Boiler corrosion dynamic sensing and early warning method based on multi-source information fusion

Abstract

The invention discloses a boiler corrosion dynamic sensing and early warning method based on multi-source information fusion, and belongs to the field of industrial equipment health monitoring, and the early warning method specifically comprises the following steps: collecting and preprocessing multi-dimensional data of operation of each boiler, and integrating the processed multi-dimensional data to generate a corresponding state feature vector; according to the invention, refined expression of the corrosion risk can be realized, the capability of identifying the corrosion hidden danger in advance and the operation safety level are effectively improved, and the perspectiveness and accuracy of early warning are improved; meanwhile, real-time monitoring and early warning of boiler corrosion can be achieved, guarantee is provided for safe operation, it is guaranteed that the early warning effect is kept optimal along with changes of time and working conditions, and the accuracy of corrosion rate prediction is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial equipment health monitoring, and in particular to a boiler corrosion dynamic perception and early warning method based on multi-source information fusion. BACKGROUND

[0002] During the long-term operation of the boiler, due to high temperature, high pressure and medium corrosion and other factors, the metal parts are prone to corrosion, wear and structure deterioration, thereby affecting the safety and service life of the boiler. The traditional boiler corrosion monitoring method relies on periodic manual inspection or single sensor data, and has problems such as discontinuous data collection, limited monitoring dimension, low prediction accuracy, and lagging early warning. At the same time, the boiler operating environment is complex and changeable, and single source data is difficult to fully reflect the corrosion evolution process, making it difficult to identify corrosion risks in a timely manner and unable to effectively guide operation and maintenance and life management.

[0003] The existing boiler corrosion dynamic perception and early warning method cannot realize the fine and visual expression of corrosion risk, and the identification of potential hazards is often lagging. In addition, the existing boiler corrosion dynamic perception and early warning method usually relies on single or a few kinds of monitoring data, lacks deep fusion and real-time analysis of multi-source heterogeneous information, and is difficult to adaptively adjust with the change of working conditions, resulting in large deviation in corrosion rate prediction. Therefore, we propose a boiler corrosion dynamic perception and early warning method based on multi-source information fusion. SUMMARY

[0004] The present application relates to the field of industrial equipment health monitoring, and in particular to a boiler corrosion dynamic perception and early warning method based on multi-source information fusion.

[0005] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows: A boiler corrosion dynamic perception and early warning method based on multi-source information fusion, the specific steps of the early warning method are as follows: Ⅰ: Collect and pre-process the multi-dimensional data of each boiler operation, and integrate the processed multi-dimensional data to generate corresponding state feature vectors; Ⅱ: Map each state feature vector to the corresponding risk node, analyze the causal relationship between each risk node, and construct the corresponding corrosion risk propagation chain; Ⅲ: According to each state feature vector and the corresponding corrosion risk propagation chain, predict the corrosion rate of the boiler under the current working condition, and calculate the corresponding remaining service life evaluation value; Ⅳ: Based on the historical corrosion data of each boiler, dynamically optimize the corresponding early warning threshold boundary, and adjust the corresponding early warning trigger condition in combination with the current operation state of each boiler; Ⅴ: Real-time monitor the operation state of each corrosion risk propagation chain, and evaluate and divide the risk level in combination with the optimized early warning threshold boundary.

[0006] As a further scheme of the present application, the specific steps of generating the corresponding state feature vector in step I are as follows: S1.1: Collect multi-dimensional data of each boiler operation, filter noise in each dimension data by using smoothing filtering, then supplement missing values of each dimension data by using linear interpolation method, and then map each dimension data to the interval [0, 1], classify each dimension data according to the boiler corrosion influencing factors, and form the corresponding influencing factor set; S1.2: Real-time collect the quality indicators of the processed multi-dimensional data, and synchronously collect the dynamic change data of the corresponding boiler operation environment, then continuously capture the reliability change trend of each dimension data and the dynamic characteristics of the external environment through real-time data sampling and statistical analysis, and based on the known field knowledge and historical data of boiler corrosion monitoring, establish the corrosion monitoring basic rules to determine the weight influence coefficient corresponding to different data quality indicators and the influence degree of environmental change on the correlation of each dimension data, and then set the initial threshold and gradient range of weight adjustment combined with the historical fusion effect feedback; S1.3: According to the real-time monitored quality indicators and environmental change, and comparing with the initialized weight adjustment corrosion monitoring basic rules, real-time calculate the fusion weight of each dimension data, and adjust the weight proportion of each dimension data according to the weight adjustment rules, and then real-time cover the corresponding historical weight value with the current weight adjustment result of each dimension data, to form the updated weight set; S1.4: Select the appropriate weighted fusion algorithm for integrated processing of each dimension data, if it is numerical data or sensor data, use weighted summation method for fusion calculation, and calculate the weight according to the real-time weight of each dimension data; if it is structured data or environmental feature data, use feature level weighted fusion method, integrate the weight into the feature extraction process, strengthen the feature contribution of the data whose weight exceeds the preset threshold, and then integrate each dimension data into the corresponding state feature vector.

[0007] As a further scheme of the present application, the specific steps of constructing the corresponding corrosion risk propagation chain in step II are as follows: S2.1: Based on the influencing factors and state feature vectors of each boiler corrosion, determine the classification standard of risk nodes, and select temperature anomaly, medium corrosion component exceeding standard, metal loss and humidity exceeding standard as key indicators, and then define each indicator as an independent corrosion risk node type, and then map each state feature vector to the corresponding risk node according to the classification standard and type of risk node, establish one-to-one or many-to-many association relationship between state feature vector and risk node, convert the continuously changing feature data into quantified indicators of node attributes, and get the specific attribute value of each node; S2.2: Based on each state feature vector, attribute quantification processing is performed on each mapped risk node to determine the attribute dimension of each risk node, and through statistical analysis and numerical conversion method, the qualitative information of each state feature vector is converted into quantitative value, and the continuously changing state feature vector is converted into quantized index of risk node attribute, and finally the attribute of each risk node is obtained; S2.3: Based on the physical and chemical principles of boiler corrosion and historical operation data, the potential causal relationship between each risk node is combed, and the interaction mechanism of the indexes corresponding to different risk nodes in the corrosion process is analyzed. If the medium corrosion component exceeds the standard, the risk node will cause the metal loss node value to rise, and if the temperature exceeds the preset range, the risk node will accelerate the associated process. Then, by calculating the correlation coefficient and time sequence correlation of each risk node attribute value, the causal associated node pairs meeting the preset requirements are preliminarily screened out. S2.4: An association analysis model is constructed and trained, and the correlation strength values between each risk node are calculated using the trained association analysis model, and are sorted from high to low. Then, taking the risk node with correlation strength value exceeding the preset threshold as the starting point, the upstream and downstream risk nodes are connected in turn according to the order of correlation strength value, forming multiple initial corrosion risk propagation paths, and then all corrosion risk propagation paths are integrated to construct the corresponding corrosion risk propagation chain.

[0008] As a further scheme of the present application, the specific training steps of the association analysis model in S2.4 are as follows: P1.1: Collect risk node data in different boiler historical corrosion cases, and organize the risk node set corresponding to each corrosion case into a node feature matrix. Then, according to the real causal relationship between each risk node, an adjacency matrix is constructed, and the correlation strength between the risk node pairs is quantified by numerical value to form a graph structure data sample. Then, the training set, validation set and test set are divided according to the preset proportion; P1.2: Construct an association analysis model, and input the graph structure data in the training set into the association analysis model in batches. The association analysis model calculates the correlation strength prediction value of each node pair through forward propagation. Then, based on the cross-entropy loss function, the error between the prediction value and the true value is calculated. Then, the error signal is transmitted back along the network layers of the association analysis model through the back propagation algorithm, and the weights and bias terms of each layer of the association analysis model are updated by using the gradient descent algorithm; P1.3: Set the number of training iterations and batch size. Record the loss value after completing each batch training. At the same time, evaluate the performance of the association analysis model with the validation set after completing each iteration, and monitor the training set loss and validation set loss change trend during the training process. If the validation set loss rises continuously for many rounds or the training reaches the preset training rounds, stop the training, and complete the training of the association analysis model.

[0009] As a further scheme of the present application, the specific steps of predicting the corrosion rate of the boiler under the current working condition in step III are as follows: S3.1: Collect the historical operation data and corrosion data of different types of boilers as source domain data, and synchronously collect the operation data of the current boilers as target domain data, respectively construct the source domain data set and the target domain data set, then extract each corrosion rate feature from the source domain data set, and based on the extracted features, construct a corrosion prediction model, select a regression algorithm as the core architecture of the corrosion prediction model, take each feature as the input, and take the corrosion rate as the output to train the corrosion prediction model, and adjust the model parameters through iterative optimization; S3.2: Compare the feature distribution difference of the source domain data and the target domain data by using statistical analysis method, calculate the feature mean and variance of the two types of data, analyze the reason of the difference, determine the difference between the current boiler and the historical boiler in all aspects, and set an adaptive loss function, and dynamically adjust the parameters of the corrosion prediction model using the target domain data; S3.3: Divide the target domain data into training set and validation set according to the proportion, input the training set into the corrosion prediction model for secondary training, update the model parameters by using the minimum batch iteration method, and focus on optimizing the parameter weight which is strongly related to the corrosion feature of the current boiler, and evaluate the corrosion rate prediction accuracy of the model after each fine-tuning using the validation set, if the prediction accuracy does not reach the preset standard, continue to adjust the learning rate and iteration times, until the prediction performance of the corrosion prediction model reaches the preset expectation; S3.4: Take each state feature vector and the corresponding corrosion risk propagation chain as input, and input them into the optimized corrosion prediction model, the corrosion prediction model outputs the corrosion rate prediction value of the current boiler under the existing operating condition by real-time calculation, and continuously receives the updated state feature vector and the corrosion risk propagation chain, and updates the corrosion rate prediction result in real time; S3.5: Collect the design life parameters and the corrosion amount data corresponding to the running time of each boiler, and combine the predicted corrosion rate prediction result, calculate the total corrosion amount of each boiler by using the cumulative corrosion amount calculation method, then combine the allowed maximum corrosion amount corresponding to each design life parameter to obtain the corresponding remaining allowed corrosion amount, and then according to each corrosion rate prediction result, calculate the use time corresponding to each remaining allowed corrosion amount, and finally output the remaining service life evaluation value of each boiler.

[0010] As a further scheme of the present application, the specific steps of dynamically optimizing the corresponding early warning threshold boundary in step IV are as follows: S4.1: Collect real-time operation status data and historical corrosion data of each boiler during long-term operation, and classify and organize them according to data attributes, dividing them into operation status feature set, corrosion status feature set and early warning result dataset. At the same time, divide each type of dataset into training sample set, verification sample set and test sample set according to preset ratio, and build an early warning threshold optimization sample library. S4.2: Using the operation state feature set, corrosion state feature set and early warning result dataset as environmental state variables, set the corresponding environmental state space, quantify the current operation scenario and early warning status of each boiler, set the action space of the agent, determine the adjustment range and adjustment step size of the early warning threshold boundary, then construct the reward function, select the reinforcement learning algorithm that is suitable for the threshold optimization task, and construct the threshold adjustment model. S4.3: Input the training sample set in the early warning threshold optimization sample library into the threshold adjustment model in batches. The agent selects the threshold adjustment action based on the current environmental state. At the same time, it calculates the loss value using the policy gradient method, updates the model parameters through the backpropagation algorithm, and optimizes the action selection strategy. After each iteration, the model performance is evaluated using the validation sample set until training is complete. Then, the operation status data and corrosion status data of each boiler are received in real time and input into the threshold adjustment model. The threshold adjustment model outputs the current optimal early warning threshold boundary adjustment scheme. S4.4: Dynamically adjust the corresponding early warning thresholds according to the early warning threshold boundary adjustment scheme, while continuously collecting new operating data and early warning feedback results from each boiler, and supplementing the early warning threshold optimization sample library with new data. Optimize model parameters using the newly added samples, then establish a multi-dimensional evaluation system, set different evaluation indicators, and then use the test sample set and actual operating data of each boiler to comprehensively verify the optimized early warning thresholds. Analyze the degree of improvement in early warning effect after the optimization of each early warning threshold. If the degree of improvement meets the preset standard, maintain the current optimization strategy; otherwise, retrain the threshold adjustment model until the preset standard is reached.

[0011] As a further aspect of the present invention, the specific calculation formula for the fusion weights of the data in each dimension in real-time calculation as described in S1.3 is as follows: ; Indicates the first Data in each dimension at time Real-time fusion weights; Indicates the first Initial weights for each dimension of data; The weighting coefficients representing the influence of data quality indicators; Indicates the first Data in each dimension at time Quality rating; a weight adjustment coefficient representing an environmental impact; a quality score of the mth dimension data at the time point t affected by the environmental change coefficient; The specific calculation formula of the fusion calculation in S1.4 using the weighted summation is as follows: , wherein, indicates the fusion result of the numerical data; , wherein, indicates the number of dimensions of the numerical data participating in the fusion; , wherein, indicates the standardized value of the mth dimension data at the time point t; The specific calculation formula of the fusion calculation in S1.4 using the feature-level weighted fusion method is as follows: , wherein, indicates the mth fusion feature; , wherein, indicates the number of dimension data related to the mth feature; , wherein, indicates the contribution value of the mth dimension data at the time point t to the mth feature. Compared with the prior art, the present application has the beneficial effects that:

[0012] ​​​​​​​​​​​​​1、The present application collects multi-dimensional data of each boiler operation, and carries out smoothing filtering denoising, linear interpolation for missing data, and then maps to the [0, 1] interval, and then classifies each dimension data according to the boiler corrosion influencing factors to form the corresponding influence factor set, simultaneously real-time obtains the quality index of each dimension data and the dynamic information of each boiler operation environment, and then combines the field knowledge and historical data of boiler corrosion monitoring to establish the corrosion monitoring basic rules, and dynamically calculates and adjusts the fusion weight of each dimension data, and then adopts weighted summation or feature level weighting according to the type of each dimension data to form the corresponding state feature vector, and then determines the classification standard of the risk node based on the influencing factors and state feature vector of each boiler corrosion, takes temperature anomaly, medium corrosion component exceeding standard, metal loss and humidity exceeding standard as the key index, and maps each state feature vector to the quantitative attribute of each risk node, and then analyzes the causal relationship between each risk node according to the physical and chemical mechanism of boiler corrosion and historical data, selects the correlation node pair through correlation coefficient and time sequence correlation degree analysis, constructs and trains the correlation analysis model, calculates the correlation strength value between each risk node, forms multiple corrosion risk propagation paths, and finally constructs the corresponding corrosion risk propagation chain, realizes the dynamic identification and propagation analysis of the boiler corrosion risk, can realize the fine expression of the corrosion risk, effectively improves the advance identification ability and operation safety level of the corrosion hidden danger, and improves the foresight and accuracy of the early warning.

[0013] The application collects different data boiler historical operation and corrosion data as source domain data, synchronously collects current boiler operation data as target domain data, constructs a corrosion prediction model, extracts each corrosion rate feature, trains an initial model by using a regression algorithm, compares the feature distribution difference of the source domain data and the target domain data, determines the difference between the current boiler and the historical boiler, sets an adaptive loss function, dynamically adjusts the corrosion prediction model parameters by using the target domain data, optimizes through secondary training and verification, until the corrosion prediction model prediction performance reaches the preset expectation, then inputs each state feature vector and the corresponding corrosion risk propagation chain into the optimized model, real-time outputs the current corrosion rate of each boiler and dynamically updates, at the same time, combines the corrosion amount data corresponding to the boiler design life and the running time, calculates the remaining allowable corrosion amount through the cumulative corrosion amount, calculates and outputs the corresponding remaining service life evaluation value according to the predicted corrosion rate result, collects the long-term real-time operation state and historical corrosion data of each boiler, classifies and arranges them into three types of feature sets and divides the sample set, constructs the early warning threshold optimization sample library, establishes the early warning threshold adaptive adjustment model based on reinforcement learning, optimizes the threshold boundary through the reward function and policy iteration, the model dynamically outputs the optimal early warning threshold under the real-time data driving, and the sample library and model parameters are continuously updated combined with the early warning threshold feedback, finally the threshold optimization effect is verified through the multi-dimensional evaluation system, which can realize the real-time monitoring and early warning of boiler corrosion, provides protection for safe operation, ensures that the early warning effect remains optimal with time and working condition changes, and significantly improves the accuracy of corrosion rate prediction. BRIEF DESCRIPTION OF DRAWINGS

[0014] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, illustrate the application together with the embodiments thereof, and explain the principles of the application, and do not constitute a limitation of the application.

[0015] Figure 1 A flow chart of a boiler corrosion dynamic perception and early warning method based on multi-source information fusion is proposed. DETAILED DESCRIPTION

[0016] REFERENCE Figure 1 A boiler corrosion dynamic perception and early warning method based on multi-source information fusion, the specific steps of the early warning method are as follows: Collect and pretreat the multi-dimensional data of each boiler operation, integrate the processed multi-dimensional data, and generate corresponding state feature vectors.

[0017] Specifically, multi-dimensional data on the operation of each boiler is collected, and noise in each dimension is filtered out using smoothing filters. Missing values ​​in each dimension are then supplemented using linear interpolation. The data is then mapped to the [0, 1] interval. Each dimension is categorized according to boiler corrosion influencing factors to form a corresponding influencing factor set. The quality indicators of the processed data are collected in real time, and dynamic changes in the boiler operating environment are simultaneously collected. Through real-time data sampling and statistical analysis, the reliability trends of each dimension and the dynamic characteristics of the external environment are continuously captured. Based on known domain knowledge and historical data of boiler corrosion monitoring, basic rules for corrosion monitoring are established to determine the weighting influence coefficients corresponding to different data quality indicators and the degree of influence of environmental changes on the correlation of each dimension's data. Finally, combined with historical fusion effect feedback, [the following steps are taken] to set... The initial threshold and gradient range for weight adjustment are determined based on real-time monitoring of quality indicators and environmental changes, and in accordance with the initial weight adjustment rules for corrosion monitoring. The fusion weights of each dimension of data are calculated in real-time, and the weight proportions of each dimension are adjusted according to the weight adjustment rules. The current weight adjustment results for each dimension are then overwritten with the corresponding historical weight values ​​in real-time, forming an updated weight set. A suitable weighted fusion algorithm is selected for each dimension of data for integration processing. For numerical data or sensor data, a weighted summation method is used for fusion calculation, and the weights are allocated according to the real-time weights of each dimension. For structured data or environmental feature data, a feature-level weighted fusion method is used, integrating the weights into the feature extraction process to enhance the feature contribution of data whose weights exceed the preset threshold. Finally, the data from each dimension are integrated into corresponding state feature vectors.

[0018] It should be further explained that the specific calculation formula for the fusion weights of data from each dimension in real time is as follows: ; Indicates the first Data in each dimension at time Real-time fusion weights; Indicates the first Initial weights for each dimension of data; The weighting coefficients representing the influence of data quality indicators; Indicates the first Data in each dimension at time Quality rating; The weighting adjustment coefficient representing the environmental impact; Indicates the first Data in each dimension at time The quality score is affected by environmental changes.

[0019] The specific calculation formula for the fusion calculation using the weighted summation method is as follows: ; In the formula, indicates the fusion result of numerical data; indicates the number of dimensions of numerical data participating in fusion; indicates the standardized value of the first dimensional data at time .

[0020] The specific calculation formula of the feature-level weighted fusion method is as follows: ; In the formula, indicates the first fusion feature; indicates the number of dimensional data related to the first feature; indicates the contribution value of the first dimensional data at time to the first feature.

[0021] Map each state feature vector to the corresponding risk node, analyze the causal relationship between each risk node, and build the corresponding corrosion risk propagation chain.

[0022] Specifically, based on the influencing factors and state feature vectors of each boiler corrosion, the classification standard of the risk node is determined, and temperature anomaly, medium corrosion component exceeding standard, metal loss amount and humidity exceeding standard are selected as key indicators, each indicator is defined as an independent corrosion risk node type, then each state feature vector is mapped to the corresponding risk node according to the classification standard and type of the risk node, a one-to-one or many-to-many association relationship between the state feature vector and the risk node is established, the continuously changing feature data is converted into the quantitative indicators of the node attributes, the specific attribute value of each node is obtained, based on each state feature vector, the attribute quantization processing is performed on each mapped risk node, the attribute dimension of each risk node is determined, and through statistical analysis and numerical conversion method, the qualitative information of each state feature vector is converted into quantitative values, the continuously changing state feature vector is converted into the quantitative indicators of the risk node attributes, finally the attributes of each risk node are obtained, based on the physical and chemical principles and historical operation data of the boiler corrosion, the potential causal relationship between the risk nodes is combed, the interaction mechanism of the indexes corresponding to different risk nodes in the corrosion process is analyzed, if the medium corrosion component exceeds the standard, the risk node will cause the metal loss amount node value to rise, if the temperature exceeds the preset range, the risk node will accelerate the associated process, then the correlation coefficient and time sequence correlation degree of the attribute values of each risk node are calculated, the causal associated node pairs meeting the preset requirements are preliminarily screened out, the associated analysis model is constructed and trained, and the correlation strength values between the risk nodes are calculated by using the trained associated analysis model, and are sorted from high to low, then the risk nodes with correlation strength values exceeding the preset threshold value are taken as the starting point, the upstream and downstream risk nodes are connected in turn according to the sorting of the correlation strength values, a plurality of initial corrosion risk propagation paths are formed, and all the corrosion risk propagation paths are integrated to construct the corresponding corrosion risk propagation chain.

[0023] It should be further explained that the specific training steps of the associated analysis model are as follows: The risk node data in different boiler historical corrosion cases is collected, and the risk node set corresponding to each corrosion case is arranged into a node feature matrix. According to the real causal relationship between the risk nodes, an adjacency matrix is constructed, and the correlation strength between the risk node pairs is quantified by a numerical value to form a graph structure data sample. Then, the training set, the validation set and the test set are divided according to a preset proportion, the correlation analysis model is constructed, and the graph structure data in the training set is input into the correlation analysis model in batches. The correlation analysis model calculates the correlation strength prediction value of each node pair by forward propagation. Then, the error between the prediction value and the true value is calculated based on the cross-entropy loss function. The error signal is transmitted in the reverse direction along the network layer of the correlation analysis model by the back propagation algorithm, and the weights and bias terms of each layer of the correlation analysis model are updated by the gradient descent algorithm. The number of training iterations and the batch size are set. After completing a batch of training, the loss value is recorded. After completing one round of iteration, the performance of the correlation analysis model is evaluated by using the validation set. The training set loss and the validation set loss are monitored during the training process. If the validation set loss increases continuously for multiple rounds or the training reaches the preset training round, the training is stopped, and the training of the correlation analysis model is completed.

[0024] Referring to Figure 1 A boiler corrosion dynamic perception and early warning method based on multi-source information fusion, the specific steps of the early warning method are as follows: According to the state feature vector and the corresponding corrosion risk propagation chain, the corrosion rate of the boiler under the current working condition is predicted, and the corresponding residual service life evaluation value is calculated.

[0025] Specifically, historical operation data and corrosion data of different types of boilers are collected as source domain data, and the current operation data of each boiler is synchronously collected as target domain data, and the source domain data set and the target domain data set are constructed, then the corrosion rate features are extracted from the source domain data set, and based on the extracted features, a corrosion prediction model is constructed, and a regression algorithm is selected as the core architecture of the corrosion prediction model, and the corrosion prediction model is trained with the features as input and the corrosion rate as output, and the model parameters are adjusted by iterative optimization, and the statistical analysis method is used to compare the feature distribution difference between the source domain data and the target domain data, and the feature mean and variance of the two types of data are calculated, the causes of the difference are analyzed, the differences between the current boiler and the historical boiler in all aspects are determined, and a self-adaptive loss function is set, the parameters of the corrosion prediction model are dynamically adjusted using the target domain data, the target domain data is divided into training set and validation set according to the proportion, the training set is input into the corrosion prediction model for secondary training, the model parameters are updated by using the minimum batch iteration method, and the parameter weight related to the corrosion feature of the current boiler is optimized, after each fine-tuning, the validation set is used to evaluate the corrosion rate prediction accuracy of the model, if the prediction accuracy does not reach the preset standard, the learning rate and the number of iterations are adjusted, until the prediction performance of the corrosion prediction model reaches the preset expectation, the state feature vector and the corresponding corrosion risk propagation chain are input into the optimized corrosion prediction model, the corrosion prediction model outputs the corrosion rate prediction value of the current boiler under the existing operating condition by real-time calculation, and the updated state feature vector and corrosion risk propagation chain are continuously received, the corrosion rate prediction result is updated in real time, the design life parameters and the corrosion amount data corresponding to the running time of each boiler are collected, and the predicted corrosion rate prediction result is combined, the cumulative corrosion amount calculation method is used to calculate the total corrosion amount of each boiler, and the allowed maximum corrosion amount corresponding to each design life parameter is combined to obtain the corresponding remaining allowed corrosion amount, then according to the corrosion rate prediction result, the use time corresponding to each remaining allowed corrosion amount is calculated, and finally the remaining service life evaluation value of each boiler is output.

[0026] Based on the historical corrosion data of each boiler, the corresponding warning threshold boundary is dynamically optimized, and the current operating state of each boiler is adjusted.

[0027] Specifically, real-time operation state data and historical corrosion data of each boiler during long-term operation are collected, classified and arranged according to data attributes, divided into operation state feature set, corrosion state feature set and early warning result data set, and each type of data set is divided into training sample set, verification sample set and test sample set according to a preset proportion, an early warning threshold optimization sample library is constructed, the operation state feature set, the corrosion state feature set and the early warning result data set are taken as environmental state variables, the corresponding environmental state space is set, the running scene and the early warning state of each boiler are quantitatively processed, the action space of the intelligent agent is set, the adjustment range and the adjustment step of the early warning threshold boundary are determined, then the reward function is constructed, the reinforcement learning algorithm suitable for threshold optimization task is selected, the threshold adjustment model is constructed, the training sample set in the early warning threshold optimization sample library is input into the threshold adjustment model in batches, the intelligent agent selects threshold adjustment action based on the current environmental state, the loss value is calculated by using the policy gradient method, the model parameters are updated by using the back propagation algorithm, the action selection strategy is optimized, the model performance is evaluated using the verification sample set after each iteration is completed, until the training is completed, then the operation state data and the corrosion state data of each boiler are received in real time and input into the threshold adjustment model, the threshold adjustment model outputs the current optimal early warning threshold boundary adjustment scheme, the corresponding early warning threshold is dynamically adjusted according to the early warning threshold boundary adjustment scheme, the new operation data and the early warning feedback results of each boiler are continuously collected, and the new data of each type is supplemented to the early warning threshold optimization sample library, the model parameters are optimized using the new samples, then a multi-dimensional evaluation system is established, different evaluation indexes are set, and the optimized early warning threshold is verified comprehensively using the test sample set and the actual operation data of each boiler, the improvement degree of the early warning effect after optimization of each early warning threshold is analyzed, if the improvement degree meets the preset standard, the current optimization strategy is maintained; otherwise, the threshold adjustment model is retrained until the preset standard is met.

[0028] The running state of each corrosion risk propagation chain is monitored in real time, and the risk level is evaluated and divided in combination with the optimized early warning threshold boundary.

Claims

1. A boiler corrosion dynamic sensing and early warning method based on multi-source information fusion, characterized in that, The early warning method comprises the following specific steps: I: Collect and pretreat multi-dimensional data of each boiler operation, integrate the processed multi-dimensional data, and generate corresponding state feature vectors; II: Map each state feature vector to a corresponding risk node, analyze the causal relationship between each risk node, and construct a corresponding corrosion risk propagation chain; III: According to each state feature vector and the corresponding corrosion risk propagation chain, predict the corrosion rate of the boiler under the current working condition, and calculate the corresponding residual service life evaluation value; IV: Based on the historical corrosion data of each boiler, dynamically optimize the corresponding early warning threshold boundary, and adjust the corresponding early warning trigger condition in combination with the current operation state of each boiler; V: Real-time monitor the operation state of each corrosion risk propagation chain, and evaluate and divide the risk level in combination with the optimized early warning threshold boundary.

2. The method according to claim 1, characterized in that, The specific steps of generating the corresponding state feature vector in step I are as follows: S1.1: Collect multi-dimensional data of each boiler operation, filter noise in each dimension data by using smoothing filter, then supplement missing values of each dimension data by using linear interpolation method, and then map each dimension data to [0, 1] interval, classify each dimension data according to boiler corrosion influencing factors, and form a corresponding influencing factor set; S1.2: Real-time collect quality indicators of the processed multi-dimensional data, and synchronously collect dynamic change data of the corresponding boiler operation environment, then continuously capture the reliability change trend of each dimension data and the dynamic characteristics of the external environment through real-time data sampling and statistical analysis, and based on the known field knowledge and historical data of boiler corrosion monitoring, establish corrosion monitoring basic rules to determine the weight influence coefficient corresponding to different data quality indicators and the influence degree of environmental change on the correlation of each dimension data, and then set the initial threshold and gradient range of weight adjustment combined with historical fusion effect feedback; S1.3: According to the real-time monitored quality indicators and environmental change, and according to the initialized weight adjustment corrosion monitoring basic rules, real-time calculate the fusion weight of each dimension data, and adjust the weight proportion of each dimension data according to the weight adjustment rule, and then real-time cover the historical weight value of each dimension data with the current weight adjustment result to form an updated weight set; S1.4: Select the appropriate weighted fusion algorithm for integration processing of each dimension data, if it is numerical data or sensor data, the weighted summation method is used for fusion calculation, and the weight is calculated according to the real-time weight of each dimension data; If it is structured data or environmental feature data, the feature level weighted fusion method is adopted, the weight is integrated into the feature extraction process, the feature contribution of the weight exceeding the preset threshold data is strengthened, and then each dimension data is integrated into a corresponding state feature vector.

3. The method according to claim 2, wherein, The specific steps of constructing the corresponding corrosion risk propagation chain in step II are as follows: S2.1: Based on the influencing factors and state feature vectors of the corrosion of each boiler, the classification standard of the risk node is determined, and temperature anomaly, medium corrosion component exceeding standard, metal loss and humidity exceeding standard are selected as key indicators. Each indicator is defined as an independent corrosion risk node type. Then, according to the classification standard and type of the risk node, each state feature vector is mapped to the corresponding risk node, and a one-to-one or many-to-many association relationship between the state feature vector and the risk node is established. The continuously changing feature data is converted into the quantified indicators of the node attributes, and the specific attribute values of each node are obtained; S2.2: Based on each state feature vector, the attribute quantification processing is performed on each mapped risk node to determine the attribute dimension of each risk node. Through statistical analysis and numerical conversion method, the qualitative information of each state feature vector is converted into quantitative value, and the continuously changing state feature vector is converted into the quantified indicators of the risk node attributes. Finally, the attributes of each risk node are obtained; S2.3: Based on the physical and chemical principles of boiler corrosion and historical operation data, the potential causal relationship between the risk nodes is analyzed, and the interaction mechanism of the indicators corresponding to different risk nodes in the corrosion process is analyzed. If the medium corrosion component exceeds the standard, the risk node will cause the metal loss node value to rise. If the temperature exceeds the preset range, the risk node will accelerate the associated process. Then, by calculating the correlation coefficient and time sequence correlation degree of the attribute values of each risk node, the causal association node pairs that meet the preset requirements are preliminarily selected; S2.4: The correlation analysis model is constructed and trained, and the correlation strength values between the risk nodes are calculated using the trained correlation analysis model. According to the order from high to low, the risk nodes with correlation strength values exceeding the preset threshold are taken as the starting point, and the upstream and downstream risk nodes are connected in turn according to the order of the correlation strength values, forming multiple initial corrosion risk propagation paths. Then, all the corrosion risk propagation paths are integrated to construct the corresponding corrosion risk propagation chain.

4. The method according to claim 3, characterized in that, The specific training steps of the correlation analysis model described in S2.4 are as follows: P1.1: Collect the risk node data in different boiler historical corrosion cases, and organize the risk node set corresponding to each corrosion case into a node feature matrix. Then, according to the real causal relationship between the risk nodes, an adjacency matrix is constructed, and the correlation strength between the risk node pairs is quantified by numerical value to form a graph structure data sample. Then, the training set, validation set and test set are divided according to the preset proportion; P1.2: The correlation analysis model is constructed, and the graph structure data in the training set is input into the correlation analysis model in batches. The correlation analysis model calculates the correlation strength prediction value of each node pair through forward propagation. Then, based on the cross-entropy loss function, the error between the prediction value and the true value is calculated. Then, the error signal is transmitted in the reverse direction along the network layers of the correlation analysis model through the back propagation algorithm, and the weights and bias terms of each layer of the correlation analysis model are updated by using the gradient descent algorithm. P1.3: Set the number of training iterations and batch size, record the loss value after each batch training, and evaluate the performance of the correlation analysis model with the validation set after each iteration, and monitor the training set loss and validation set loss trend during training, if the validation set loss rises continuously for many rounds or the training reaches the preset training rounds, stop training, complete the training of the correlation analysis model.

5. The method according to claim 3, characterized in that, The specific steps of predicting the corrosion rate of the boiler under the current working condition in step III are as follows: S3.1: Collect historical operation data and corrosion data of different types of boilers as source domain data, and synchronously collect operation data of the current boilers as target domain data, construct source domain data set and target domain data set respectively, then extract each corrosion rate feature from the source domain data set, and based on the extracted features, build a corrosion prediction model, select a regression algorithm as the core architecture of the corrosion prediction model, input each feature, and output the corrosion rate to train the corrosion prediction model by iteratively optimizing and adjusting the model parameters; S3.2: Use statistical analysis method to compare the feature distribution difference of source domain data and target domain data, and calculate the feature mean and variance of two kinds of data, analyze the reason of difference, determine the difference of current boiler and historical boiler in all aspects, and set adaptive loss function, and dynamically adjust the parameters of corrosion prediction model by using target domain data; S3.3: Divide the target domain data into training set and validation set according to the proportion, input the training set into the corrosion prediction model for secondary training, update the model parameters by using the minimum batch iteration method, and focus on optimizing the parameter weight which is strongly related to the corrosion feature of the current boiler, and evaluate the corrosion rate prediction accuracy of the model after each fine tuning by using the validation set, if the prediction accuracy does not reach the preset standard, continue to adjust the learning rate and iteration times, until the prediction performance of the corrosion prediction model reaches the preset expectation; S3.4: Input each state feature vector and corresponding corrosion risk propagation chain into the optimized corrosion prediction model, and the corrosion prediction model outputs the corrosion rate prediction value of each boiler under the existing operating condition by real-time calculation, and continuously receives the updated state feature vector and corrosion risk propagation chain, and updates the corrosion rate prediction result in real time; S3.5: Collect the design life parameters and corrosion amount data corresponding to the running time of each boiler, combine the predicted corrosion rate prediction result, calculate the total corrosion amount of each boiler by using the cumulative corrosion amount calculation method, and combine the corresponding allowed maximum corrosion amount of each design life parameter to obtain the corresponding remaining allowed corrosion amount, then according to each corrosion rate prediction result, calculate the use time corresponding to each remaining allowed corrosion amount, and finally output the remaining service life evaluation value of each boiler.

6. The method according to claim 1, characterized in that, The specific steps of dynamically optimizing the corresponding early warning threshold boundary in step IV are as follows: S4.1: Collect real-time operation state data and historical corrosion data of each boiler during long-term operation, classify and organize them according to data attributes, divide them into operation state feature set, corrosion state feature set and early warning result data set, and divide each data set into training sample set, validation sample set and test sample set according to a preset proportion, and construct early warning threshold optimization sample library; S4.2: Taking the operation state feature set, the corrosion state feature set and the early warning result data set as the environmental state variables, setting the corresponding environmental state space, quantifying the running scene and the early warning state of each boiler, setting the action space of the agent, determining the adjustment range and adjustment step of the early warning threshold boundary, then constructing the reward function, selecting the reinforcement learning algorithm suitable for threshold optimization task, and constructing the threshold adjustment model; S4.3: Input the training sample set in the early warning threshold optimization sample library into the threshold adjustment model in batches, the agent selects the threshold adjustment action based on the current environmental state, calculates the loss value by using the policy gradient method, updates the model parameters by using the back propagation algorithm, optimizes the action selection strategy, evaluates the model performance using the validation sample set after each iteration is completed, until the training is completed, then receives the operation state data and the corrosion state data of each boiler in real time, and inputs them into the threshold adjustment model, the threshold adjustment model outputs the current optimal early warning threshold boundary adjustment scheme; S4.4: According to the early warning threshold boundary adjustment scheme, dynamically adjust the corresponding early warning threshold, continuously collect new operation data and early warning feedback results of each boiler, and supplement new data to the early warning threshold optimization sample library, optimize the model parameters using the new samples, then establish a multi-dimensional evaluation system, set different evaluation indexes, and use the test sample set and the actual operation data of each boiler to verify the optimized early warning threshold, analyze the improvement degree of the early warning effect after optimizing each early warning threshold, if the improvement degree meets the preset standard, maintain the current optimization strategy; Otherwise, retrain the threshold adjustment model until the preset standard is met.

7. The method according to claim 2, wherein the method is characterized by, The specific calculation formula of the real-time calculation of the fusion weight of each dimension data in S1.3 is as follows: ； represents the real-time fusion weight of the first dimensional data at time ; represents the initial weight of the first dimensional data; represents the weight influence coefficient of the data quality index; represents the quality score of the first dimensional data at time ; represents the weight adjustment coefficient of the environmental influence; represents the influence coefficient of the quality score of the first dimensional data at time ; the environmental change; The specific calculation formula of the fusion calculation by using the weighted summation method in S1.4 is as follows: ； In the formula, This represents the fusion result of numerical data; Indicates the number of numerical data dimensions involved in the fusion; Indicates the first Data in each dimension at time The standardized value; The specific calculation formula of the feature-level weighted fusion method in S1.4 is as follows: ； In the formula, represents the first fusion feature; represents the number of dimension data related to the first feature; represents the contribution value of the first dimension data at the time point to the first feature.

Images