerability through time-varying differential-integral methods of thermal stress, and finally fusing to generate a fouling phase transition state vector includes:

[0015] Based on the input scaling process characteristics and evolution trajectory, the thermo-mechanical coupling covariance coefficient, the fluidization coupling inverse correlation factor, and the spatiotemporal correlation time lag mutual information are extracted.

[0016] A dynamic convolutional kernel group is constructed for identifying the scaling phase transition stage: a mutation detection kernel is used to scan the steep drop point of the thermo-mechanical coupling covariance coefficient, a continuously growing kernel is used to capture the monotonically increasing interval of the fluidization coupling inverse correlation factor, and an oscillating attenuation kernel is used to analyze the periodic fluctuations of the time lag mutual information of the spatiotemporal correlation; the phase transition event marker sequence of each region in the time domain is output by sliding the feature evolution trajectory of the convolutional kernel group; the scaling phase transition stage index is quantified according to the density distribution and duration of the phase transition event marker sequence and encoded using numbers 1 to 4;

[0017] Simultaneously, the regional vulnerability score is calculated based on the thermo-mechanical coupling covariance coefficient: the absolute value of the time-varying differential of the thermo-mechanical coupling covariance coefficient is integrated over time with its peak value.

[0018] ,

[0019] In the formula, Score the vulnerability of the region; Time-varying weighting factor; Spatial concentration weighting factor; The thermo-mechanical coupling covariance coefficient; This represents the peak value of the thermo-mechanical coupling covariance coefficient within the current time window; This is the integration time interval, corresponding to the boiler operating time in the time information. This represents the absolute value of the instantaneous rate of change of the thermo-mechanical coupling covariance coefficient;

[0020] By integrating the scaling phase transition stage index and the regional vulnerability score, a scaling phase transition state vector is generated.

[0021] Furthermore, based on the topological dependencies of the scaling phase transition state vector, the process of constructing a scaling risk transmission network and outputting network structure data containing node-edge relationships and risk transmission strength coefficients includes:

[0022] The boiler tube bundle system is abstracted as a dynamic weighted directed network, including node definition and directed edge definition: each tube segment unit is an independent node, and the scaling phase change stage index and regional vulnerability score are used as node attributes; adjacent nodes are connected according to the actual fluid flow path, and the weight of the directed edge is dynamically calculated by the product of the extracted fluidization coupling inverse correlation factor and the tube bundle topology spacing.

[0023] Perform downstream risk propagation at each node, i.e., iterative calculation of risk transmission intensity: calculate the scaling risk transmission intensity coefficient at each node. Along the directed edge direction, transfer the scaling risk transmission intensity coefficient of the current node. The values ​​are scaled according to edge weights and then superimposed onto downstream nodes: Downstream node risk value = Initial risk value + Scaling risk transmission strength coefficient × Directed edge weights; Set global convergence conditions, traverse all nodes to complete global iterative calculations, and update the risk values ​​of each node in the network;

[0024] Key region identification is performed based on node and edge attributes, including key transmission hub node identification and high-risk isolated node cluster identification. For key transmission hub node identification, the in-degree of each node is calculated. If the in-degree of a node is greater than a preset node connection density threshold and its risk value is greater than the average risk value, then the node is marked as a key transmission hub node. For high-risk isolated node cluster identification, the out-degree of each node is calculated. If the out-degree of a node is zero and its regional vulnerability score is greater than a preset vulnerability threshold, then the node is marked as a candidate isolated node. Isolated node clusters are generated through spatial clustering. The regional vulnerability score of the cluster is calculated using a formula. If the regional vulnerability score of the cluster is greater than a preset cluster vulnerability threshold, then the isolated node cluster is marked as a high-risk isolated node cluster.

[0025] Finally, a scaling risk transmission network is generated, and the network structure data is output, including a set of node attributes, a set of directed edge attributes, and a list of key area identifiers. The set of node attributes includes the scaling phase transition stage index, regional vulnerability score, and risk value for each node; the set of directed edge attributes includes the real-time weight of each edge and the scaling risk transmission intensity coefficient. Values; the list of key area identifiers includes key transmission hub node IDs and high-risk isolated node cluster IDs.

[0026] Furthermore, the process of dynamically optimizing the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol includes:

[0027] The output of the scaling risk transmission network is analyzed, and key parameters are extracted: the scaling phase transition stage index, regional vulnerability score, and risk value of each node are obtained based on node attributes; the directed edge weights and scaling risk transmission strength coefficients are obtained based on edge attributes. Simultaneously, load the set of key transmission hub node IDs and the set of high-risk isolated node cluster IDs from the list of key area identifiers as trigger conditions for subsequent cleaning strategies.

[0028] Based on the extracted key parameters, a three-level cleaning strategy is dynamically generated: a cleaning mode selection strategy is implemented: for the nodes corresponding to the key transmission hub node IDs, the cleaning mode is assigned according to their scaling phase change stage index.

[0029] Implement a dynamic adjustment strategy for cleaning parameters: For nodes in the high-risk isolated node cluster ID, dynamically allocate cleaning agent concentration gradient and pulse duration according to their regional vulnerability score;

[0030] Implement a cleaning operation chain generation strategy: sort operations in reverse order along the fluid flow direction, prioritizing those with a high scaling risk transmission intensity coefficient. Downstream node clusters exceeding a preset intensity threshold are formed into a cleaning priority sequence; if the priority sequence contains a key transmission hub node ID, its cleaning priority is raised to the highest level; at the same time, the cleaning resource allocation weight is adjusted according to the edge weight.

[0031] The final output is an adaptive cleaning protocol, which includes generating a cleaning operation chain with flow direction constraints, clearly marking the cleaning order and parameters of key transmission hub node IDs and high-risk isolated node cluster IDs; and dynamically updating the cleaning protocol to match the real-time scaling risk transmission network status.

[0032] A second aspect of the present invention provides an evaluation system based on boiler cleaning monitoring data, comprising:

[0033] Boiler cleaning assessment data analysis module: acquires boiler cleaning assessment dataset, extracts multi-physics coupled feature tensors, and constructs feature evolution trajectory of scaling process;

[0034] Scaling phase transition state vector generation module: Based on the characteristic evolution trajectory of the scaling process, dynamic convolution kernel group is used to identify the phase transition stage, and regional vulnerability is calculated through time-varying differential integral of thermal stress. Finally, the scaling phase transition state vector is generated by fusion.

[0035] Scaling risk transmission network construction module: Based on the topological dependency relationship of the scaling phase transition state vector, construct the scaling risk transmission network and output network structure data including node-edge relationships and risk transmission strength coefficients;

[0036] Cleaning strategy optimization protocol module: Based on the scaling risk transmission network, the three-level cleaning strategy is dynamically optimized to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

[0037] A third aspect of the present invention provides an evaluation device based on boiler cleaning monitoring data, including a processor, a memory, and a communication bus;

[0038] The memory stores a computer-readable program that can be executed by the processor;

[0039] The communication bus enables communication between the processor and the memory;

[0040] When the processor executes the computer-readable program, it performs an evaluation method based on boiler cleaning monitoring data as described in this invention.

[0041] Compared with existing technologies, the beneficial effects of this invention are as follows: It comprehensively acquires boiler cleaning assessment datasets covering multiple dimensions such as temperature and strain through multiple sources, providing a rich data foundation for subsequent analysis; it simultaneously analyzes and integrates multi-physics interaction features to form a multi-physics coupling feature tensor, thereby accurately capturing the complex relationships between various physical fields during boiler operation; it constructs the feature evolution trajectory of the scaling process based on continuous changes, intuitively presenting the dynamic development of scaling over time, providing a clear framework for subsequent analysis; it employs dynamic convolution kernel groups, which can accurately identify the scaling phase transition stage, and different convolution kernels scan for different feature changes, improving the accuracy and sensitivity of identification; it calculates regional vulnerability through time-varying differential-integral thermal stress calculation, comprehensively considering the temporal and spatial factors of thermal stress changes, thereby scientifically assessing the risk; and it integrates the phase transition stage and regional vulnerability information by fusing the generated scaling phase transition state vector, providing key indicators for accurately judging the boiler scaling state; through… The boiler tube bundle system is abstracted as a dynamic weighted directed network, clearly defining nodes and directed edges to intuitively reflect the system structure. Through iterative calculation of risk transmission intensity, the propagation process of scaling risk in the network can be accurately simulated, clarifying the risk changes at each node. By identifying key areas, critical transmission hub nodes and high-risk isolated node clusters are identified, providing direction for concentrated resource treatment of high-risk areas, thereby improving risk control efficiency and reducing the probability of accidents. By analyzing the output of the scaling risk transmission network and extracting key parameters, a basis for formulating cleaning strategies is provided. A dynamically generated three-level cleaning strategy allocates cleaning modes, parameters, and work chains according to different node characteristics, achieving personalized cleaning. Adopting a reverse sorting and priority-boosting strategy along the fluid flow direction ensures that high-risk areas are treated first, improving cleaning effectiveness. By dynamically updating the cleaning protocol to match the real-time network status, the cleaning strategy is guaranteed to always adapt to the boiler scaling situation, extending boiler lifespan and reducing operating costs. Attached Figure Description

[0042] Figure 1 This is a flowchart of the evaluation method based on boiler cleaning monitoring data proposed in this invention.

[0043] Figure 2 This is a block diagram of the evaluation system based on boiler cleaning monitoring data proposed in this invention. Detailed Implementation

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] Example 1

[0046] Please see Figure 1 This invention provides an evaluation method based on boiler cleaning monitoring data, including:

[0047] S1. Obtain the boiler cleaning assessment dataset, extract the multi-physics coupled feature tensor, and construct the feature evolution trajectory of the scaling process;

[0048] S2. Based on the characteristic evolution trajectory of the scaling process, a dynamic convolution kernel group is used to identify the phase transition stage. At the same time, the vulnerability of the region is calculated by the time-varying differential integral of thermal stress, and finally the scaling phase transition state vector is generated by fusion.

[0049] S3. Based on the topological dependency relationship of the scaling phase change state vector, construct the scaling risk transmission network and output the network structure data including node-edge relationships and risk transmission strength coefficients;

[0050] S4. Dynamically optimize the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

[0051] It should be further explained that, in the specific implementation process, the process of obtaining the boiler cleaning assessment dataset, extracting the multiphysics coupled feature tensor, and constructing the feature evolution trajectory of the scaling process includes:

[0052] Boiler cleaning assessment datasets were obtained through multiple sources, including pipe wall temperature distribution data, pipe wall strain data, scale adhesion strength data, fluid velocity data, turbulent kinetic energy data, ion deposition related data, scale thickness data, and time information.

[0053] Understandably, the tube wall temperature distribution data covers the real-time temperature of each micro-region at different locations on the boiler tube wall, used to calculate the temperature gradient; the tube wall strain data includes the real-time strain increment of each micro-region on the tube wall, combined with the temperature gradient to calculate the covariance coefficient; the scale adhesion strength data involves the adhesion strength of the scale at different locations on the tube wall, used for spatial mapping analysis with the thermal stress distribution; the fluid velocity data refers to the flow velocity information of the fluid at different locations within the boiler, used to calculate the fluid shear stress field; the turbulent kinetic energy data is the information on the peak region of the fluid turbulent kinetic energy; the ion deposition related data includes the fluid ion concentration at different locations, the deposit concentration, and their decay rate over time; the scale thickness data is the scale thickness of the tube bundle in different areas of the boiler and its change over time, used to calculate the scale growth rate; the time information includes the time point of measurement of various data, boiler operating time, cleaning cycle, etc., used to analyze the spatial transmission delay effect; the multi-source methods include, but are not limited to, industrial field measurements, simulation generation, and public resource access;

[0054] From the boiler cleaning assessment dataset, multi-physics interaction features are simultaneously analyzed and integrated: by calculating the spatial mapping relationship between the pipe wall thermal stress distribution and the real-time scale adhesion strength, thermo-mechanical coupling features, namely the thermo-mechanical coupling covariance coefficient, are obtained. Specifically, based on the pipe wall temperature distribution data, the temperature gradient of each micro-region at different locations on the boiler pipe wall is calculated using the numerical difference method; combined with the real-time strain increment of each micro-region in the pipe wall strain data, the covariance coefficient between the temperature gradient and the local strain increment is calculated. This coefficient reflects the spatial mapping relationship between the pipe wall thermal stress distribution and the real-time scale adhesion strength. For example, a region with a large covariance coefficient may indicate that the thermal stress has a significant impact on the scale adhesion strength in that region.

[0055] Based on the dynamic inhibition effect of the fluid shear stress field and ion deposition rate, the fluidization-chemical coupling characteristics, namely the fluidization-chemical coupling inverse correlation factor, are obtained. Specifically, based on fluid velocity data, the shear stress field of the fluid at different locations in the boiler is calculated using relevant fluid mechanics formulas to obtain the distribution of shear stress magnitude of the fluid on the pipe wall at each location. Combined with the sediment concentration data in the ion deposition data, the changes in sediment concentration at different times and in different fluid shear stress regions are analyzed to calculate the inverse correlation factor between the peak region of turbulent kinetic energy and the sediment concentration decay rate. This inverse correlation factor reflects the dynamic inhibition relationship between the fluid shear stress field and the ion deposition rate. For example, a large inverse correlation factor indicates that the fluid shear stress has a stronger inhibitory effect on ion deposition in this region. That is, as the fluid shear stress increases, the sediment concentration decay rate accelerates and the ion deposition rate decreases.

[0056] By capturing the spatial transmission delay effect of scale growth rate in different boiler areas, spatiotemporal correlation characteristics are obtained, namely, spatiotemporal correlation time lag mutual information. Specifically, based on scale thickness data, the scale growth rate is calculated at each time point using the scale growth rate calculation formula (i.e., scale growth rate = (current scale thickness - initial scale thickness) / time interval). Combining the measurement time point in the time information and the flow direction of the medium in the boiler, the change of scale growth rate between adjacent tube bundles along the medium flow direction with time is analyzed. The mutual information method in time series analysis is used to calculate the time lag mutual information of scale thickness growth between adjacent tube bundles. This time lag mutual information reflects the spatial transmission delay effect of scale growth rate in different boiler areas. For example, when the time lag mutual information is large and there is a significant time difference, it indicates that the scale growth of the upstream tube bundle has a certain time delay effect on the downstream tube bundle. That is, the change in scale growth of the downstream tube bundle will only appear some time after the change in scale growth of the upstream tube bundle. Moreover, the larger the mutual information value, the stronger this spatial transmission correlation.

[0057] The extracted multi-physics interaction features (thermal-mechanical coupling covariance coefficient, fluidization coupling inverse correlation factor, and spatiotemporal correlation time lag mutual information) are integrated according to the spatiotemporal dimension to form a multi-physics coupling feature tensor. This tensor comprehensively reflects the interaction of multi-physics fields and their spatiotemporal variation characteristics during boiler scaling.

[0058] The scaling process feature evolution trajectory is constructed based on the continuous change of the multi-physics coupling feature tensor within the cleaning cycle. Specifically, a scaling process feature evolution trajectory diagram is plotted, with the horizontal axis representing time (days), for example, covering 3 cleaning cycles totaling 90 days; and the vertical axis representing the thermo-mechanical coupling covariance coefficient, the fluidization coupling inverse correlation factor, and the spatiotemporal correlation time lag mutual information, respectively. Three curves are plotted using professional plotting software (such as Origin, Matplotlib, etc.), marking the three feature values ​​calculated at each time point on the corresponding coordinate positions. Then, adjacent data points are connected with a smooth curve to obtain the scaling process feature evolution trajectory diagram. This trajectory diagram allows for a direct observation of the dynamic changes in the boiler scaling process within different cleaning cycles, providing an important basis for boiler cleaning assessment and optimization.

[0059] It should be further explained that, in the specific implementation process, the process of using dynamic convolutional kernel groups to identify the phase transition stage, and simultaneously calculating regional vulnerability through time-varying differential-integral thermal stress, ultimately fusing to generate a fouling phase transition state vector, includes:

[0060] Based on the input scaling process characteristics and evolution trajectory, the thermo-mechanical coupling covariance coefficient, the fluidization coupling inverse correlation factor, and the spatiotemporal correlation time lag mutual information are extracted.

[0061] Constructing dynamic convolutional kernel sets for identifying fouling phase transition stages:

[0062] Z1. Employ mutation detection to scan the steep drop point of the thermo-mechanical coupling covariance coefficient in the time domain: By monitoring the mutation of the covariance coefficient value in real time (such as the sudden drop caused by thermal stress release), the critical time point of scale layer peeling or structural failure is accurately marked. For example, when the covariance coefficient drops by more than 30% within 2 hours, a peeling event warning is triggered.

[0063] Z2. Use a continuously growing kernel to capture the monotonically increasing interval of the fluid-chemical coupling inverse correlation factor: track the dynamic inhibition relationship between fluid shear stress and ion deposition rate. When the fluid-chemical coupling inverse correlation factor shows an upward trend for 3 consecutive days (e.g., turbulence weakening leads to a 5% increase in deposit concentration per day), the scaling acceleration period is identified and marked as a high-risk growth interval.

[0064] Z3. Analyze the periodic fluctuations of time-space correlation time lag mutual information through oscillation decay kernel: quantify the delay effect of upstream tube scale growth on downstream tubes (e.g., delay time of 2-3 days). When the time-space correlation time lag mutual information exhibits an oscillation decay characteristic with a period of 1 day, it is marked as a time-space coupling event of scale propagation.

[0065] Z4. By sliding the convolutional kernel group to scan the feature evolution trajectory, output the phase transition event label sequence of each region in the time domain: slide the feature evolution trajectory along the time axis with a 3-day window. Within each window, three kernels simultaneously output local feature values. Specifically: the mutation detection kernel generates a thermo-mechanical coupling steep drop label (0 / 1 binary), the continuous growth kernel generates a fluidization coupling growth label (0-1 continuous value), and the oscillation decay kernel generates a spatiotemporal correlated oscillation label (sine wave amplitude). After traversing the data of the entire cleaning cycle (e.g., 90 days), integrate to generate a full-time-domain phase transition event sequence, in the format [time point, region ID, event type (level 1-4), Duration], where the event type is encoded based on a combination of features (e.g., a sharp drop + growth marker appearing simultaneously is encoded as a level 3 phase change event); for example, on the 45th day of boiler operation, the sliding window detected: the thermo-mechanical coupling covariance coefficient dropped sharply from 0.8 to 0.3 (triggered by a mutation detection kernel); the fluidization coupling inverse correlation factor increased for 5 consecutive days (triggered by a continuous growth kernel); the time lag mutual information of the spatiotemporal correlation showed an oscillation with a period of 2 days (triggered by an oscillation decay kernel); the system outputs the marker sequence: [45 days, region A (e.g., boiler section 2), 4 (stripping period), 2 days];

[0066] Z5. Based on the density distribution of the phase change event marker sequence, quantify the scaling phase change stage index and encode it using numbers 1 to 4. The number of event markers per unit time is used as the density distribution to quantify phase change activity. The numerical encoding specifically includes: 1. Nucleation stage: fluctuations in all three characteristics are less than 10% of the baseline value; 2. Growth stage: fluid-chemical coupling inverse correlation factor continuously increases and mutation nuclei show no response; 3. Stable stage: oscillations of spatiotemporal correlation time lag mutual information decay and thermo-mechanical coupling covariance coefficient tends to stabilize; 4. Peeling stage: thermo-mechanical coupling covariance coefficient drops sharply by more than 30% and is accompanied by disordered spatiotemporal correlation time lag mutual information.

[0067] Simultaneously, the regional vulnerability score is calculated based on the thermo-mechanical coupling covariance coefficient: the absolute value of the time-varying differential of the thermo-mechanical coupling covariance coefficient is integrated over time with its peak value.

[0068] ,

[0069] In the formula, Score the vulnerability of the region; The time-varying sensitive weighting factor is adjusted based on the pipe wall strain data; The spatial concentration weighting factor is dynamically adjusted based on the distribution entropy of the thermo-mechanical coupling covariance coefficient. The thermo-mechanical coupling covariance coefficient; This represents the peak value of the thermo-mechanical coupling covariance coefficient within the current time window, used to characterize the long-term stress concentration effect. This is the integration time interval, corresponding to the boiler operating time in the time information. This represents the absolute value of the instantaneous rate of change of the thermo-mechanical coupling covariance coefficient;

[0070] Understandably, the regional vulnerability score is a quantitative result of integrating the absolute value of the time-varying differential of the thermo-mechanical coupling covariance coefficient with the peak value over time. It is used to reflect the cumulative damage effect and spatial distribution vulnerability of the boiler tube wall region caused by thermal stress shock.

[0071] By integrating the scaling phase transition stage index and the regional vulnerability score, a scaling phase transition state vector is generated.

[0072] It should be further explained that, in the specific implementation process, the process of constructing a scaling risk transmission network based on the topological dependency relationship of the scaling phase transition state vector, and outputting network structure data containing node-edge relationships and risk transmission strength coefficients, includes:

[0073] The boiler tube bundle system is abstracted as a dynamic weighted directed network, including node definition and directed edge definition: each tube segment unit is an independent node, and the scaling phase change stage index and regional vulnerability score are used as node attributes; adjacent nodes are connected according to the actual fluid flow path, and the weight of the directed edge is dynamically calculated by multiplying the extracted fluidization coupling inverse correlation factor and the tube bundle topology spacing; wherein, the tube bundle topology spacing is calculated based on the actual pipeline layout distance of the boiler tube bundle system, that is, the distance between adjacent tube segment units is determined based on the actual connection length or relative position relationship of the pipeline in physical space;

[0074] Perform downstream risk propagation at each node, i.e., iterative calculation of risk transmission intensity: calculate the scaling risk transmission intensity coefficient at each node. :

[0075] ,

[0076] In the formula, This is the time sensitivity coefficient; Spatial sensitivity coefficient; This is a time dimension term used to characterize the rate of scale development; This represents the index change during the scaling phase transition stage. This represents the scaling phase transition stage index at the current monitoring time point. This represents the scaling phase transition stage index at the previous effective monitoring time point; For time intervals, For the current monitoring time point, This refers to the previous valid monitoring time point. Indexed by time point; This is a spatial dimension term used to characterize the risk spatial gradient; This represents the change in the regional vulnerability score. Score the regional vulnerability of the current node. Score the regional vulnerability of adjacent nodes; This refers to the distance between adjacent nodes, i.e., the distance between pipe layouts.

[0077] Along the directed edge direction, transfer the scaling risk transmission intensity coefficient of the current node. The values ​​are scaled according to edge weights and then superimposed onto downstream nodes: Downstream node risk value = Initial risk value + Scaling risk transmission strength coefficient × Directed edge weight;

[0078] Among them, the initial risk value of each node is calculated based on node attributes. :

[0079] ,

[0080] In the formula, , where are the weighting coefficients for the scaling phase transition stage index and the regional vulnerability score, respectively; For nodes The scaling phase transition stage index; For nodes Regional vulnerability score;

[0081] Set global convergence conditions, traverse all nodes to complete global iterative calculations, and update the risk values ​​of each node in the network.

[0082] The global convergence condition is: In the formula, A preset convergence threshold is set; simultaneously, the risk value of each node in the network is updated using the following formula:

[0083] ,

[0084] In the formula, For the updated node The risk value; This represents the number of iterations. For nodes To the node The directed edge risk transmission strength coefficient; For nodes To the node The directed edge weights (calculated by the product of the fluidization coupling inverse correlation factor and the topological spacing); upstream nodes Represents all directly oriented nodes Upstream pipe section unit for conveying fluid;

[0085] Key areas are determined based on node and edge attributes, including key transmission hub nodes and high-risk isolated node clusters.

[0086] In determining key transmission hub nodes, the in-degree of each node (the number of directed edges pointing to that node) is calculated. If the in-degree of a node is greater than a preset node connection density threshold and its risk value is greater than the average risk value, then the node is marked as a key transmission hub node. Node connection density refers to the degree of tightness of the connections around a node as reflected by the number of directed edges (in-degree) pointing to a node relative to the overall topology in the dynamic weighted directed network of the boiler tube bundle system.

[0087] For identifying high-risk isolated node clusters, the out-degree of each node (the number of directed edges emanating from that node) is calculated. If a node's out-degree is zero (no downstream propagation path) and its regional vulnerability score is greater than a preset vulnerability threshold, then that node is marked as a candidate isolated node. Isolated node clusters are generated through spatial clustering: specifically, the 3D coordinates of all candidate isolated nodes (from the boiler tube bundle layout) are extracted, spatially adjacent nodes are connected using the Delaunay triangulation algorithm, and nodes with a topological distance less than the tube diameter threshold are merged to generate isolated node clusters. The regional vulnerability score of the cluster is calculated using a formula. If the regional vulnerability score of the cluster is greater than a preset cluster vulnerability threshold, then the isolated node cluster is marked as a high-risk isolated node cluster. The regional vulnerability score of the cluster... The calculation formula is:

[0088] ,

[0089] In the formula, Indicates the number within the cluster Regional vulnerability score for each candidate isolated node; Indicates the number within the cluster The equivalent weight of pipe segments for each candidate island node, where the equivalent weight of pipe segment = pipe segment length × scale thickness gradient, is used to quantify the spatial importance of different pipe segments in the transmission of scaling risk. Index of candidate isolated nodes in the cluster; This represents the number of candidate isolated nodes in the cluster.

[0090] Finally, a scaling risk transmission network is generated, and the network structure data is output, including a set of node attributes (scaling phase transition stage index, regional vulnerability score, and risk value for each node) and a set of directed edge attributes (real-time weight and scaling risk transmission strength coefficient for each edge). Value) and a list of key area identifiers (key transmission hub node ID, high-risk isolated node cluster ID).

[0091] It should be further explained that, in the specific implementation process, the process of dynamically optimizing the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol includes:

[0092] The output of the scaling risk transmission network was analyzed, and key parameters were extracted: the scaling phase transition stage index (1-4), regional vulnerability score, and risk value of each node were obtained based on node attributes; the directed edge weights (fluidization coupling inverse correlation factor × tube bundle topology spacing) and scaling risk transmission intensity coefficients were obtained based on edge attributes. Simultaneously, load the set of key transmission hub node IDs and the set of high-risk isolated node cluster IDs from the list of key area identifiers as trigger conditions for subsequent cleaning strategies.

[0093] Based on the extracted key parameters, a three-level cleaning strategy is dynamically generated: A cleaning mode selection strategy is implemented: For nodes corresponding to key transmission hub node IDs, cleaning modes are assigned according to their scaling phase change stage index. Specifically, if the scaling phase change stage index is 1 (nucleation stage), a low-concentration chemical cleaning mode is used; if the scaling phase change stage index is 2-3 (growth stage, stable stage), a mechanical cleaning + chemical composite mode is used; if the scaling phase change stage index is 4 (stripping stage), a high-concentration chemical cleaning + pulse flushing mode is used; if the node also belongs to a high-risk isolated node cluster ID, it is forcibly upgraded to a high-concentration chemical + pulse mode.

[0094] Implement a dynamic adjustment strategy for cleaning parameters: For nodes in the high-risk isolated node cluster ID, dynamically allocate the cleaning agent concentration gradient and pulse duration according to their regional vulnerability score; understandably, the higher the regional vulnerability score, the greater the risk of scaling or pipe wall damage, and the higher the proportion of pulse duration extension, in order to enhance the cleaning effect and control the risk; for example, for every 10% increase in regional vulnerability score, the cleaning agent concentration gradient increases by 15% and the pulse duration is extended by 25%; where pulse duration refers to the duration of a single pulse or the total pulse duration within the entire cleaning cycle when pulsed cleaning (such as high-pressure water jet pulse, chemical cleaning agent pulse injection, etc.) is used during the cleaning process;

[0095] Implement a cleaning operation chain generation strategy: sort operations in reverse order along the fluid flow direction, prioritizing those with a high scaling risk transmission intensity coefficient. Downstream node clusters exceeding the preset intensity threshold are formed into a cleaning priority sequence; if the priority sequence contains a key transmission hub node ID, its cleaning priority is raised to the highest level; at the same time, the cleaning resource allocation weight is adjusted according to the edge weight (the larger the edge weight, the more resources are allocated), and the edge weight connecting the high-risk isolated node cluster ID is given an additional 20% resource weight.

[0096] The final output is an adaptive cleaning protocol, which includes generating a cleaning operation chain with flow direction constraints, clearly marking the cleaning order and parameters of key transmission hub node IDs and high-risk isolated node cluster IDs; and dynamically updating the cleaning protocol to match the real-time scaling risk transmission network status to ensure the blocking of high-risk paths.

[0097] Example 2

[0098] Please see Figure 2 This invention provides an evaluation system based on boiler cleaning monitoring data, comprising:

[0099] Boiler cleaning assessment data analysis module: acquires boiler cleaning assessment dataset, extracts multi-physics coupled feature tensors, and constructs feature evolution trajectory of scaling process;

[0100] Scaling phase transition state vector generation module: Based on the characteristic evolution trajectory of the scaling process, dynamic convolution kernel group is used to identify the phase transition stage, and regional vulnerability is calculated through time-varying differential integral of thermal stress. Finally, the scaling phase transition state vector is generated by fusion.

[0101] Scaling risk transmission network construction module: Based on the topological dependency relationship of the scaling phase transition state vector, construct the scaling risk transmission network and output network structure data including node-edge relationships and risk transmission strength coefficients;

[0102] Cleaning strategy optimization protocol module: Based on the scaling risk transmission network, the three-level cleaning strategy is dynamically optimized to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

[0103] Example 3

[0104] This invention provides an evaluation device based on boiler cleaning monitoring data, including a processor, a memory, and a communication bus;

[0105] The memory stores a computer-readable program that can be executed by the processor;

[0106] The communication bus enables communication between the processor and the memory;

[0107] When the processor executes the computer-readable program, it performs an evaluation method based on boiler cleaning monitoring data as described in this invention.

[0108] Furthermore, the formulas mentioned above are all dimensionless calculations, derived from software simulation using a large amount of collected data to approximate the real situation. The proportionality coefficients in the formulas and the preset thresholds in the analysis process are set by those skilled in the art based on the actual situation or obtained through large-scale data simulation. The magnitude of the proportionality coefficient is a specific value obtained by quantifying each parameter to facilitate subsequent comparison. The magnitude of the proportionality coefficient depends on the amount of sample data and the processing coefficients initially set by those skilled in the art for each set of sample data. As long as it does not affect the proportional relationship between the parameter and the quantified value, it is acceptable.

[0109] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. The focus of each embodiment is on its differences from other embodiments. In particular, the apparatus embodiments are described simply because they are fundamentally based on the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0110] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0111] Those skilled in the art will understand that embodiments of the present invention can be provided as 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.

[0112] 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, and 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 processor, 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 illustrations and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] 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.

[0114] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device 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.

[0115] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0116] In conclusion, the above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An evaluation method based on boiler cleaning monitoring data, characterized in that: S1. Obtain the boiler cleaning assessment dataset, extract the multi-physics coupled feature tensor, and construct the feature evolution trajectory of the scaling process; S2. Based on the characteristic evolution trajectory of the scaling process, a dynamic convolution kernel group is used to identify the phase transition stage. At the same time, the vulnerability of the region is calculated by the time-varying differential integral of thermal stress, and finally the scaling phase transition state vector is generated by fusion. S3. Based on the topological dependency relationship of the scaling phase change state vector, construct the scaling risk transmission network and output the network structure data including node-edge relationships and risk transmission strength coefficients; S4. Dynamically optimize the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

2. The evaluation method based on boiler cleaning monitoring data according to claim 1, characterized in that: The process of obtaining a boiler cleaning assessment dataset, extracting multiphysics coupled feature tensors, and constructing the feature evolution trajectory of the scaling process includes: Boiler cleaning assessment datasets were obtained through multiple sources, including pipe wall temperature distribution data, pipe wall strain data, scale adhesion strength data, fluid velocity data, turbulent kinetic energy data, ion deposition related data, scale thickness data, and time information. From the boiler cleaning assessment dataset, multi-physics interaction features are simultaneously analyzed and fused: by calculating the spatial mapping relationship between pipe wall thermal stress distribution and real-time scale adhesion intensity, thermo-mechanical coupling features, i.e., thermo-mechanical coupling covariance coefficients, are obtained; based on the dynamic inhibition effect of quantified fluid shear stress field and ion deposition rate, fluidization coupling features, i.e., fluidization coupling inverse correlation factors, are obtained; by capturing the spatial transmission delay effect of scale growth rate in different boiler areas, spatiotemporal correlation features, i.e., spatiotemporal correlation time lag mutual information, are obtained; the extracted multi-physics interaction features are integrated according to the spatiotemporal dimension to form a multi-physics coupling feature tensor. The feature evolution trajectory of the scaling process is constructed based on the continuous change of the multi-physics field coupling feature tensor during the cleaning cycle.

3. The evaluation method based on boiler cleaning monitoring data according to claim 1, characterized in that: The process of identifying phase transition stages using dynamic convolutional kernel groups, calculating regional vulnerability through time-varying differential-integral methods of thermal stress, and finally fusing these methods to generate a fouling phase transition state vector includes: Based on the input scaling process characteristics and evolution trajectory, the thermo-mechanical coupling covariance coefficient, the fluidization coupling inverse correlation factor, and the spatiotemporal correlation time lag mutual information are extracted. A dynamic convolutional kernel group is constructed for identifying the scaling phase transition stage: a mutation detection kernel is used to scan the steep drop point of the thermo-mechanical coupling covariance coefficient, a continuously growing kernel is used to capture the monotonically increasing interval of the fluidization coupling inverse correlation factor, and an oscillating attenuation kernel is used to analyze the periodic fluctuations of the time lag mutual information of the spatiotemporal correlation; the phase transition event marker sequence of each region in the time domain is output by sliding the feature evolution trajectory of the convolutional kernel group; the scaling phase transition stage index is quantified according to the density distribution and duration of the phase transition event marker sequence and encoded using numbers 1 to 4; Simultaneously, the regional vulnerability score is calculated based on the thermo-mechanical coupling covariance coefficient: the absolute value of the time-varying differential of the thermo-mechanical coupling covariance coefficient is integrated over time with its peak value. ， In the formula, Score the vulnerability of the region; Time-varying weighting factor; Spatial concentration weighting factor; The thermo-mechanical coupling covariance coefficient; This represents the peak value of the thermo-mechanical coupling covariance coefficient within the current time window; This is the integration time interval, corresponding to the boiler operating time in the time information. This represents the absolute value of the instantaneous rate of change of the thermo-mechanical coupling covariance coefficient; By integrating the scaling phase transition stage index and the regional vulnerability score, a scaling phase transition state vector is generated.

4. The evaluation method based on boiler cleaning monitoring data according to claim 1, characterized in that: The process of constructing a scaling risk transmission network based on the topological dependency of scaling phase transition state vectors and outputting network structure data containing node-edge relationships and risk transmission strength coefficients includes: The boiler tube bundle system is abstracted as a dynamic weighted directed network, including node definition and directed edge definition: each tube segment unit is an independent node, and the scaling phase change stage index and regional vulnerability score are used as node attributes; adjacent nodes are connected according to the actual fluid flow path, and the weight of the directed edge is dynamically calculated by the product of the extracted fluidization coupling inverse correlation factor and the tube bundle topology spacing. Perform downstream risk propagation at each node, i.e., iterative calculation of risk transmission intensity: calculate the scaling risk transmission intensity coefficient at each node. Along the directed edge direction, transfer the scaling risk transmission intensity coefficient of the current node. The values ​​are scaled according to edge weights and then superimposed onto downstream nodes: Downstream node risk value = Initial risk value + Scaling risk transmission strength coefficient × Directed edge weights; Set global convergence conditions, traverse all nodes to complete global iterative calculations, and update the risk values ​​of each node in the network; Key region identification is performed based on node and edge attributes, including key transmission hub node identification and high-risk isolated node cluster identification. For key transmission hub node identification, the in-degree of each node is calculated. If the in-degree of a node is greater than a preset node connection density threshold and its risk value is greater than the average risk value, then the node is marked as a key transmission hub node. For high-risk isolated node cluster identification, the out-degree of each node is calculated. If the out-degree of a node is zero and its regional vulnerability score is greater than a preset vulnerability threshold, then the node is marked as a candidate isolated node. Isolated node clusters are generated through spatial clustering. The regional vulnerability score of the cluster is calculated using a formula. If the regional vulnerability score of the cluster is greater than a preset cluster vulnerability threshold, then the isolated node cluster is marked as a high-risk isolated node cluster. Finally, a scaling risk transmission network is generated, and the network structure data is output, including a set of node attributes, a set of directed edge attributes, and a list of key area identifiers. The set of node attributes includes the scaling phase transition stage index, regional vulnerability score, and risk value for each node; the set of directed edge attributes includes the real-time weight of each edge and the scaling risk transmission intensity coefficient. Values; the list of key area identifiers includes key transmission hub node IDs and high-risk isolated node cluster IDs.

5. The evaluation method based on boiler cleaning monitoring data according to claim 4, characterized in that: The process of dynamically optimizing the three-level cleaning strategy based on the scaling risk transmission network to form an adaptive cleaning protocol includes: The output of the scaling risk transmission network is analyzed, and key parameters are extracted: the scaling phase transition stage index, regional vulnerability score, and risk value of each node are obtained based on node attributes; the directed edge weights and scaling risk transmission strength coefficients are obtained based on edge attributes. Simultaneously, load the set of key transmission hub node IDs and the set of high-risk isolated node cluster IDs from the list of key area identifiers as trigger conditions for subsequent cleaning strategies. Based on the extracted key parameters, a three-level cleaning strategy is dynamically generated: a cleaning mode selection strategy is implemented: for the nodes corresponding to the key transmission hub node IDs, the cleaning mode is assigned according to their scaling phase change stage index. Implement a dynamic adjustment strategy for cleaning parameters: For nodes in the high-risk isolated node cluster ID, dynamically allocate cleaning agent concentration gradient and pulse duration according to their regional vulnerability score; Implement a cleaning operation chain generation strategy: sort operations in reverse order along the fluid flow direction, prioritizing those with a high scaling risk transmission intensity coefficient. Downstream node clusters exceeding a preset intensity threshold are formed into a cleaning priority sequence; if the priority sequence contains a key transmission hub node ID, its cleaning priority is raised to the highest level; at the same time, the cleaning resource allocation weight is adjusted according to the edge weight. The final output is an adaptive cleaning protocol, which includes generating a cleaning operation chain with flow direction constraints, clearly marking the cleaning order and parameters of key transmission hub node IDs and high-risk isolated node cluster IDs; and dynamically updating the cleaning protocol to match the real-time scaling risk transmission network status.

6. An evaluation system based on boiler cleaning monitoring data, characterized in that, The system, applied to the evaluation method based on boiler cleaning monitoring data as described in any one of claims 1-5, comprises: Boiler cleaning assessment data analysis module: acquires boiler cleaning assessment dataset, extracts multi-physics coupled feature tensors, and constructs feature evolution trajectory of scaling process; Scaling phase transition state vector generation module: Based on the characteristic evolution trajectory of the scaling process, dynamic convolution kernel group is used to identify the phase transition stage, and regional vulnerability is calculated through time-varying differential integral of thermal stress. Finally, the scaling phase transition state vector is generated by fusion. Scaling risk transmission network construction module: Based on the topological dependency relationship of the scaling phase transition state vector, construct the scaling risk transmission network and output network structure data including node-edge relationships and risk transmission strength coefficients; Cleaning strategy optimization protocol module: Based on the scaling risk transmission network, the three-level cleaning strategy is dynamically optimized to form an adaptive cleaning protocol; the three-level cleaning strategy includes a cleaning mode selection strategy, a cleaning parameter dynamic adjustment strategy, and a cleaning operation chain generation strategy.

7. An evaluation device based on boiler cleaning monitoring data, characterized in that: It is implemented based on the evaluation method based on boiler cleaning monitoring data as described in any one of claims 1-5, and includes a processor, a memory, and a communication bus; The memory stores a computer-readable program that can be executed by the processor; The communication bus enables communication between the processor and the memory; When the processor executes the computer-readable program, it implements the evaluation method based on boiler cleaning monitoring data as described in any one of claims 1-5.

Images