Equipment abnormity identification method and system in SCR (Selective Catalytic Reduction) external denitration process

Abstract

The invention relates to an equipment abnormity identification method in an SCR external denitration process, and the method comprises the steps: obtaining high-frequency sound wave signal data and temperature gradient difference data of catalyst equipment, and constructing a corresponding data set based on a public time axis; generating an equipment health index by using a temperature field matrix analysis module and an equipment health index calculation model in combination with a multi-modal feature fusion algorithm; further, a real-time equipment health index is calculated by monitoring a three-dimensional feature tensor and real-time data, and a probability value of an abnormal type is output based on an abnormal type probability model; and finally, analyzing the abnormal probability value by using a dynamic risk early warning judgment model, judging whether the equipment is abnormal or not, generating an equipment abnormality detection report, and pushing the equipment abnormality detection report to operation and maintenance personnel. According to the invention, the equipment abnormity can be accurately identified in real time, and the operation reliability and maintenance efficiency of the equipment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring of off - furnace denitration, and in particular to a method and system for identifying equipment anomalies in the SCR off - furnace denitration process. Background Art

[0002] Selective catalytic reduction (SCR) technology is a flue gas denitration technology. Under the action of a catalyst, it converts nitrogen oxides in the flue gas into nitrogen and water, thereby reducing harmful emissions. The SCR denitration technology is applicable to a variety of industrial fields, including power, metallurgy, building materials, light industry, chemical industry, etc. It is a widely used environmental protection technology that helps control nitrogen oxide emissions and promote ecological balance.

[0003] In the existing SCR off - furnace denitration process, equipment anomalies in the SCR process are usually detected by monitoring the data of each link. When the monitoring parameters of the equipment do not exceed the preset monitoring thresholds, it is difficult to detect equipment anomalies in a timely manner, reducing the accuracy of equipment anomaly identification in SCR treatment production, resulting in untimely equipment anomaly identification and further causing production losses. Therefore, improvement is needed. Summary of the Invention

[0004] In order to improve the accuracy of equipment anomaly identification in SCR treatment production, the present application provides a method and system for identifying equipment anomalies in the SCR off - furnace denitration process.

[0005] In the first aspect, the above - mentioned object of the present invention is achieved through the following technical solutions:

[0006] A method for identifying equipment anomalies in the SCR off - furnace denitration process, the method comprising the steps of:

[0007] Obtain the high - frequency acoustic wave signal data collected by the piezoelectric sensor array of the catalyst equipment, and record the high - frequency acoustic wave signal data based on a preset common time axis to construct a high - frequency acoustic wave signal data set corresponding to the catalyst equipment;

[0008] Obtain the temperature gradient difference data before and after the catalyst layer in the catalyst equipment and record the temperature gradient difference data based on the common time axis to construct a temperature gradient difference data set;

[0009] A preset temperature field matrix analysis module analyzes the temperature gradient difference data set based on a machine learning algorithm to generate a temperature field matrix;

[0010] Obtain the electromagnetic flowmeter data of each spray gun to record the instantaneous flow rate value data of ammonia, and generate the flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0011] The pre-set device health index calculation model processes and calculates the high-frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate data based on a multi-modal feature fusion algorithm to generate the device health index of the catalyst device;

[0012] Based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, a monitoring three-dimensional feature tensor is formed, where the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree of hydrodynamic simulation;

[0013] Obtain real-time temperature gradient difference data and instantaneous flow rate data to calculate the corresponding real-time device health index, and output the probability value of the abnormal type according to the pre-set abnormal type probability model, where the abnormal types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0014] Based on the pre-set dynamic risk warning judgment model, analyze the probability value of the abnormal type and compare it with the pre-set abnormal threshold, and judge whether there is a device abnormality. Generate a device abnormality detection report based on the abnormality judgment result and push it to the operation and maintenance personnel.

[0015] By adopting the above technical solutions, introducing the acoustic emission activity index can monitor the acoustic emission signals inside the catalyst device in real time, and timely detect the minute changes inside the device, such as catalyst wear, blockage, etc., so as to early warn of potential device failures. The present invention also considers the temperature gradient difference data. By analyzing the temperature changes before and after the catalyst layer, it can accurately judge the activity state of the catalyst and the thermodynamic performance of the device, and avoid device damage caused by temperature abnormalities. By introducing the spray gun flow distribution data, by calculating the flow deviation of each spray gun, problems such as spray gun scaling and blockage can be timely detected, ensuring the uniform distribution of ammonia gas and improving the denitrification efficiency. By analyzing the gas concentration data set, an oxygen concentration correction factor is generated, further optimizing the calculation of the device health index and making the evaluation result more accurate and reliable. In summary, through the fusion analysis of multi-dimensional data, the accurate identification and comprehensive evaluation of device abnormalities in the SCR out-of-furnace denitrification process are realized, effectively improving the operation reliability and maintenance efficiency of the device. Different from the single-parameter monitoring in the prior art, this solution monitors the device abnormalities through multi-dimensional comprehensive calculations, timely discovers and warns of potential device failures, and improves the accuracy and timeliness of device abnormality identification in SCR processing production.

[0016] In a preferred example of the present application, it can be further configured as follows: in the step where the pre-set device health index calculation model processes and calculates the high-frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate data based on a multi-modal feature fusion algorithm to generate the device health index of the catalyst device,

[0017] The acoustic emission activity index obtained by analyzing and calculating the high-frequency acoustic wave signal dataset using a pre-set acoustic emission activity analysis model; the calculation formula for the equipment health index H is as follows:

[0018]

[0019] α + β + γ = 1

[0020] where α is the acoustic emission signal weight value, β is the temperature gradient weight value, γ is the spray gun uniformity weight value, I AE is the acoustic emission activity index, ΔT is the temperature gradient difference data, T max is the pre-set catalyst tolerance limit value, Q i is the real-time flow data value of the i-th spray gun, is the calculated average flow rate of the spray gun.

[0021] By adopting the above technical solution, comprehensively considering the acoustic emission activity index, temperature gradient difference, spray gun flow distribution, and oxygen concentration correction factor, the equipment health index can comprehensively evaluate the operating status of the catalyst equipment. This multi-dimensional evaluation method can timely detect potential abnormalities of the equipment, improve the operating reliability and maintenance efficiency of the equipment, reduce the risk of equipment failure and maintenance costs. Compared with the existing method of judging whether there is an abnormality of the equipment through a single monitoring threshold, the equipment health index of this solution can be comprehensively considered from multiple indicators. When the monitoring parameters of the equipment do not exceed the pre-set monitoring threshold, the abnormal problems of the equipment can also be timely detected, improving the accuracy and timeliness of equipment abnormality identification in SCR processing production.

[0022] In a preferred example of this application, it can be further configured as follows: in the step of obtaining the acoustic emission activity index by analyzing and calculating the high-frequency acoustic wave signal dataset using a pre-set acoustic emission activity analysis model, the pre-set calculation formula for the acoustic emission activity index of the acoustic emission activity analysis model is as follows:

[0023]

[0024] where E f is the energy data value of each frequency band after wavelet packet decomposition, N events is the acoustic emission event count, t sample is the sampling duration.

[0025] By adopting the above technical solution, by real-time monitoring the acoustic emission activity index, the abnormal acoustic emission activities inside the catalyst equipment, such as catalyst blockage or wear, can be timely detected, providing an important basis for the early fault diagnosis of the equipment, helping to take maintenance measures in advance to avoid further damage to the equipment.

[0026] In a preferred example, the present application can be further configured as follows: after the step of generating the device health index of the catalyst device by processing and calculating the high-frequency acoustic wave signal data, the temperature gradient difference data, and the instantaneous flow rate data based on a preset device health index calculation model using a multimodal feature fusion algorithm, the following steps are included:

[0027] Obtain the NO x concentration C NOx and O 2 concentration C O2 , and construct a gas concentration data set based on the recording on the common time axis;

[0028] A preset oxygen concentration correction model analyzes the gas concentration data set based on a machine learning algorithm to generate an oxygen concentration correction factor, where the calculation formula for the device health index H is:

[0029]

[0030] α + β + γ + δ = 1,

[0031] where δ is the flue gas component weight value and ε is the oxygen concentration correction factor.

[0032] By adopting the above technical solution, by calculating the oxygen concentration correction factor, the concentration relationship between NO x and O 2 can be quantified, so as to more accurately evaluate the operating state of the catalyst device. This helps to optimize the operating conditions of the catalyst, improve the denitration efficiency, reduce emissions, and at the same time provide an important basis for the maintenance and management of the device.

[0033] In a preferred example, the present application can be further configured as follows: in the step of analyzing the probability value of the abnormal type based on a preset dynamic risk warning judgment model and comparing it with a preset abnormal threshold to determine whether there is a device abnormality, the following steps are included:

[0034] If there is an abnormal situation, a preset abnormal location analysis module analyzes the probability value of the abnormal type based on a machine learning algorithm to locate the abnormal area and generate abnormal area data;

[0035] Specifically, obtain the probability value of the output abnormal type and calculate the gradient data of the feature tensor according to a preset gradient calculation module;

[0036] A preset feature map extraction module calculates the monitored three-dimensional feature tensor through a spatial feature map generation algorithm to extract the corresponding feature map data;

[0037] The pre-set heat map generation module calculates the feature map data of the convolutional layer based on the weighted fusion algorithm to generate highlighted data;

[0038] Perform a highlighting operation on the feature map data based on the highlighted data to describe the abnormal position in the feature map, thereby locating the abnormal area and generating abnormal area data.

[0039] By adopting the above technical solution, the probability value of the output abnormal type is obtained, and the gradient data of the feature tensor is calculated by using the pre-set gradient calculation module, so as to accurately identify the abnormal type; through the feature map extraction module, the spatial feature map generation algorithm is used to calculate the monitored three-dimensional feature tensor, and the key feature map data is extracted, providing a basis for the location of the abnormal area; the heat map generation module calculates the feature map data of the convolutional layer based on the weighted fusion algorithm to generate highlighted data, intuitively displaying the abnormal area of the device; finally, a highlighting operation is performed on the feature map data based on the highlighted data to accurately locate the abnormal position and generate abnormal area data, facilitating the rapid response and processing of the operation and maintenance personnel.

[0040] In a preferred example of the present application, it can be further configured as follows: after the step of performing a highlighting operation on the feature map data based on the highlighted data to describe the abnormal position in the feature map, thereby locating the abnormal area and generating abnormal area data, the following steps are included:

[0041] The pre-set device topology graph construction model constructs a device topology graph based on the pre-set node data and edge weight data;

[0042] Obtain the nodes identified as abnormal sources, and calculate the estimated influence value of the abnormal nodes on the downstream nodes through the pre-set propagation influence calculation model;

[0043] Compare the estimated influence value with the pre-set influence threshold to determine whether the abnormal node is the main abnormal reason, and generate an abnormal node report according to the judgment result.

[0044] In a second aspect, the above object of the present invention of the present application is achieved through the following technical solutions:

[0045] An equipment abnormal identification device in an SCR off-site denitration process, the device includes: a high-frequency acoustic wave signal data set construction unit, configured to obtain high-frequency acoustic wave signal data collected by a piezoelectric sensor array of a catalyst device, and record the high-frequency acoustic wave signal data based on a pre-set common time axis to construct a high-frequency acoustic wave signal data set corresponding to the catalyst device;

[0046] A temperature gradient difference dataset construction unit is configured to obtain temperature gradient difference data before and after a catalyst layer in a catalyst device and record the temperature gradient difference data based on the common time axis to construct a temperature gradient difference dataset;

[0047] A temperature field matrix generation unit is configured to preset a temperature field matrix analysis module to analyze the temperature gradient difference dataset based on a machine learning algorithm to generate a temperature field matrix;

[0048] A spray gun flow distribution generation unit is configured to obtain electromagnetic flowmeter data of each spray gun to record instantaneous flow rate value data of ammonia, and generate a flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0049] An equipment health index generation unit is configured to preset an equipment health index calculation model to process and calculate the high-frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate value data based on a multi-modal feature fusion algorithm to generate an equipment health index of the catalyst device;

[0050] A monitoring three-dimensional feature tensor composition unit is configured to compose a monitoring three-dimensional feature tensor based on the equipment health index, the flow distribution of each spray gun, and the temperature field matrix, where the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree of fluid dynamics simulation;

[0051] An abnormal type probability value output unit is configured to obtain real-time temperature gradient difference data and instantaneous flow rate value data to calculate the corresponding real-time equipment health index, and output the probability value of the abnormal type according to a preset abnormal type probability model, where the abnormal types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0052] An equipment anomaly detection report generation unit is configured to analyze the probability value of the abnormal type based on a preset dynamic risk warning judgment model and compare it with a preset anomaly threshold, and determine whether there is an equipment anomaly situation, generate an equipment anomaly detection report based on the anomaly judgment result, and push it to the operation and maintenance personnel.

[0053] In a third aspect, the above object of the present application is achieved by the following technical solutions:

[0054] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the equipment anomaly recognition method in the above SCR off-furnace denitration process are implemented.

[0055] In a fourth aspect, the above object of the present application is achieved by the following technical solutions:

[0056] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the device anomaly recognition method in the above-mentioned SCR off-furnace denitration process are implemented.

[0057] In summary, the present application includes at least one of the following beneficial technical effects:

[0058] 1. By introducing the acoustic emission activity index, it is possible to monitor the acoustic emission signals inside the catalyst device in real time, and timely detect small changes inside the device, such as catalyst wear and blockage, etc., so as to early warn of potential device failures. The present invention also considers the temperature gradient difference data. By analyzing the temperature changes before and after the catalyst layer, it is possible to accurately judge the activity state of the catalyst and the thermodynamic performance of the device, and avoid device damage caused by abnormal temperature. By introducing the spray gun flow distribution data, by calculating the flow deviation of each spray gun, it is possible to timely detect problems such as fouling and blockage of the spray gun, ensure the uniform distribution of ammonia, and improve the denitration efficiency. Through the analysis of the gas concentration data set, an oxygen concentration correction factor is generated, which further optimizes the calculation of the device health index, making the evaluation result more accurate and reliable. In summary, through the fusion analysis of multi-dimensional data, the accurate identification and comprehensive evaluation of device anomalies in the SCR off-furnace denitration process are realized, effectively improving the operation reliability and maintenance efficiency of the device. Different from the single-parameter monitoring in the prior art, this solution monitors the device anomalies through multi-dimensional comprehensive calculations, timely discovers and warns of potential device failures, and improves the accuracy and timeliness of device anomaly identification in SCR treatment production;

[0059] 2. By comprehensively considering the acoustic emission activity index, temperature gradient difference, spray gun flow distribution, and oxygen concentration correction factor, the device health index can comprehensively evaluate the operation state of the catalyst device. This multi-dimensional evaluation method can timely detect potential device anomalies, improve the operation reliability and maintenance efficiency of the device, and reduce the risk of device failure and maintenance cost. Compared with the existing method of judging whether there is an anomaly in the device through a single monitoring threshold, the device health index of this solution can be considered comprehensively from multiple indicators. When the monitoring parameters of the device do not exceed the preset monitoring threshold, the anomaly problems of the device can also be timely detected, improving the accuracy and timeliness of device anomaly identification in SCR treatment production;

[0060] 3. By real-time monitoring the acoustic emission activity index, it is possible to timely detect abnormal acoustic emission activities inside the catalyst device, such as catalyst blockage or wear, etc., providing an important basis for the early fault diagnosis of the device, helping to take maintenance measures in advance to avoid further damage to the device;

[0061] 4. By obtaining the probability value of the output abnormal type and using a preset gradient calculation module to calculate the gradient data of the feature tensor, the abnormal type can be accurately identified. Through the feature map extraction module, the spatial feature map generation algorithm is used to calculate the monitored three-dimensional feature tensor, and the key feature map data is extracted, providing a basis for the positioning of the abnormal area. The heat map generation module calculates the feature map data of the convolutional layer based on the weighted fusion algorithm to generate high-brightness data, intuitively displaying the abnormal area of the device. Finally, based on the high-brightness data, the feature map data is highlighted to accurately locate the abnormal position and generate abnormal area data, facilitating the rapid response and processing of the operation and maintenance personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flowchart of a method for identifying equipment anomalies in an SCR off-furnace denitration process according to an embodiment of the present application;

[0063] Figure 2 is a schematic block diagram of a device for identifying equipment anomalies in an SCR off-furnace denitration process according to an embodiment of the present application;

[0064] Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present application.

[0065] Reference Numerals in the Drawings:

[0066] 1. High-frequency acoustic wave signal dataset construction unit; 2. Temperature gradient difference dataset construction unit; 3. Temperature field matrix generation unit; 4. Spray gun flow distribution generation unit; 5. Equipment health index generation unit; 6. Monitored three-dimensional feature tensor composition unit; 7. Probability value output unit of abnormal type; 8. Equipment anomaly detection report generation unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The present application will be further described in detail below with reference to the accompanying drawings.

[0068] In one embodiment, as Figure 1 shown, the present application discloses a method for identifying equipment anomalies in an SCR off-furnace denitration process, specifically including the following steps:

[0069] S10: Obtain the high-frequency acoustic wave signal data collected by the piezoelectric sensor array of the catalyst device, and record the high-frequency acoustic wave signal data based on a preset common time axis to construct a high-frequency acoustic wave signal dataset corresponding to the catalyst device;

[0070] Specifically, in the SCR off - furnace denitration process of this application embodiment, multiple piezoelectric sensors are installed inside the catalyst device. These sensors can collect high - frequency acoustic wave signals generated during the operation of the device in real time. In this application, the frequency range is 20 kHz - 1 MHz. These sensors can capture acoustic emission events (count rate, energy, rise time) generated by micro - cracks in the catalyst. It should be noted that in this application, interdigital electrodes are arranged on the catalyst surface, and an AC signal of 0.1 - 10 MHz is applied to measure the complex impedance to calculate the distribution of relaxation time (DRT).

[0071] By collecting high - frequency acoustic wave signal data in real time, abnormal acoustic wave changes inside the catalyst device, such as catalyst blockage or wear, can be detected in a timely manner, providing an important basis for subsequent anomaly recognition.

[0072] S20: Obtain the temperature gradient difference data before and after the catalyst layer in the catalyst device and record the temperature gradient difference data based on the common time axis to construct a temperature gradient difference data set;

[0073] Specifically, taking the "temperature gradient difference data" as an example, in the catalyst device, the temperature sensor in front of the catalyst layer measures the temperature as 300 °C at time point t1, and the temperature sensor behind the catalyst layer measures the temperature as 280 °C at time point t2, and the temperature gradient difference is 20 °C. These data are synchronously recorded through the common time axis to form a temperature gradient difference data set. By monitoring the temperature gradient difference before and after the catalyst layer in real time, abnormal temperature changes in the catalyst device, such as reduced catalyst activity or local overheating, can be detected in a timely manner.

[0074] S30: The pre - set temperature field matrix analysis module analyzes the temperature gradient difference data set based on the machine learning algorithm to generate a temperature field matrix;

[0075] Specifically, in this application embodiment, an infrared thermal imager arranged along the reactor cross - section obtains the temperature gradient ΔT before and after the catalyst layer. The temperature field matrix analysis module uses the machine learning algorithm to analyze the temperature gradient difference data set and generates a two - dimensional temperature field matrix. Suppose at a certain moment, the temperature field matrix shows that the temperature distribution inside the catalyst device is uniform, but at another moment, the matrix shows that the temperature of a certain area has increased abnormally, which may indicate that there is a local overheating problem in this area.

[0076] S40: Obtain the electromagnetic flowmeter data of each spray gun to record the instantaneous flow rate value data of ammonia, and generate the flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0077] Specifically, in the SCR system, there are multiple spray guns for injecting ammonia. Suppose the instantaneous flow rate value of spray gun A at time point t1 is 100 L / min, and the instantaneous flow rate value of spray gun B at time point t2 is 120 L / min. These data are recorded by electromagnetic flow meters, and a flow distribution map of each spray gun is generated. The data of the instantaneous flow rate values of each spray gun are recorded as Q i , where i represents the i-th spray gun. By real-time monitoring the instantaneous flow rate values of each spray gun, abnormal changes in the flow rate of the spray gun, such as spray gun blockage or uneven flow rate, can be detected in a timely manner, providing an important basis for subsequent anomaly identification.

[0078] S50: The pre-set device health index calculation model processes and calculates the high-frequency acoustic signal data, temperature gradient difference data, and instantaneous flow rate value data based on a multi-modal feature fusion algorithm to generate the device health index of the catalyst device;

[0079] Specifically, the device health index calculation model uses a multi-modal feature fusion algorithm to fuse and process the high-frequency acoustic signal data, temperature gradient difference data, and instantaneous flow rate value data.

[0080] S60: Based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, a monitoring three-dimensional feature tensor is formed. Among them, the spray gun is used as a node, and the node features include flow deviation and position coordinates. The edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics;

[0081] Specifically, based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, a three-dimensional feature tensor is constructed. Suppose the flow deviation of spray gun A is 10%, and the position coordinates are (1, 2), and the flow deviation of spray gun B is 5%, and the position coordinates are (3, 4). The edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics, forming a three-dimensional feature tensor.

[0082] By constructing the monitoring three-dimensional feature tensor, the operating state of the catalyst device can be comprehensively displayed, including the flow distribution, position information, and ammonia diffusion correlation degree of the spray gun, providing an important basis for subsequent anomaly identification.

[0083] S70: Obtain the real-time temperature gradient difference data and instantaneous flow rate value data to calculate the corresponding real-time device health index, and output the probability value of the anomaly type according to the pre-set anomaly type probability model;

[0084] Wherein the anomaly types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0085] Specifically, the device health index, the flow rate distribution of each spray gun, and the temperature field matrix are combined to form a three-dimensional feature tensor. Taking the spray gun as a node, the node features include flow rate deviation and position coordinates, and the edge weights are determined by the ammonia diffusion correlation degree simulated by fluid mechanics. Gated temporal convolution and spatial graph convolution are alternately used to extract features, and the probability of the abnormal type is output through a softmax classifier, including three types: catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit.

[0086] S80: Analyze the probability value of the abnormal type based on a pre-set dynamic risk warning judgment model and compare it with a pre-set abnormal threshold, and judge whether there is an equipment abnormality. Generate an equipment abnormality detection report based on the abnormality judgment result and push it to the operation and maintenance personnel;

[0087] Specifically, the dynamic risk warning judgment model analyzes the probability value of the abnormal type and compares it with a pre-set abnormal threshold. Suppose the probability value of catalyst blockage is 0.3, exceeding the pre-set threshold of 0.2. It is judged that there is a catalyst blockage abnormality, and an equipment abnormality detection report is generated based on the abnormality judgment result and pushed to the operation and maintenance personnel.

[0088] The solution of the present invention can, by introducing the acoustic emission activity index, monitor the acoustic emission signal inside the catalyst device in real time, and timely detect the minute changes inside the device, such as the wear and blockage of the catalyst, etc., so as to early warn of the potential faults of the device. The present invention also considers the temperature gradient difference data. By analyzing the temperature changes before and after the catalyst layer, it can accurately judge the activity state of the catalyst and the thermodynamic performance of the device, and avoid equipment damage caused by abnormal temperature. By introducing the spray gun flow rate distribution data, by calculating the flow rate deviation of each spray gun, problems such as scaling and blockage of the spray gun can be detected in time, ensuring the uniform distribution of ammonia gas and improving the denitration efficiency. By analyzing the gas concentration data set, an oxygen concentration correction factor is generated, further optimizing the calculation of the device health index and making the evaluation result more accurate and reliable. In summary, through the fusion analysis of multi-dimensional data, the accurate identification and comprehensive evaluation of equipment abnormalities in the SCR off-site denitration process are realized, effectively improving the operation reliability and maintenance efficiency of the equipment. Different from the monitoring of a single parameter in the prior art, this solution monitors the equipment abnormalities through multi-dimensional comprehensive calculations, and timely discovers and warns of the potential faults of the equipment, improving the accuracy and timeliness of equipment abnormality identification in SCR treatment production.

[0089] After the step S50: The pre-set device health index calculation model processes and calculates the high-frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate value data based on a multi-modal feature fusion algorithm to generate the device health index of the catalyst device, the following steps are included:

[0090] S51: The acoustic emission activity index obtained by analyzing and calculating the high-frequency acoustic wave signal dataset using a pre-set acoustic emission activity analysis model;

[0091] S52: Obtain the NO x concentration C NOx and O 2 concentration C O2 , and record based on the public time axis to construct a gas concentration dataset;

[0092] S53: The pre-set oxygen concentration correction model analyzes the gas concentration dataset based on a machine learning algorithm to generate an oxygen concentration correction factor, where the calculation formula for the equipment health index H is:

[0093]

[0094] α + β + γ + δ = 1,

[0095] where α is the acoustic emission signal weight value, β is the temperature gradient weight value, γ is the spray gun uniformity weight value, I AE is the acoustic emission activity index, ΔT is the temperature gradient difference data, T max is the pre-set catalyst tolerance limit value, Q i is the real-time flow data value of the i-th spray gun, is the calculated average flow value of the spray gun, δ is the flue gas composition weight value, and ε is the oxygen concentration correction factor.

[0096] Specifically, by monitoring the acoustic emission activity index in real time, abnormal acoustic emission activities inside the catalyst equipment, such as catalyst blockage or wear, can be detected in a timely manner, providing an important basis for the early fault diagnosis of the equipment, helping to take maintenance measures in advance, and avoiding further damage to the equipment; in the SCR system, NO x and O 2 concentration sensors are installed, and these sensors can monitor the NO x and O 2 concentrations at the inlet and outlet of the catalyst equipment in real time. Suppose at a certain moment, the NO x concentration C NOx is 200 ppm, and the O 2 concentration C O2 is 5%, and these data are synchronously recorded through the public time axis to construct a gas concentration dataset; by monitoring NO x and O 2Concentration, it can accurately grasp the denitration efficiency of the catalyst equipment and the change of oxygen content, provide data support for subsequent oxygen concentration correction, help optimize the operation state of the catalyst, and improve the denitration efficiency; the oxygen concentration correction model uses machine self-learning algorithm to analyze the gas concentration data set and generate the oxygen concentration correction factor ε. For example, according to historical data and real-time monitoring data, the model calculates that the current oxygen concentration correction factor ε is 1.1, indicating that the oxygen concentration needs to be appropriately corrected to optimize the operation conditions of the catalyst. It should be noted that in the embodiments of the present application, the oxygen concentration correction factor ε prevents division-by-zero errors, and the default value is 0.1% vol.

[0097] In summary, by comprehensively considering the acoustic emission activity index, temperature gradient difference, spray gun flow distribution, and oxygen concentration correction factor, the equipment health index H can comprehensively evaluate the operation state of the catalyst equipment. This multi-dimensional evaluation method can timely detect potential abnormalities of the equipment, improve the operation reliability and maintenance efficiency of the equipment, reduce the risk of equipment failure and maintenance cost. Compared with the existing method of judging whether there is an abnormality of the equipment through a single monitoring threshold, the equipment health index of this solution can be comprehensively considered from multiple indicators. When the monitoring parameters of the equipment do not exceed the preset monitoring threshold, the abnormal problems of the equipment can also be timely detected, improving the accuracy and timeliness of equipment abnormality identification in SCR treatment production.

[0098] In the step S51: the acoustic emission activity index calculated by the preset acoustic emission activity analysis model for analyzing the high-frequency acoustic wave signal data set, the acoustic emission activity analysis model is preset with the following acoustic emission activity index calculation formula:

[0099]

[0100] Where E f is the energy data value of each frequency band after wavelet packet decomposition, N events is the acoustic emission event count, and t sample is the sampling duration.

[0101] Suppose in the SCR off-furnace denitration process, multiple piezoelectric sensors are installed inside the catalyst equipment, and these sensors can collect high-frequency acoustic wave signals generated during the operation of the equipment in real time. The acoustic emission activity analysis model analyzes these signals and calculates the acoustic emission activity index I AW . For example, in the normal operation state, the acoustic emission activity index I AE is 1.2, indicating that the acoustic emission activity inside the equipment is relatively stable; while in abnormal situations such as catalyst blockage or wear, the acoustic emission activity index I AE may rise to 2.5, indicating a significant increase in the acoustic emission activity inside the equipment.

[0102] By real-time monitoring of the acoustic emission activity index, abnormal acoustic emission activities inside the catalyst equipment, such as catalyst blockage or wear, can be detected in a timely manner, providing an important basis for the early fault diagnosis of the equipment, helping to take maintenance measures in advance and avoid further damage to the equipment.

[0103] In step S80: Analyzing the probability value of the abnormal type based on a preset dynamic risk warning judgment model and comparing it with a preset abnormal threshold, and determining whether there is an equipment abnormality, the following steps are included:

[0104] S81: If there is an abnormal situation, the preset abnormal location analysis module analyzes the probability value of the abnormal type based on the machine learning algorithm to locate the abnormal area and generate abnormal area data;

[0105] Specifically, S811: Obtain the probability value of the output abnormal type and calculate the gradient data of the feature tensor according to the preset gradient calculation module;

[0106] S812: The preset feature map extraction module calculates the monitored three-dimensional feature tensor through the spatial feature map generation algorithm to extract the corresponding feature map data;

[0107] S813: The preset heat map generation module calculates the feature map data of the convolutional layer based on the weighted fusion algorithm to generate highlight data;

[0108] S814: Perform a highlighting operation on the feature map data based on the highlight data to describe the abnormal position in the feature map, thereby locating the abnormal area and generating abnormal area data.

[0109] For steps S811 - S814, in the embodiment of the present application, the gradient formula for calculating the gradient of the output category probability with respect to the input feature tensor is: c is the probability of the c-th abnormal category, is the k-th layer feature map, (i, j) is the spatial position, and Z is the preset normalization coefficient;

[0110] In the step of generating the heat map in S813, weighted fusion is performed on the feature map of the convolutional layer, and the highlighted area in the heat map corresponds to the physical position of the equipment (such as the spray gun number, catalyst module coordinates), thereby locating the abnormal area.

[0111] For example: When catalyst blockage is detected, the heat map shows that the activation value in the lower right quadrant of the reactor (coordinates X: 80 - 120 cm, Y: 40 - 60 cm) exceeds the threshold of 0.7, corresponding to catalyst module numbers C - 23 to C - 27.

[0112] The system automatically retrieves the historical ash cleaning records of this area. If the most recent ash cleaning time exceeds 72 hours, it is marked as a high-probability blockage area.

[0113] In the embodiment of the present application, the formula for automatically adjusting the ammonia injection distribution strategy is where μ is the adjustment rate factor and P is the abnormal probability value.

[0114] Regarding steps S811 - S814: The probability value of the output abnormal type is obtained through step S811, and the gradient data of the feature tensor is calculated using a preset gradient calculation module to accurately identify the abnormal type; through the feature map extraction module in step S812, the spatial feature map generation algorithm is used to calculate the monitored three-dimensional feature tensor, and the key feature map data is extracted to provide a basis for the positioning of the abnormal area; in step S813, the heat map generation module calculates the feature map data of the convolutional layer based on the weighted fusion algorithm to generate highlighted data, intuitively displaying the abnormal area of the device; finally, in step S814, the feature map data is highlighted based on the highlighted data to accurately locate the abnormal position and generate abnormal area data, facilitating the rapid response and processing of maintenance personnel. In addition, when calculating the gradient of the input feature tensor for the output category probability in the present invention, a normalization coefficient Z is introduced to optimize the gradient calculation, avoid the problems of gradient disappearance or explosion, improve the model training effect and abnormal recognition accuracy; at the same time, by calculating the gradient of the feature map of the k-th layer, multi-layer feature information is fused to comprehensively evaluate the operating state of the device, enhancing the robustness and accuracy of abnormal recognition; considering the influence of the spatial position (i, j), the model can accurately capture the abnormal changes of the device in space, further improving the accuracy of abnormal positioning.

[0115] S82: The pre-set device topology graph construction model constructs a device topology graph based on the pre-set node data and edge weight data;

[0116] In the present application, each spray gun, catalyst module, and sensor are used as independent nodes based on the ammonia diffusion path intensity simulated by fluid dynamics.

[0117] S83: Obtain the nodes identified as abnormal sources, and calculate the estimated influence value of the abnormal nodes on the downstream nodes through a pre-set propagation influence calculation model;

[0118] S84: Compare the estimated influence value with the pre-set influence threshold to determine whether the abnormal node is the main abnormal cause, and generate an abnormal node report according to the judgment result.

[0119] Specifically, when node i is identified as an abnormal source, calculate its influence weight on downstream node j. If the influence weight exceeds a preset threshold, then determine that node i is the main cause, and calculate the device health index corresponding to node i, as well as the result values of each monitoring item after weight processing, to determine the corresponding abnormal type, and perform abnormal area positioning based on the position of node i.

[0120] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0121] In one embodiment, there is provided a device for identifying equipment anomalies in an SCR off - furnace denitration process. The device for identifying equipment anomalies in the SCR off - furnace denitration process corresponds one - to - one with the method for identifying equipment anomalies in the SCR off - furnace denitration process in the above embodiments. As Figure 2 shown, the device for identifying equipment anomalies in the SCR off - furnace denitration process includes a high - frequency acoustic wave signal dataset construction unit 1, which is used to obtain the high - frequency acoustic wave signal data collected by the piezoelectric sensor array of the catalyst equipment, and record the high - frequency acoustic wave signal data based on a preset common time axis to construct a high - frequency acoustic wave signal dataset corresponding to the catalyst equipment;

[0122] A temperature gradient difference dataset construction unit 2, which is used to obtain the temperature gradient difference data before and after the catalyst layer in the catalyst equipment and record the temperature gradient difference data based on the common time axis to construct a temperature gradient difference dataset;

[0123] A temperature field matrix generation unit 3, which is used to preset a temperature field matrix analysis module to analyze the temperature gradient difference dataset based on a machine learning algorithm to generate a temperature field matrix;

[0124] A spray gun flow distribution generation unit 4, which is used to obtain the electromagnetic flowmeter data of each spray gun to record the instantaneous flow rate value data of ammonia, and generate the flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0125] A device health index generation unit 5, which is used to preset a device health index calculation model to process and calculate the high - frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate value data based on a multi - modal feature fusion algorithm to generate the device health index of the catalyst equipment;

[0126] A monitoring three - dimensional feature tensor composition unit 6, which is used to compose a monitoring three - dimensional feature tensor based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, where the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics;

[0127] The probability value output unit 7 of the abnormal type is configured to obtain real-time temperature gradient difference data and instantaneous flow rate value data, calculate the corresponding real-time device health index, and output the probability value of the abnormal type according to a preset abnormal type probability model, where the abnormal types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0128] The device anomaly detection report generation unit 8 is configured to analyze the probability value of the abnormal type based on a preset dynamic risk warning judgment model, compare it with a preset anomaly threshold, determine whether there is a device anomaly, generate a device anomaly detection report based on the anomaly judgment result, and push it to the operation and maintenance personnel.

[0129] For the specific limitations of the device anomaly recognition device in the SCR off-furnace denitration process, reference can be made to the limitations of the device anomaly recognition method in the SCR off-furnace denitration process described above, which will not be elaborated here. Each module in the above device anomaly recognition device in the SCR off-furnace denitration process can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or stored in the memory of the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0130] In one embodiment, an electronic device is provided. The electronic device can be a server, and its internal structure diagram can be as Figure 3 shown. The electronic device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store the database. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a device anomaly recognition method in the SCR off-furnace denitration process.

[0131] In one embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0132] Obtain the high-frequency acoustic wave signal data collected by the piezoelectric sensor array of the catalyst device, and record the high-frequency acoustic wave signal data based on a preset common time axis to construct a high-frequency acoustic wave signal dataset corresponding to the catalyst device;

[0133] Obtain the temperature gradient difference data before and after the catalyst layer in the catalyst device and record the temperature gradient difference data based on the preset common time axis to construct a temperature gradient difference data set;

[0134] A preset temperature field matrix analysis module analyzes the temperature gradient difference data set based on a machine learning algorithm to generate a temperature field matrix;

[0135] Obtain the electromagnetic flowmeter data of each spray gun to record the instantaneous flow rate value data of ammonia, and generate the flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0136] A preset device health index calculation model processes and calculates the high-frequency acoustic wave signal data, temperature gradient difference data, and instantaneous flow rate value data based on a multi-modal feature fusion algorithm to generate the device health index of the catalyst device;

[0137] Based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, a monitoring three-dimensional feature tensor is formed, where the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics;

[0138] Obtain the real-time temperature gradient difference data and instantaneous flow rate value data to calculate the corresponding real-time device health index, and output the probability value of the abnormal type according to a preset abnormal type probability model, where the abnormal types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0139] Based on a preset dynamic risk warning judgment model, analyze the probability value of the abnormal type and compare it with a preset abnormal threshold, and judge whether there is an equipment abnormality. Generate an equipment abnormality detection report based on the abnormality judgment result and push it to the operation and maintenance personnel.

[0140] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0141] Obtain the high-frequency acoustic wave signal data collected by the piezoelectric sensor array of the catalyst device, and record the high-frequency acoustic wave signal data based on a preset common time axis to construct a high-frequency acoustic wave signal data set corresponding to the catalyst device;

[0142] Obtain the temperature gradient difference data before and after the catalyst layer in the catalyst device and record the temperature gradient difference data based on the preset common time axis to construct a temperature gradient difference data set;

[0143] A preset temperature field matrix analysis module analyzes the temperature gradient difference data set based on a machine learning algorithm to generate a temperature field matrix;

[0144] Obtain the electromagnetic flowmeter data of each spray gun to record the instantaneous flow rate value data of ammonia, and generate the flow distribution of each spray gun according to the instantaneous flow rate value data of each spray gun;

[0145] The pre-set device health index calculation model processes and calculates the high-frequency acoustic wave signal data, temperature gradient difference data and instantaneous flow rate value data based on the multi-modal feature fusion algorithm to generate the device health index of the catalyst device;

[0146] Based on the device health index, the flow distribution of each spray gun, and the temperature field matrix, a monitoring three-dimensional feature tensor is formed, where the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics;

[0147] Obtain the real-time temperature gradient difference data and instantaneous flow rate value data to calculate the corresponding real-time device health index, and output the probability value of the abnormal type according to the pre-set abnormal type probability model, where the abnormal types include catalyst blockage, spray gun scaling, and ammonia escape exceeding the limit;

[0148] Based on the pre-set dynamic risk warning judgment model, analyze the probability value of the abnormal type and compare it with the pre-set abnormal threshold, and judge whether there is a device abnormality. Generate a device abnormality detection report based on the abnormality judgment result and push it to the operation and maintenance personnel.

[0149] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database or other media used in the various embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0150] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0151] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for identifying equipment abnormality in an SCR furnace denitration process, characterized in that: The method comprises the steps of: acquiring high-frequency acoustic wave signal data collected by a piezoelectric sensor array of a catalyst device, and recording the high-frequency acoustic wave signal data based on a preset common time axis to construct a high-frequency acoustic wave signal data set corresponding to the catalyst device; Acquiring temperature gradient difference data before and after the catalyst layer in the catalyst device and recording the temperature gradient difference data based on the common time axis to construct a temperature gradient difference data set; The preset temperature field matrix analysis module analyzes the temperature gradient difference data set based on a machine self-learning algorithm to generate a temperature field matrix; Acquire electromagnetic flowmeter data of each spray gun to record instantaneous flow value data of ammonia, and generate flow distribution of each spray gun according to the instantaneous flow value data of each spray gun; The preset equipment health index calculation model processes and calculates the high-frequency acoustic wave signal data, the temperature gradient difference data and the instantaneous flow value data based on a multimodal feature fusion algorithm to generate an equipment health index of the catalyst equipment; Based on the equipment health index, the flow distribution of each spray gun, and the temperature field matrix, a three-dimensional feature tensor is monitored, wherein the spray gun is used as a node, the node features include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation of fluid mechanics simulation; Acquire real-time temperature gradient difference data and instantaneous flow value data to calculate the corresponding real-time equipment health index, and output the probability value of the abnormal type according to the preset abnormal type probability model, wherein the abnormal type includes catalyst blockage, spray gun fouling and ammonia escape exceeding the limit; Based on the preset dynamic risk warning judgment model, the probability value of the abnormal type is analyzed and compared with the preset abnormal threshold, and it is determined whether there is an equipment abnormality. Based on the abnormal judgment result, an equipment abnormality detection report is generated and pushed to the operation and maintenance personnel.

2. The method for identifying equipment abnormality in an SCR furnace denitration process according to claim 1, characterized in that: In the step of processing and calculating the high-frequency acoustic wave signal data, the temperature gradient difference data and the instantaneous flow value data based on the multimodal feature fusion algorithm by the preset equipment health index calculation model to generate the equipment health index of the catalyst equipment, The preset acoustic emission activity analysis model analyzes and calculates the high-frequency acoustic wave signal data set to obtain the acoustic emission activity index; the calculation formula of the equipment health index H is: α+β+γ=1 Among them, α is the weight value of the acoustic emission signal, β is the weight value of the temperature gradient, γ is the weight value of the spray gun uniformity, and I AE is the acoustic emission activity index, ΔT is the temperature gradient difference data, T max is the preset catalyst tolerance limit, Q i is the real-time flow data value of the i-th spray gun, Calculated value for average flow rate of the spray gun.

3. The method for identifying equipment abnormality in an SCR furnace denitration process according to claim 2, characterized in that: In the step of obtaining the acoustic emission activity index by analyzing and calculating the high-frequency acoustic wave signal data set using the preset acoustic emission activity analysis model, the acoustic emission activity analysis model is preset with the following calculation formula for the acoustic emission activity index: Where E f is the energy data value of each frequency band after wavelet packet decomposition, N events is the acoustic emission event count, t sample is the sampling duration.

4. The method for identifying equipment abnormality in an SCR furnace denitration process according to claim 2, characterized in that: After the preset equipment health index calculation model processes and calculates the high-frequency acoustic wave signal data, the temperature gradient difference data and the instantaneous flow value data based on the multimodal feature fusion algorithm to generate the equipment health index of the catalyst equipment, the following steps are included: Obtaining NO from Catalyst Equipment x Concentration C NOx With O2 concentration C O2 , and constructing a gas concentration data set based on the common time axis records; The preset oxygen concentration correction model analyzes the gas concentration data set based on a machine self-learning algorithm to generate an oxygen concentration correction factor, wherein the calculation formula of the equipment health index H is: α+β+γ+δ=1, Where δ is the weight value of the flue gas component and ε is the oxygen concentration correction factor.

5. The method for identifying equipment abnormality in an SCR furnace denitration process according to claim 1, characterized in that: The step of analyzing the probability value of the abnormal type based on the preset dynamic risk warning judgment model and comparing it with the preset abnormal threshold, and judging whether there is an equipment abnormality, includes the following steps: If there is an abnormal situation, the preset abnormal location analysis module analyzes the probability value of the abnormal type based on the machine self-learning algorithm to locate the abnormal area and generate abnormal area data; Specifically, the probability value of the output abnormal type is obtained and the gradient data of the feature tensor is calculated according to the preset gradient calculation module; The preset feature map extraction module calculates the monitoring three-dimensional feature tensor through a spatial feature map generation algorithm to extract corresponding feature map data; The preset heat map generation module calculates the feature map data of the convolution layer based on the weighted fusion algorithm to generate highlight data; The feature map data is highlighted based on the highlight data to describe the abnormal position in the feature map, thereby locating the abnormal area and generating abnormal area data.

6. The method for identifying equipment abnormality in an SCR furnace denitration process according to claim 5, characterized in that: After the step of performing a highlighting operation on the feature map data based on the highlight data to describe the abnormal position in the feature map, thereby locating the abnormal area and generating abnormal area data, the following steps are included: The preset device topology map construction model constructs a device topology map based on preset node data and edge weight data; Obtain the nodes identified as abnormal sources, and calculate the estimated impact value of the abnormal nodes on downstream nodes through a preset propagation impact calculation model; The estimated impact value is compared with a preset impact threshold to determine whether the abnormal node is the main abnormal cause, and an abnormal node report is generated based on the determination result.

7. A device for identifying equipment abnormality in an SCR furnace denitration process, applied to a method for identifying equipment abnormality in an SCR furnace denitration process according to any one of claims 1 to 6, characterized in that: The device comprises: a high-frequency acoustic wave signal data set construction unit (1), which is used to obtain high-frequency acoustic wave signal data collected by a piezoelectric sensor array of a catalyst device, and record the high-frequency acoustic wave signal data based on a preset common time axis to construct a high-frequency acoustic wave signal data set corresponding to the catalyst device; A temperature gradient difference data set construction unit (2) is used to obtain temperature gradient difference data before and after the catalyst layer in the catalyst device and record the temperature gradient difference data based on the common time axis to construct a temperature gradient difference data set; A temperature field matrix generating unit (3) is used to pre-set a temperature field matrix analyzing module to analyze the temperature gradient difference data set based on a machine self-learning algorithm to generate a temperature field matrix; A spray gun flow distribution generating unit (4) is used to obtain electromagnetic flow meter data of each spray gun to record instantaneous flow value data of ammonia gas, and generate flow distribution of each spray gun according to the instantaneous flow value data of each spray gun; An equipment health index generating unit (5) is used to pre-set an equipment health index calculation model to process and calculate the high-frequency sound wave signal data, the temperature gradient difference data and the instantaneous flow value data based on a multi-modal feature fusion algorithm to generate an equipment health index of the catalyst equipment; A monitoring three-dimensional feature tensor composition unit (6) is used to monitor the three-dimensional feature tensor based on the equipment health index, the flow distribution of each spray gun, and the temperature field matrix, wherein the spray gun is used as a node, the node characteristics include flow deviation and position coordinates, and the edge weight is determined by the ammonia diffusion correlation degree simulated by fluid mechanics; An abnormal type probability value output unit (7) is used to obtain real-time temperature gradient difference data and instantaneous flow value data to calculate the corresponding real-time equipment health index, and output the probability value of the abnormal type according to a preset abnormal type probability model, wherein the abnormal type includes catalyst blockage, spray gun fouling and ammonia escape exceeding the limit; The device abnormality detection report generating unit (8) is used to analyze the probability value of the abnormality type based on a preset dynamic risk warning judgment model and compare it with a preset abnormality threshold, and judge whether there is an abnormality in the device, generate a device abnormality detection report based on the abnormality judgment result, and push it to the operation and maintenance personnel.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for identifying equipment abnormality in the SCR external denitration process according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a method for identifying equipment abnormality in an SCR external denitration process as described in any one of claims 1 to 6 are implemented.

Images