Petroleum chemical equipment operation abnormality early warning system

Abstract

The application discloses a kind of petroleum chemical equipment operation abnormal early warning system, it is related to equipment abnormal early warning technical field, including matching module and early warning module, matching module calls deformation abnormal rule to carry out first matching to deformation parameter, if there is abnormality in first matching result, then call image abnormal rule to carry out second matching to inner wall image data, if first matching result is no abnormality, then only output first matching result, early warning module is based on first matching result and second matching result constructs dynamic mapping relationship, obtains health degree evaluation value by coupling analysis, generates graded abnormal early warning signal according to health degree evaluation value, and executes corresponding grade early warning operation, through deformation and fouling image multidimensional collaborative matching, solve one-sidedness of single parameter detection, realize equipment abnormal comprehensive identification, based on entropy weight method constructs coupling degree function and dynamic health degree evaluation system, combined with correction and attenuation coefficient accurately quantifies equipment state, improve evaluation dynamic and accuracy.

Description

Technical Field

[0001] This invention relates to the field of equipment anomaly early warning technology, and in particular to an early warning system for abnormal operation of petrochemical equipment. Background Technology

[0002] In recent years, petrochemical equipment, such as reactors, has become a core component of the process industry. The stability of its operation directly affects production safety, process efficiency, and equipment lifespan. With the acceleration of industrial intelligence, reactor equipment anomaly early warning technology has become an increasingly important support for ensuring the continuous and reliable operation of petrochemical production. By identifying potential equipment anomalies in advance, unplanned shutdowns, equipment failures, and even safety accidents can be effectively avoided. This has irreplaceable significance for improving the reliability of equipment throughout its entire life cycle and reducing operation and maintenance costs.

[0003] However, existing equipment anomaly early warning technologies have many limitations. On the one hand, they mostly focus on single-dimensional detection and lack collaborative matching and dynamic correlation of multiple abnormal parameters such as deformation and scaling parameters, which can easily lead to one-sided anomaly identification. On the other hand, even when multi-parameter fusion is involved, simple threshold superposition or linear correlation are often used instead of combining historical fault data to build dynamic mapping relationships and coupling analysis, making it difficult to accurately quantify the overall health status of the equipment. In addition, early warning classification often relies on empirical threshold division and does not comprehensively consider factors such as parameter coupling degree and the attenuation of reactor equipment over time, resulting in insufficient accuracy and timeliness of early warnings, which cannot meet the urgent needs of petrochemical equipment for comprehensive, accurate and dynamic anomaly early warning. Summary of the Invention

[0004] The technical problem solved by this invention is that existing equipment anomaly early warning technologies have many limitations. On the one hand, they often focus on single-dimensional detection and lack collaborative matching and dynamic correlation of multiple abnormal parameters such as deformation parameters and scaling parameters, which can easily lead to one-sided anomaly identification. On the other hand, even when multi-parameter fusion is involved, simple threshold superposition or linear correlation is often used instead of combining historical fault data to build dynamic mapping relationships and coupling analysis, making it difficult to accurately quantify the overall health status of the equipment. In addition, early warning classification often relies on empirical threshold division and does not comprehensively consider factors such as parameter coupling degree and the attenuation of reactor equipment over time, resulting in insufficient accuracy and timeliness of early warning, which cannot meet the urgent needs of petrochemical equipment for comprehensive, accurate and dynamic anomaly early warning.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a petrochemical equipment abnormal operation early warning system, including a matching module and an early warning module;

[0006] The matching module calls a preset deformation anomaly rule to perform a first match on the deformation parameters. If the first match result is abnormal, it calls a preset image anomaly rule to perform a second match on the inner wall image data. If the first match result is without anomaly, it only outputs the first match result.

[0007] The early warning module constructs a dynamic mapping relationship based on the first matching result and the second matching result, obtains a health assessment value through coupling analysis, generates a graded abnormality early warning signal based on the health assessment value, and executes the corresponding level of early warning operation.

[0008] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system of the present invention, the deformation anomaly rules are constructed based on the historical deformation parameters of the reactor, and the scaling parameters are extracted based on the historical inner wall image data of the reactor, and the image anomaly rules are constructed based on the scaling parameters.

[0009] The configuration logic of the deformation anomaly rule includes substituting historical deformation parameters and preset benchmark deformation parameters into the cosine similarity formula to calculate the similarity value.

[0010] Abnormal deformation parameters are filtered out based on the similarity value. The filtering logic is that historical deformation parameters with similarity values ​​lower than a preset threshold are judged as abnormal deformation parameters.

[0011] A first rule threshold is set based on the abnormal deformation parameters, and an abnormal deformation rule is constructed based on the first rule threshold.

[0012] The configuration logic of the image anomaly rule is as follows: extract scaling parameters based on historical inner wall image data, and construct image anomaly rules based on scaling parameters. Specifically, this includes using an adaptive threshold segmentation algorithm to segment the historical inner wall image data to obtain the scaling area and the inner wall background area, and using the Canny edge detection algorithm to extract the contour features of the scaling area.

[0013] By combining contour features and image pixel scale, the scale thickness, scale area ratio, and scale distribution density are calculated.

[0014] Scaling parameters can be configured as a single dimension or a combination of multiple dimensions;

[0015] Based on the numerical range of scale thickness and scale area ratio, the first abnormal feature quantity is obtained. The first abnormal feature quantity includes the scale thickness exceeding the standard coefficient and the scale area ratio exceeding the limit value.

[0016] The second abnormal characteristic quantity is obtained from the scale distribution density in the scale parameters. The second abnormal characteristic quantity is the proportion of dense scale area and the scale unevenness.

[0017] The first and second abnormal feature quantities are combined to obtain the judgment feature quantity, which is then converted into the second rule threshold. After being set according to the scaling parameter safety standard, the image anomaly rule is constructed.

[0018] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system of the present invention, the deformation parameters are first matched by calling the deformation anomaly rule. If the first matching result is abnormal, the inner wall image data is second matched by calling the image anomaly rule to obtain the second matching result. Specifically, the deformation parameters of the reactor are obtained, the preset deformation anomaly rule is called, the deformation parameters are compared with the first rule threshold in the deformation anomaly rule, and if the deformation parameters exceed the first rule threshold, the first matching result is determined to be abnormal. The first matching result including the deformation parameters, the first threshold comparison difference and the anomaly identifier is output.

[0019] The first threshold comparison difference is the difference between the deformation parameter and the first rule threshold.

[0020] Synchronously acquire image data of the inner wall of the reactor, call the preset image anomaly rules, extract scaling parameters from the inner wall image data based on the image anomaly rules, and compare the scaling parameters with the second rule threshold in the image anomaly rules;

[0021] If the scaling parameter exceeds the second rule threshold, the second matching result is determined to be abnormal, and the output includes the scaling parameter, the difference between the second threshold and the abnormality indicator of the second matching result.

[0022] The second threshold comparison difference is the difference between the scaling parameters and the second rule threshold.

[0023] If the scaling parameters do not exceed the threshold of the second rule, the second matching result is determined to be without abnormality, and the output includes the scaling parameters and the no abnormality indicator of the second matching result.

[0024] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system of the present invention, if the first matching result is no anomaly, the call to the anomaly rule is terminated, and only the first matching result is output. Specifically, if the deformation parameter does not exceed the first rule threshold, the first matching result is determined to be no anomaly, the call process of the image anomaly rule is directly terminated, and only the first matching result including the deformation parameter and the no anomaly indicator is output.

[0025] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system of the present invention, the dynamic mapping relationship includes a first dynamic mapping relationship and a second dynamic mapping relationship.

[0026] The logic for constructing the first dynamic mapping relationship is as follows: when the matching result includes the first matching result and the second matching result, the deformation parameter, the first threshold comparison difference and the abnormal identifier in the first matching result are used as the first mapping dimension, and the scaling parameter, the second threshold comparison difference and the abnormal identifier in the second matching result are used as the second mapping dimension. The first dynamic mapping relationship is constructed based on the first mapping dimension and the second mapping dimension.

[0027] The deformation anomaly level is first divided based on the first threshold comparison difference to obtain different deformation anomaly levels. The first division logic is as follows: the first threshold comparison difference in the left-closed-right-open interval from 0 to A corresponds to the mild deformation anomaly level; the left-closed-right-open interval from A to B corresponds to the moderate deformation anomaly level; and the left-closed-right-open interval from B to +∞ corresponds to the severe deformation anomaly level. A and B are both preset positive numbers, and A is less than B.

[0028] The scaling anomaly level is divided into two categories based on the comparison difference of the second threshold. The second classification logic is as follows: the left-closed and right-open interval between 0 and C corresponds to the mild scaling anomaly level; the left-closed and right-open interval between C and D corresponds to the moderate scaling anomaly level; and the left-closed and right-open interval between D and +∞ corresponds to the severe scaling anomaly level. C and D are both preset positive numbers, and C is less than D.

[0029] The first dynamic mapping relationship includes the dynamic mapping relationship between deformation anomaly level and scaling anomaly level, specifically including mild deformation anomaly level corresponding to mild scaling anomaly level, moderate deformation anomaly level corresponding to moderate scaling anomaly level, and severe deformation anomaly level corresponding to severe scaling anomaly level.

[0030] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system described in this invention, the second dynamic mapping relationship construction logic is as follows: when the matching result only includes the first matching result, the second matching result corresponding to the image anomaly rule that has not been called is set as the default result.

[0031] The default result includes a scaling parameter of 0, an anomaly identifier of no anomaly, and a second threshold comparison difference of 0. The deformation parameter, the first threshold comparison difference, and the anomaly identifier in the first matching result are used as the main dimension, and the default result is used as the auxiliary dimension to construct a second dynamic mapping relationship.

[0032] The first dynamic mapping relationship and the second dynamic mapping relationship use the anomaly dimension as the classification benchmark, and associate the parameter values ​​of the corresponding dimension to form a structured relationship, providing standardized data input for subsequent coupling analysis;

[0033] The abnormality dimensions include deformation dimension and scaling dimension;

[0034] The parameter values ​​include deformation parameters, scaling parameters, first threshold comparison difference and second threshold comparison difference, and are also bound to the corresponding dimension of the anomaly identifier.

[0035] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system described in this invention, the coupling analysis includes constructing an efficacy function and a coupling degree function;

[0036] The logic for constructing the performance function is as follows: when the deformation parameter has a positive effect, the expression of the deformation performance function is:

[0037] ;

[0038] in, This represents the deformation efficiency function value corresponding to the j-th deformation parameter. This represents the measured value of the j-th dimension of the deformation parameter. , These represent the lower and upper risk limits for the j-th dimension of the deformation parameter, respectively. These are preset constants used to standardize the efficacy function benchmark value when deformation parameters and scaling parameters reach health goals;

[0039] When the deformation parameter has a negative effect, the expression for the deformation effect function is:

[0040] ;

[0041] When the scaling parameters have a positive effect, the expression for the scaling effect function is:

[0042] ;

[0043] in, This represents the scaling efficiency function value corresponding to the Kth scaling parameter. This represents the measured value of the Kth dimension of the scaling parameters. , These represent the lower and upper risk limits for the Kth dimension of the scaling parameter, respectively.

[0044] When the scaling parameter has a negative effect, the expression for the scaling effect function is:

[0045] ;

[0046] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system described in this invention, the coupling degree function construction logic is as follows, and the coupling degree function expression of the deformation parameter is as follows:

[0047] ;

[0048] In the first dynamic mapping relationship, m represents the m dimensions of deformation parameters. This indicates that the deformation parameters contribute to the overall health by linear weighting. Let represent the first baseline weight of the j-th dimension of the corresponding deformation parameter. This first baseline weight is determined from historical fault data using the entropy weight method and satisfies the following:

[0049] ;

[0050] The coupling degree function expression for the scaling parameters is:

[0051] ;

[0052] In the first dynamic mapping relationship, n represents n dimensions of scaling parameters. This indicates the contribution of scaling parameters to overall health, obtained through linear weighting. Let the second baseline weight be the k-th dimension of the corresponding scaling parameter. This second baseline weight is determined from historical fault data using the entropy weight method and satisfies the following:

[0053] ;

[0054] when and When any one of the values ​​is not zero, the coupling function expression between the deformation parameter and the scaling parameter is:

[0055] ;

[0056] Wherein, the coupling degree C is a closed interval from 0 to 1;

[0057] when and When all values ​​are 0, the coupling function between the deformation parameter and the scaling parameter is 0.

[0058] As a preferred embodiment of the petrochemical equipment operation anomaly early warning system described in this invention, the calculation logic for the health assessment value is as follows, and the expression for the health assessment value is:

[0059] ;

[0060] Based on the deformation anomaly level and scaling anomaly level in the first dynamic mapping relationship, , Assign dynamic correction coefficients respectively , ;

[0061] For each level increase in the deformation anomaly grade, Increase by 0.1, the aforementioned The interval is a closed interval from 1 to 1.3;

[0062] For each level increase in the scaling abnormality level, Increase by 0.1, the aforementioned The interval is a closed interval from 1 to 1.3;

[0063] Where T is the decay coefficient of the reactor operating time, and the expression for T is:

[0064] ;

[0065] Where T is a double-closed interval from 0.8 to 1;

[0066] For the second dynamic mapping relationship, the measured values ​​of scaling parameters are set as the baseline value of 0, and the weights of each dimension of the scaling parameters are... Set the corresponding second benchmark weight. When the scaling parameter is a single dimension, the weight is 1. Calculate according to the health assessment value calculation logic to obtain H.

[0067] The health assessment value H indicates that the lower the value, the worse the operating condition of the reactor.

[0068] As a preferred embodiment of the abnormal operation early warning system for petrochemical equipment described in this invention, the health assessment value H is divided into a first interval, a second interval, a third interval, and a fourth interval.

[0069] The first interval is where the H value belongs to the double-closed interval from the first value to the second value, corresponding to the first level of abnormal warning level;

[0070] The second interval is where the H value falls within the double-closed interval between the second and third values, corresponding to a level two anomaly warning level;

[0071] The third interval is where the H value belongs to the double-closed interval between the third and fourth values, corresponding to the third level of abnormal warning.

[0072] The fourth interval is where the H value belongs to the double-closed interval between the fourth and fifth values, corresponding to the fourth level of abnormal warning level;

[0073] If it is a Level 1 warning, a red audible and visual alarm signal will be generated to trigger the emergency shutdown preparation procedure of the reactor in advance, and an emergency maintenance SMS instruction with fault location and trend prediction will be sent to the operation and maintenance personnel.

[0074] If it is a Level II warning, a yellow audible and visual alarm signal will be generated, triggering the reactor load reduction plan in advance, and sending a maintenance reminder SMS instruction with abnormal parameter tracing and development trend analysis to the operation and maintenance personnel;

[0075] If it is a Level 3 warning, a blue audible and visual alert signal will be generated, and an abnormal warning email and planned inspection suggestions will be sent to the operation and maintenance personnel in advance, prompting them to pay attention to the operating status of the reactor and start preventive data monitoring.

[0076] If there is no abnormality level, no abnormality warning signal will be generated. Only the health data of the reactor will be recorded for historical analysis and early warning model optimization and iteration.

[0077] The beneficial effects of this invention are as follows: By multi-dimensional collaborative matching of deformation and scaling images, the one-sidedness of single-parameter detection is solved, enabling comprehensive identification of equipment anomalies. Based on the entropy weight method, a coupling degree function and dynamic health assessment system are constructed. Combined with the anomaly level correction coefficient and the runtime decay coefficient, the overall health status of the reactor is accurately quantified, significantly improving the dynamism and accuracy of the assessment. Furthermore, through a graded early warning strategy, targeted handling measures are triggered based on health differences, effectively reducing the risk of unplanned downtime and equipment failure, lowering operation and maintenance costs, and ultimately significantly improving the reliability and intelligent operation and maintenance level of petrochemical equipment. Attached Figure Description

[0078] Figure 1 This is a basic flowchart of a petrochemical equipment operation anomaly early warning system provided in one embodiment of the present invention. Detailed Implementation

[0079] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0080] Example, refer to Figure 1 As an embodiment of the present invention, a petrochemical equipment abnormal operation early warning system is provided, including a matching module and an early warning module;

[0081] The matching module calls the preset deformation anomaly rules to perform a first match on the deformation parameters. If the first match result is abnormal, it calls the preset image anomaly rules to perform a second match on the inner wall image data. If the first match result is without anomalies, it only outputs the first match result.

[0082] The early warning module constructs a dynamic mapping relationship based on the first and second matching results, obtains a health assessment value through coupling analysis, generates a graded abnormality early warning signal based on the health assessment value, and executes the corresponding level of early warning operation.

[0083] In one embodiment, the matching module first calls the deformation anomaly rule to perform a first match on the deformation parameters. If the first match result is abnormal, it further calls the image anomaly rule to perform a second match on the inner wall image data. If the first match result is not abnormal, it only outputs the first match result. The early warning module constructs a dynamic mapping relationship based on the first and second match results, obtains a health assessment value through coupling analysis, and then generates a graded anomaly early warning signal based on the health assessment value and executes the corresponding level of early warning operation. Through the hierarchical matching logic of deformation priority and image supplementation and the health measurement mechanism of coupling analysis, the accurate identification and graded handling of anomalies in petrochemical equipment are realized. This not only improves the pertinence and efficiency of anomaly identification, but also ensures the timeliness and precision of equipment operation and maintenance through dynamic mapping and graded early warning. It effectively reduces the risk of failure caused by abnormal operation of petrochemical equipment and provides efficient and reliable technical support for intelligent operation and maintenance of equipment.

[0084] Deformation anomaly rules are constructed based on the historical deformation parameters of the reactor, and scaling parameters are extracted based on the historical inner wall image data of the reactor. Image anomaly rules are then constructed based on the scaling parameters.

[0085] The configuration logic for deformation anomaly rules includes substituting historical deformation parameters and preset baseline deformation parameters into the cosine similarity formula to calculate the similarity value.

[0086] Abnormal deformation parameters are filtered out based on similarity values. The filtering logic is that historical deformation parameters with similarity values ​​lower than a preset threshold are judged as abnormal deformation parameters.

[0087] A first rule threshold is set based on the abnormal deformation parameters, and deformation anomaly rules are constructed based on the first rule threshold.

[0088] The configuration logic of the image anomaly rules is as follows: extract scaling parameters based on historical inner wall image data, and construct image anomaly rules based on scaling parameters. Specifically, the process includes using an adaptive threshold segmentation algorithm to segment the historical inner wall image data to obtain the scaling area and the inner wall background area, and using the Canny edge detection algorithm to extract the contour features of the scaling area.

[0089] By combining contour features and image pixel scale, the scale thickness, scale area ratio, and scale distribution density are calculated.

[0090] Scaling parameters can be configured as a single dimension or a combination of multiple dimensions;

[0091] Based on the numerical range of scale thickness and scale area ratio, the first abnormal characteristic quantity is obtained. The first abnormal characteristic quantity includes the scale thickness exceeding the standard coefficient and the scale area ratio exceeding the limit value.

[0092] The second abnormal characteristic quantity is obtained from the scale distribution density in the scale parameters. The second abnormal characteristic quantity is the proportion of dense scale area and the scale non-uniformity.

[0093] The first and second abnormal feature quantities are combined to obtain the judgment feature quantity, which is then converted into the second rule threshold. After being set according to the scaling parameter safety standard, the image anomaly rule is constructed.

[0094] In one embodiment, deformation anomaly rules are constructed based on the historical deformation parameters of the reactor. Simultaneously, scaling parameters are extracted from historical internal wall image data of the reactor to construct image anomaly rules. In configuring the deformation anomaly rules, a preset baseline deformation parameter is the reactor's designed rated deformation range, such as a maximum allowable deformation of 0.5mm in the diameter direction and 1.0mm in the length direction. The historical deformation parameters and this baseline deformation parameter are substituted into a cosine similarity formula to calculate the similarity value. The cosine similarity formula is a commonly used parameter similarity calculation method in the prior art. A preset similarity threshold of 0.85 is set based on statistical historical reactor failure data. Historical deformation parameters with similarity values ​​below 0.85 are selected as abnormal deformation parameters. These abnormal deformation parameters are set with a first rule threshold of ±0.8mm. The basis for this setting is to take the minimum extreme value of the abnormal deformation parameter and add a 10% safety redundancy to ensure early identification of deformation risks and avoid alarms only when the fault threshold is approached. This is used to construct deformation anomaly rules. In configuring image anomaly rules, an adaptive threshold segmentation algorithm is first used to segment historical inner wall image data to obtain the scaled area and the inner wall background area. The adaptive threshold segmentation algorithm is a conventional image segmentation technique in this field and can be directly adapted to the scaled area identification scenario of inner wall images. Then, the Canny edge detection algorithm is used to extract the contour features of the scaled area. Combining the contour features with the image pixel scale, the scale thickness, scale area ratio, and scale distribution density are calculated. The image pixel ratio... For example, 1 pixel corresponds to an actual thickness of 0.1 mm. Scaling parameters can be configured as a single dimension or a combination of multiple dimensions. The first abnormal characteristic quantity is set based on the numerical range of scale thickness and scale area ratio. The safe standard for scale thickness is less than or equal to 0.3 mm, corresponding to a first rule threshold of 0.3 mm. The scale thickness exceedance coefficient is equal to the measured thickness of 0.3 mm, based on the fact that when the thickness exceeds 0.3 mm, the heat transfer efficiency of the reactor decreases too quickly and significantly, easily leading to localized overheating. The safe standard for scale area ratio is less than or equal to 5%, corresponding to a first rule threshold of 5%. The scale area ratio exceeding the limit is equal to -5% of the measured ratio, based on the fact that when the area ratio exceeds 5%, the risk of internal wall corrosion increases significantly. As the scale density increases, the probability of failure increases. A second abnormal characteristic quantity is set based on the scale distribution density. The safe standard for scale distribution density is less than or equal to 10 scale particles / m². The threshold for the proportion of densely scaled areas is set at 15%, based on the fact that when the proportion of densely scaled areas exceeds 15%, local stress concentration is significant, easily inducing deformation superposition failures. The threshold for scale non-uniformity is set at 0.6, based on the fact that when the non-uniformity exceeds 0.6, the vibration amplitude of the reactor equipment increases, affecting stability. The first and second abnormal characteristic quantities are combined to obtain judgment characteristic quantities, which are then converted into second rule thresholds, such as a scale thickness exceeding the standard coefficient greater than or equal to 1.2, an area proportion exceeding the limit greater than or equal to 3%, a densely scaled area proportion greater than or equal to 15%, and a non-uniformity greater than or equal to 0.6. An anomaly is determined by meeting any two of the following criteria: image anomaly rules are constructed after setting safety standards for scaling parameters. Through clear and realistic threshold and numerical settings, combined with cosine similarity, adaptive threshold segmentation, and Canny edge detection technology, the objective accuracy of deformation anomaly rule construction is ensured, and the precision of scaling parameter extraction and image anomaly rule construction is achieved. This provides a solid and practical rule foundation for the effective identification of subsequent equipment malfunctions, significantly improving the reliability and practicality of anomaly early warning for petrochemical equipment and effectively reducing false alarms and missed alarms caused by unreasonable rule thresholds.

[0095] The deformation parameters are first matched by calling the deformation anomaly rule. If the first matching result is abnormal, the image anomaly rule is called to perform a second matching on the inner wall image data to obtain the second matching result. Specifically, the deformation parameters of the reactor are obtained, the preset deformation anomaly rule is called, the deformation parameters are compared with the first rule threshold in the deformation anomaly rule, and if the deformation parameters exceed the first rule threshold, the first matching result is determined to be abnormal. The first matching result including the deformation parameters, the difference between the first threshold and the anomaly identifier is output.

[0096] The first threshold comparison difference is the difference between the deformation parameter and the first rule threshold.

[0097] Synchronously acquire image data of the inner wall of the reactor, call the preset image anomaly rules, extract scaling parameters from the inner wall image data based on the image anomaly rules, and compare the scaling parameters with the second rule threshold in the image anomaly rules;

[0098] If the scaling parameter exceeds the second rule threshold, the second matching result is determined to be abnormal, and the output includes the scaling parameter, the difference between the second threshold and the abnormality indicator of the second matching result.

[0099] The second threshold comparison difference is the difference between the scaling parameters and the second rule threshold.

[0100] If the scaling parameters do not exceed the threshold of the second rule, the second matching result is determined to be without abnormality, and the output includes the scaling parameters and the no abnormality indicator of the second matching result.

[0101] In one embodiment, the matching module first calls a preset deformation anomaly rule to perform a first match on the deformation parameters of the reactor. Specifically, a high-precision displacement sensor deployed on the outer wall of the reactor (a conventional deformation monitoring device in the prior art with a measurement accuracy of 0.01 mm) is used to acquire deformation parameters in real time. These parameters include real-time deformation data in the diameter and length directions. Then, the deformation anomaly rule constructed above is called to compare the real-time deformation parameters with the first rule threshold of ±0.8 mm. If the real-time deformation parameters exceed... If the measured deformation in the diameter direction is 1.0 mm, the first matching result is determined to be abnormal. The output includes deformation parameters, the first threshold comparison difference, and an anomaly indicator. The deformation parameters are, for example, 1.0 mm in the diameter direction and 0.9 mm in the length direction. The first threshold comparison difference is, for example, 0.2 mm in the diameter direction and 0.1 mm in the length direction. The anomaly indicator is, for example, abnormal deformation or moderate deformation. The first threshold comparison difference is the difference between the measured deformation parameters and the first rule threshold, i.e., 1.0 mm - 0.8 mm = 0.2 mm. If the first matching result is abnormal, the inner wall image data is acquired using an embedded high-definition industrial camera. This embedded high-definition industrial camera has 20 megapixels and a frame rate of 10 frames per second, suitable for high-temperature and high-pressure conditions. The second matching is then performed by calling preset image anomaly rules. Specifically, based on the algorithm logic in the image anomaly rules, the image is segmented using an adaptive threshold segmentation algorithm to obtain the scaling area. After extracting the contour using the Canny edge detection algorithm, and combined with an image pixel scale of 0.1mm per pixel, scaling parameters are calculated: scaling thickness 0.4mm, scaling area percentage 8%, scaling dense area percentage 20%, and scaling unevenness 0.7. These parameters are then compared with the second rule thresholds: scaling thickness exceeding the limit by a factor greater than or equal to 1.2, area percentage exceeding the limit by a factor greater than or equal to 3%, scaling dense area percentage greater than or equal to 15%, and scaling unevenness exceeding the limit by a factor greater than or equal to 15%. If the value is equal to 0.6, any two of the following conditions must be met to determine an anomaly in the comparison. In this case, the scaling thickness exceedance coefficient = 0.4mm / 0.3mm ≈ 1.33, and the area ratio exceedance value = 8% - 5% = 3%. Both of these conditions are met, so the second matching result is determined to be an anomaly. The output includes scaling parameters, the second threshold comparison difference, and anomaly indicators. Scaling parameters include, for example, a thickness of 0.4mm and an area ratio of 8%. The second threshold comparison difference includes, for example, a thickness exceedance coefficient of 0.13 and an area ratio exceedance value of 3%. The anomaly indicators include, for example, scaling anomaly or moderate scaling. The second threshold comparison difference is the quantitative value of the scaling parameters exceeding the second rule threshold. If the scaling parameters do not reach the threshold, such as a scaling thickness of 0.2mm and an area ratio of 3%, the second matching result is determined to be without anomalies, and the output includes the above scaling parameters and no anomaly indicators.By employing a deformation-priority and anomaly-linked matching logic, combined with clear hardware selection, parameter thresholds, and comparison standards, the system ensures the accuracy of anomaly matching. Layered matching reduces invalid image analysis, and image matching is only initiated when deformation anomalies occur, improving system operating efficiency. At the same time, the output quantified difference and anomaly identifier provide accurate data support for the health assessment of the subsequent early warning module, effectively avoiding anomaly omissions caused by ambiguous matching logic and further enhancing the reliability of device early warning.

[0102] If the first matching result is no anomaly, the call to the anomaly rule is terminated, and only the first matching result is output. Specifically, if the deformation parameter does not exceed the threshold of the first rule, the first matching result is determined to be no anomaly, the call process of the image anomaly rule is directly terminated, and only the first matching result including the deformation parameter and the no anomaly indicator is output.

[0103] In one embodiment, a high-precision displacement sensor deployed on the outer wall of the reactor measures deformation parameters in the diameter and length directions in real time with an accuracy of 0.01 mm. Then, a preset deformation anomaly rule is invoked to accurately compare the measured deformation parameters with the previously set ±0.8 mm first rule threshold. If the measured deformation parameters do not exceed the threshold, for example, the measured deformation in the diameter direction is 0.6 mm and the measured deformation in the length direction is 0.7 mm, both of which are within the safe range of ±0.8 mm, then the first matching result is directly determined to be without anomaly. At this time, the system will immediately terminate the subsequent image anomaly rule invocation process and will not start the image acquisition and scaling parameter analysis of the inner wall industrial camera. It will only output the first matching result including specific deformation parameters, such as diameter 0.6 mm, length 0.7 mm, and a deformation no-anomaly indicator. By terminating the system when no anomalies are detected, the system effectively avoids unnecessary resource consumption caused by starting image analysis when deformation parameters are normal. Compared to executing a dual-matching process throughout the entire process, this logic reduces computing power consumption in anomaly-free scenarios, while also reducing the start-up and shutdown frequency of industrial cameras and image processors, thus extending equipment lifespan. On the other hand, the clear anomaly detection and the synchronous output of specific deformation parameters provide maintenance personnel with a clear basis for the status of the reactor equipment, and also accumulate accurate basic data for subsequent historical health data archiving and early warning model iteration, further strengthening the dual advantages of accurate monitoring and efficient operation of this system.

[0104] Dynamic mapping relationships include first dynamic mapping relationships and second dynamic mapping relationships;

[0105] The logic for constructing the first dynamic mapping relationship is as follows: when the matching result includes the first matching result and the second matching result, the deformation parameter, the first threshold comparison difference and the abnormal identifier in the first matching result are used as the first mapping dimension, and the scaling parameter, the second threshold comparison difference and the abnormal identifier in the second matching result are used as the second mapping dimension. The first dynamic mapping relationship is constructed based on the first mapping dimension and the second mapping dimension.

[0106] The deformation anomaly level is first divided based on the first threshold comparison difference to obtain different deformation anomaly levels. The first division logic is as follows: the first threshold comparison difference in the left-closed-right-open interval from 0 to A corresponds to the mild deformation anomaly level; the left-closed-right-open interval from A to B corresponds to the moderate deformation anomaly level; and the left-closed-right-open interval from B to +∞ corresponds to the severe deformation anomaly level. A and B are both preset positive numbers, and A is less than B.

[0107] The scaling anomaly level is further divided based on the difference between the second thresholds to obtain different scaling anomaly levels. The second division logic is as follows: the left-closed-right-open interval between 0 and C corresponds to a mild scaling anomaly level; the left-closed-right-open interval between C and D corresponds to a moderate scaling anomaly level; and the left-closed-right-open interval between D and +∞ corresponds to a severe scaling anomaly level. C and D are both preset positive numbers, and C is less than D.

[0108] The first dynamic mapping relationship includes the dynamic mapping relationship between deformation anomaly level and scaling anomaly level, specifically including the relationship between mild deformation anomaly level and mild scaling anomaly level, moderate deformation anomaly level and moderate scaling anomaly level, and severe deformation anomaly level and severe scaling anomaly level.

[0109] The second dynamic mapping relationship construction logic is that when the matching result only includes the first matching result, the second matching result corresponding to the image anomaly rule that was not called is set as the default result;

[0110] The default results include scaling parameters of 0, anomaly identification of no anomaly, and second threshold comparison difference of 0. The deformation parameters, first threshold comparison difference, and anomaly identification in the first matching results are used as the main dimensions, and the default results are used as the auxiliary dimensions to construct the second dynamic mapping relationship.

[0111] The first and second dynamic mapping relationships use the anomaly dimension as the classification benchmark, and associate the parameter values ​​of the corresponding dimension to form a structured relationship, providing standardized data input for subsequent coupling analysis;

[0112] The abnormal dimensions include deformation dimension and scaling dimension;

[0113] The parameter values ​​include deformation parameters, scaling parameters, the difference between the first threshold and the second threshold, and are also bound to the corresponding dimension's anomaly identifier.

[0114] In one embodiment, a first dynamic mapping relationship and a second dynamic mapping relationship are constructed based on different matching result scenarios, clarifying dimensional association rules and numerical standards to provide a unified data input format for health assessment. When the matching result includes both the first and second matching results, a first dynamic mapping relationship is constructed, using the deformation parameter, the first threshold comparison difference, and the anomaly identifier in the first matching result as the first mapping dimension, and the scaling parameter, the second threshold comparison difference, and the anomaly identifier in the second matching result as the second mapping dimension, forming the first dynamic mapping relationship through structured association. The deformation anomaly level is classified based on the difference between the first threshold and the pre-set thresholds A=0.1mm and B=0.3mm. This classification is based on the previous rule threshold of ±0.8mm, and is divided according to the fault risk gradient. 0.1mm corresponds to mild risk, and 0.3mm corresponds to severe risk, conforming to the statistical patterns of historical fault data. A difference between the first threshold and the left-closed-right-open interval (0-0.1mm) corresponds to mild deformation anomaly; 0.1mm to 0.3mm corresponds to moderate deformation anomaly; and 0.3mm to +∞ corresponds to severe deformation anomaly. Scaling anomalies are also classified based on the difference between the second threshold and the pre-set threshold. The scale level is set with preset values ​​of C=1% and D=5%, based on the scaling safety standard. 1% is considered a slight exceedance and 5% is a severe exceedance, matching the gradient of decreased heat transfer efficiency and increased corrosion risk caused by scaling. The second threshold comparison difference is 0% to 1% in the left-closed-right-open range, corresponding to mild scaling anomaly; 1% to 5% in the left-closed-right-open range, corresponding to moderate scaling anomaly; and 5% to +∞ in the left-closed-right-open range, corresponding to severe scaling anomaly. In the first dynamic mapping relationship, deformation and scaling anomaly level are correlated, i.e., mild deformation corresponds to mild scaling, moderate deformation corresponds to moderate scaling, and severe deformation corresponds to severe scaling.

[0115] When the matching result only includes the first matching result, a second dynamic mapping relationship is constructed. The second matching result corresponding to the image anomaly rule that has not been called is set as the default result. Specifically, the scaling parameter is equal to 0, the anomaly identifier is equal to no anomaly, and the second threshold comparison difference is equal to 0. The deformation parameter, the first threshold comparison difference, and the anomaly identifier in the first matching result are used as the main dimension, and the default result is used as the auxiliary dimension. The second dynamic mapping relationship is formed through dimension association.

[0116] Both dynamic mapping relationships use the deformation and scaling dimensions of the anomaly dimension as classification benchmarks. The corresponding parameter values, such as deformation parameters, scaling parameters, first threshold comparison difference, and second threshold comparison difference, are bound to the anomaly identifier to form structured data. This design achieves standardized transformation of different matching results by clearly defining dimension division, level thresholds, and association rules. This ensures the uniformity and accuracy of subsequent coupled analysis data input, and reflects the synergy of parameter anomalies through level correspondence, reducing data redundancy processing and improving the efficiency of health assessment. At the same time, the structured data format facilitates rapid system access and model iteration optimization, enhancing the practicality and scalability of the early warning module.

[0117] Coupling analysis includes constructing a power function and a coupling degree function;

[0118] The logic for constructing the effect function is as follows: when the deformation parameter has a positive effect, the expression of the deformation effect function is:

[0119] ;

[0120] in, This represents the deformation efficiency function value corresponding to the j-th deformation parameter. This represents the measured value of the j-th dimension of the deformation parameter. , These represent the lower and upper risk limits for the j-th dimension of the deformation parameter, respectively. These are preset constants used to standardize the efficacy function benchmark value when deformation parameters and scaling parameters reach health goals;

[0121] When the deformation parameter has a negative effect, the expression for the deformation effect function is:

[0122] ;

[0123] When the scaling parameters have a positive effect, the expression for the scaling effect function is:

[0124] ;

[0125] in, This represents the scaling efficiency function value corresponding to the Kth scaling parameter. This represents the measured value of the Kth dimension of the scaling parameters. , These represent the lower and upper risk limits for the Kth dimension of the scaling parameter, respectively.

[0126] When the scaling parameter has a negative effect, the expression for the scaling effect function is:

[0127] ;

[0128] In one embodiment, the power function of the coupling analysis is used to quantify the individual contributions of deformation parameters and scaling parameters to the equipment health status, providing basic data for subsequent coupling degree and health assessment. The specific construction logic is as follows:

[0129] For deformation parameters, a deformation efficacy function is constructed to distinguish between positive and negative efficacy scenarios, and a risk lower bound is preset for the j-th dimension of the deformation parameters. For example, -0.8mm in the diameter direction and -1.0mm in the length direction (the lower limit of the safety range of the first rule threshold mentioned above) are set as the upper limit of the risk. For example, +0.8mm in the diameter direction and +1.0mm in the length direction are set based on the upper limit of the safety range of the first rule threshold mentioned above, and a preset constant is also used. =1, set based on the efficacy benchmark value of the health target of uniformly quantified deformation and scaling parameters, to facilitate the comparison of the contribution of parameters. When the deformation parameter has a positive effect, such as the diameter deformation being closer to the upper limit, it is more conducive to the stability of the reactor equipment. The efficacy function expression of the j-th deformation parameter is:

[0130] For example, measured values ​​in the diameter direction =0.6mm, substituting, we get: When the deformation parameter has a negative effect, such as the longer the length deformation is, the healthier the result is, the expression is:

[0131] For example, measured value in the length direction =0.7mm, substituting, we get: .

[0132] For scaling parameters, a scaling effect function is constructed by distinguishing between positive and negative effect scenarios, and a risk lower limit is preset for the Kth dimension of the scaling parameters. If the scale thickness is 0mm and the area percentage is 0%, the setting is based on the safe starting value of the scale parameters and the upper limit of risk. For example, if the scale thickness is 0.3 mm and the area accounts for 5%, the threshold is set based on the critical threshold of the scale safety standard mentioned above. Still taking 1, when the scaling parameter has a positive effect, such as the closer the scaling distribution density is to the lower limit, the healthier the scale is, the expression for the effect function of the Kth scaling parameter is:

[0133] For example, the measured value of scale thickness =0.1mm, substituting, we get: When scaling parameters have a negative effect, such as a healthier scale uniformity as it deviates further from the upper limit, the expression is: For example, the measured value of scale unevenness =0.4, upper limit of risk =0.6, substituting gives This efficacy function, through clearly defined risk upper and lower limits, positive and negative efficacy scenarios, and quantitative formulas, achieves precise quantification of the health contribution of deformation and scaling parameters. It provides a unified numerical basis for the subsequent multi-parameter weighting of the coupling function, while ensuring the comparability between different types of parameters. This effectively improves the accuracy of coupling analysis and health assessment, and lays a quantitative analysis foundation for the accuracy of abnormal early warning of reactor equipment.

[0134] The coupling degree function is constructed using the following logic: the coupling degree function expression for the deformation parameter is as follows:

[0135] ;

[0136] In the first dynamic mapping relationship, m represents the m dimensions of deformation parameters. This indicates that the deformation parameters contribute to the overall health by linear weighting. Let represent the first baseline weight of the j-th dimension of the corresponding deformation parameter. This first baseline weight is determined from historical fault data using the entropy weight method and satisfies the following:

[0137] ;

[0138] The coupling degree function expression for the scaling parameters is:

[0139] ;

[0140] In the first dynamic mapping relationship, n represents n dimensions of scaling parameters. This indicates the contribution of scaling parameters to overall health, obtained through linear weighting. Let the second baseline weight be the k-th dimension of the corresponding scaling parameter. This second baseline weight is determined from historical fault data using the entropy weight method and satisfies the following:

[0141] ;

[0142] when and When any one of the values ​​is not zero, the coupling function expression between the deformation parameter and the scaling parameter is:

[0143] ;

[0144] Wherein, the coupling degree C is a closed interval from 0 to 1;

[0145] when and When all values ​​are 0, the coupling function between the deformation parameter and the scaling parameter is 0.

[0146] In one embodiment, the coupling degree function is used to quantify the comprehensive contribution and synergistic relationship between deformation parameters and scaling parameters on the health of the reactor equipment, providing core quantitative basis for health assessment. The specific construction logic is as follows:

[0147] For the deformation parameter, its coupling degree function expression is:

[0148] And satisfy Where m represents the number of deformation parameter dimensions, for example, taking m=2 corresponds to two deformation dimensions: diameter and length. The power function values ​​for each deformation dimension, such as the diameter direction. =0.875, length direction =0.15, The first baseline weight is determined based on historical fault data using the entropy weight method, such as the diameter dimension weight. =0.6, Length dimension weight =0.4, reflecting that the proportion of failures caused by diameter deformation in historical failures is higher. The contribution of deformation parameters to the overall health is obtained by linear weighting: U1=0.6×0.875+0.4×0.15=0.585.

[0149] For scaling parameters, the coupling degree function expression is as follows: And satisfy Where n represents the number of scaling parameters. For example, taking n=2 corresponds to two dimensions: scaling thickness and scaling non-uniformity. For each scaling dimension, such as scale thickness, the performance function value is provided. =0.333, non-uniformity =0.333, The second benchmark weight is determined based on historical fault data using the entropy weighting method, such as the weight of scale thickness. =0.7, non-uniformity weight =0.3, reflecting that the proportion of failures caused by scale thickness in historical failures is higher. The contribution of scale parameters to overall health is obtained by linear weighting: U2 = 0.7 × 0.333 + 0.3 × 0.333 ≈ 0.333. When either U1 or U2 is not 0, the coupling function expression between deformation and scale parameters is:

[0150] The coupling degree C belongs to a closed interval between 0 and 1. If both U1 and U2 are 0, then C=0. Taking U1=0.585 and U2=0.333 as an example, substituting... The result is C≈0.985, which reflects a high degree of synergistic influence between deformation and scaling parameters on equipment health.

[0151] The coupling function ensures the objectivity of the weights through the entropy weighting method. The linear weighting and coupling formula realize the accurate quantification of the individual contributions and synergistic effects of deformation and scaling parameters, providing standardized coupling relationship data for subsequent health assessments and effectively improving the accuracy and precision of equipment anomaly early warning.

[0152] The calculation logic for the health assessment value is as follows, and the expression for the health assessment value is:

[0153] ;

[0154] Based on the deformation anomaly level and scaling anomaly level in the first dynamic mapping relationship, , Assign dynamic correction coefficients respectively , ;

[0155] For each level increase in the deformation anomaly grade, Increase by 0.1, The interval is a closed interval from 1 to 1.3;

[0156] For each level increase in the scaling abnormality level, Increase by 0.1, The interval is a closed interval from 1 to 1.3;

[0157] Where T is the decay coefficient of the reactor operating time, and the expression for T is:

[0158] ;

[0159] Where T is a double-closed interval from 0.8 to 1;

[0160] For the second dynamic mapping relationship, the measured values ​​of scaling parameters are set as the baseline value of 0, and the weights of each dimension of the scaling parameters are... Set the corresponding second benchmark weight. When the scaling parameter is a single dimension, the weight is 1. Calculate according to the health assessment value calculation logic to obtain H.

[0161] The health assessment value H indicates that the lower the value, the worse the operating condition of the reactor.

[0162] In one embodiment, the health assessment value is used to comprehensively quantify the operating status of the reactor, and its calculation logic is as follows: The health assessment value expression is:

[0163] ;

[0164] Wherein, U1 is the contribution of deformation parameters to overall health, such as U1=0.585 in the previous example; U2 is the contribution of scaling parameters to overall health, such as U2≈0.333 in the previous example; C is the coupling degree between deformation and scaling parameters, such as C≈0.985 in the previous example; α and β are dynamic correction coefficients, set based on the abnormality levels of deformation and scaling. For each level increase in abnormality level, α or β increases by 0.1, with a value range of 1 to 1.3. For example, α=1.1 for moderate deformation abnormality and β=1.1 for moderate scaling abnormality. The setting is based on the gradient matching between abnormality level and failure risk; T is the reactor operating time decay coefficient, expressed as follows: The value ranges from 0.8 to 1. For example, when the design life is 10 years and the cumulative operation is 5 years, The health assessment is based on the positive correlation between the operating time of the reactor equipment and the risk of aging failure. Taking specific values ​​as an example, substituting U1=0.585, U2≈0.333, C≈0.985, α=1.1, β=1.1, and T=0.9 into the formula, we calculate H=100×[(0.585×1.1+0.333×1.1)×(1−0.985)+(0.585×0.333×0.985)]×0.9≈18.9. The lower the health assessment value H, the worse the operating condition of the reactor. This health assessment logic integrates the abnormal level risk and the decay coefficient to reflect the impact of equipment aging through dynamic correction coefficients. Combined with the synergistic effect of coupled quantitative parameters, it achieves a multi-dimensional, dynamic, and accurate assessment of the reactor's operating status. This provides a quantitative basis for graded decision-making for abnormal early warning and effectively improves the foresight and precision of equipment operation and maintenance.

[0165] The health assessment value H is divided into four intervals: the first interval, the second interval, the third interval, and the fourth interval.

[0166] The first interval is where the H value falls within the double-closed interval from the first value to the second value, corresponding to the first-level abnormal warning level;

[0167] The second interval is where the H value falls within the double-closed interval between the second and third values, corresponding to a level two anomaly warning level.

[0168] The third interval is where the H value falls within the double-closed interval between the third and fourth values, corresponding to a level three anomaly warning level.

[0169] The fourth interval is where the H value falls within the double-closed interval between the fourth and fifth values, corresponding to the fourth level of abnormal warning.

[0170] If it is a Level 1 warning, a red audible and visual alarm signal will be generated to trigger the emergency shutdown preparation procedure of the reactor in advance, and an emergency maintenance SMS instruction with fault location and trend prediction will be sent to the operation and maintenance personnel.

[0171] If it is a Level II warning, a yellow audible and visual alarm signal will be generated, triggering the reactor load reduction plan in advance, and sending a maintenance reminder SMS instruction with abnormal parameter tracing and development trend analysis to the operation and maintenance personnel;

[0172] If it is a Level 3 warning, a blue audible and visual alert signal will be generated, and an abnormal warning email and planned inspection suggestions will be sent to the operation and maintenance personnel in advance, prompting them to pay attention to the operating status of the reactor and start preventive data monitoring.

[0173] If there is no abnormality level, no abnormality warning signal will be generated. Only the health data of the reactor will be recorded for historical analysis and early warning model optimization and iteration.

[0174] In one embodiment, to achieve graded early warning of reactor anomalies, the health assessment value H is divided into four intervals corresponding to different warning levels. The specific logic is as follows: The first interval is set as a double-closed interval where the H value is between 0 and 20, corresponding to the first-level anomaly warning level; the second interval is set as a double-closed interval where the H value is between 20 and 50, corresponding to the second-level anomaly warning level; the third interval is a double-closed interval where the H value is between 50 and 80, corresponding to the third-level anomaly warning level; and the fourth interval is a double-closed interval where the H value is between 80 and 100, corresponding to the fourth-level no-anomaly level. The interval setting is based on the gradient matching between the health value and the reactor equipment failure risk. The lower the H value, the worse the operating status and the higher the failure risk. For example, if the health assessment value H ≈ 18.9, it belongs to the first interval, and a first-level warning is triggered: a red audible and visual alarm signal is generated, the reactor emergency shutdown preparation procedure is triggered in advance, and an emergency maintenance SMS instruction with fault location, such as the abnormal area of ​​diameter deformation and trend prediction, is sent to the operation and maintenance personnel; if H = If H=35, it falls into the second range, triggering a level-two warning: a yellow audible and visual alarm signal is generated, the reactor load reduction plan is initiated in advance, and a maintenance reminder SMS is sent to maintenance personnel with abnormal parameter tracing, such as excessive scale thickness and development trend analysis. If H=60, it falls into the third range, triggering a level-three warning: a blue audible and visual alert signal is generated, and an abnormality warning email and planned inspection suggestions are sent to maintenance personnel in advance, prompting them to pay attention to the reactor's operating status and initiate preventive data monitoring. If H=90, it falls into the fourth range, which is determined to be an abnormality level, no abnormality warning signal is generated, and only the reactor health data is recorded for historical analysis and proactive optimization and iteration of the warning model. This graded warning logic, by binding clear numerical ranges with risk levels and combining differentiated audible and visual signals, handling procedures, and notification methods, achieves accurate grading and rapid response to reactor abnormalities, effectively improving the timeliness of equipment operation and maintenance and the pertinence of fault handling, and significantly reducing equipment failure losses caused by untimely abnormality handling.

[0175] This invention solves the problem of the one-sidedness of single-parameter detection by multi-dimensional collaborative matching of deformation and scaling images, and realizes comprehensive identification of equipment anomalies. Based on the entropy weight method, it constructs a coupling degree function and dynamic health assessment system, and combines anomaly level correction coefficient and runtime attenuation coefficient to accurately quantify the overall health status of the reactor, greatly improving the dynamism and accuracy of the assessment. Furthermore, through a graded early warning strategy, it triggers targeted disposal measures based on health differences, effectively reducing the risk of unplanned downtime and equipment failure, lowering operation and maintenance costs, and ultimately significantly improving the reliability and intelligent operation and maintenance level of petrochemical equipment.

[0176] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A petrochemical equipment abnormality early warning system, characterized in that, Includes a matching module and an early warning module; The matching module calls a preset deformation anomaly rule to perform a first match on the deformation parameters. If the first match result is abnormal, it calls a preset image anomaly rule to perform a second match on the inner wall image data. If the first match result is without anomaly, it only outputs the first match result. The early warning module constructs a dynamic mapping relationship based on the first matching result and the second matching result, obtains a health assessment value through coupling analysis, generates a graded abnormality early warning signal based on the health assessment value, and executes the corresponding level of early warning operation. Based on historical internal wall image data of the reactor, scaling parameters are extracted, and image anomaly rules are constructed according to the scaling parameters. The configuration logic of the image anomaly rule is as follows: extract scaling parameters based on historical inner wall image data, and construct image anomaly rules based on scaling parameters. Specifically, the process includes: using an adaptive threshold segmentation algorithm to segment the historical inner wall image data to obtain the scaling area and the inner wall background area; and using the Canny edge detection algorithm to extract the contour features of the scaling area. By combining contour features and image pixel scale, the scale thickness, scale area ratio, and scale distribution density are calculated. Scaling parameters can be configured as a single dimension or a combination of multiple dimensions; Based on the numerical range of scale thickness and scale area ratio, the first abnormal feature quantity is obtained. The first abnormal feature quantity includes the scale thickness exceeding the standard coefficient and the scale area ratio exceeding the limit value. The second abnormal characteristic quantity is obtained from the scale distribution density in the scale parameters. The second abnormal characteristic quantity is the proportion of dense scale area and the scale unevenness. The first and second abnormal features are combined to obtain the judgment feature, which is then converted into the second rule threshold. After being set according to the scaling parameter safety standard, the image anomaly rule is constructed. The dynamic mapping relationship includes a first dynamic mapping relationship and a second dynamic mapping relationship; The logic for constructing the first dynamic mapping relationship is as follows: when the matching result includes the first matching result and the second matching result, the deformation parameter, the first threshold comparison difference and the abnormal identifier in the first matching result are used as the first mapping dimension, and the scaling parameter, the second threshold comparison difference and the abnormal identifier in the second matching result are used as the second mapping dimension. The first dynamic mapping relationship is constructed based on the first mapping dimension and the second mapping dimension. The second dynamic mapping relationship construction logic is as follows: when the matching result only includes the first matching result, the second matching result corresponding to the image anomaly rule that has not been called is set as the default result; The default result includes a scaling parameter of 0, an anomaly identifier of no anomaly, and a second threshold comparison difference of 0. The deformation parameter, the first threshold comparison difference, and the anomaly identifier in the first matching result are used as the main dimension, and the default result is used as the auxiliary dimension to construct a second dynamic mapping relationship. The coupling analysis includes constructing a power function and a coupling degree function; The coupling degree function is constructed using the following logic: the coupling degree function expression for the deformation parameter is as follows: ； In the first dynamic mapping relationship, m represents the m dimensions of deformation parameters. This indicates that the deformation parameters contribute to the overall health by linear weighting. Let represent the first baseline weight of the j-th dimension of the corresponding deformation parameter. This first baseline weight is determined from historical fault data using the entropy weight method and satisfies the following: ； The coupling degree function expression for the scaling parameters is: ； In the first dynamic mapping relationship, n represents n dimensions of scaling parameters. This indicates the contribution of scaling parameters to overall health, obtained through linear weighting. This represents the scaling efficiency function value corresponding to the Kth scaling parameter. Let the second baseline weight be the k-th dimension of the corresponding scaling parameter. This second baseline weight is determined from historical fault data using the entropy weight method and satisfies the following: ； when and When any one of the values ​​is not zero, the coupling function expression between the deformation parameter and the scaling parameter is: ； Wherein, the coupling degree C is a closed interval from 0 to 1; when and When all values ​​are 0, the coupling function between the deformation parameter and the scaling parameter is 0.

2. The petrochemical equipment operation anomaly early warning system as described in claim 1, characterized in that, The deformation anomaly rules are constructed based on the historical deformation parameters of the reactor. The configuration logic of the deformation anomaly rule includes substituting historical deformation parameters and preset benchmark deformation parameters into the cosine similarity formula to calculate the similarity value. Abnormal deformation parameters are filtered out based on the similarity value. The filtering logic is that historical deformation parameters with similarity values ​​lower than a preset threshold are judged as abnormal deformation parameters. A first rule threshold is set based on the abnormal deformation parameters, and an abnormal deformation rule is constructed based on the first rule threshold.

3. The petrochemical equipment operation anomaly early warning system as described in claim 2, characterized in that, The deformation parameters are first matched by calling the deformation anomaly rule. If the first matching result is abnormal, the image anomaly rule is called to perform a second matching on the inner wall image data to obtain the second matching result. Specifically, the deformation parameters of the reactor are obtained, the preset deformation anomaly rule is called, the deformation parameters are compared with the first rule threshold in the deformation anomaly rule, and if the deformation parameters exceed the first rule threshold, the first matching result is determined to be abnormal. The first matching result including the deformation parameters, the difference between the first threshold and the anomaly identifier is output. The first threshold comparison difference is the difference between the deformation parameter and the first rule threshold. Synchronously acquire image data of the inner wall of the reactor, call the preset image anomaly rules, extract scaling parameters from the inner wall image data based on the image anomaly rules, and compare the scaling parameters with the second rule threshold in the image anomaly rules; If the scaling parameter exceeds the second rule threshold, the second matching result is determined to be abnormal, and the output includes the scaling parameter, the difference between the second threshold and the abnormality indicator of the second matching result. The second threshold comparison difference is the difference between the scaling parameters and the second rule threshold. If the scaling parameters do not exceed the threshold of the second rule, the second matching result is determined to be without abnormality, and the output includes the scaling parameters and the no abnormality indicator of the second matching result.

4. The petrochemical equipment operation anomaly early warning system as described in claim 3, characterized in that, If the first matching result is no anomaly, the call to the anomaly rule is terminated, and only the first matching result is output. Specifically, if the deformation parameter does not exceed the threshold of the first rule, the first matching result is determined to be no anomaly, the call process of the image anomaly rule is directly terminated, and only the first matching result including the deformation parameter and the no anomaly indicator is output.

5. The petrochemical equipment operation anomaly early warning system as described in claim 4, characterized in that, The deformation anomaly level is first divided based on the first threshold comparison difference to obtain different deformation anomaly levels. The first division logic is as follows: the first threshold comparison difference in the left-closed-right-open interval from 0 to A corresponds to the mild deformation anomaly level; the left-closed-right-open interval from A to B corresponds to the moderate deformation anomaly level; and the left-closed-right-open interval from B to +∞ corresponds to the severe deformation anomaly level. A and B are both preset positive numbers, and A is less than B. The scaling anomaly level is divided into two categories based on the comparison difference of the second threshold. The second classification logic is as follows: the left-closed and right-open interval between 0 and C corresponds to the mild scaling anomaly level; the left-closed and right-open interval between C and D corresponds to the moderate scaling anomaly level; and the left-closed and right-open interval between D and +∞ corresponds to the severe scaling anomaly level. C and D are both preset positive numbers, and C is less than D. The first dynamic mapping relationship includes the dynamic mapping relationship between deformation anomaly level and scaling anomaly level, specifically including mild deformation anomaly level corresponding to mild scaling anomaly level, moderate deformation anomaly level corresponding to moderate scaling anomaly level, and severe deformation anomaly level corresponding to severe scaling anomaly level.

6. The petrochemical equipment operation anomaly early warning system as described in claim 5, characterized in that, The first dynamic mapping relationship and the second dynamic mapping relationship use the anomaly dimension as the classification benchmark, and associate the parameter values ​​of the corresponding dimension to form a structured relationship, providing standardized data input for subsequent coupling analysis; The abnormality dimensions include deformation dimension and scaling dimension; The parameter values ​​include deformation parameters, scaling parameters, first threshold comparison difference and second threshold comparison difference, and are also bound to the corresponding dimension of the anomaly identifier.

7. The petrochemical equipment operation anomaly early warning system as described in claim 6, characterized in that, The logic for constructing the performance function is as follows: when the deformation parameter has a positive effect, the expression of the deformation performance function is: ； in, This represents the deformation efficiency function value corresponding to the j-th deformation parameter. This represents the measured value of the j-th dimension of the deformation parameter. , These represent the lower and upper risk limits for the j-th dimension of the deformation parameter, respectively. These are preset constants used to standardize the efficacy function benchmark value when deformation parameters and scaling parameters reach health goals; When the deformation parameter has a negative effect, the expression for the deformation effect function is: ； When the scaling parameters have a positive effect, the expression for the scaling effect function is: ； in, This represents the scaling efficiency function value corresponding to the Kth scaling parameter. This represents the measured value of the Kth dimension of the scaling parameters. , These represent the lower and upper risk limits for the Kth dimension of the scaling parameter, respectively. When the scaling parameter has a negative effect, the expression for the scaling effect function is: 。 8. The petrochemical equipment operation anomaly early warning system as described in claim 7, characterized in that, The calculation logic for the health assessment value is as follows, and the expression for the health assessment value is: ； Based on the deformation anomaly level and scaling anomaly level in the first dynamic mapping relationship, , Assign dynamic correction coefficients respectively , ; For each level increase in the deformation anomaly grade, Increase by 0.1, the aforementioned The interval is a closed interval from 1 to 1.3; For each level increase in the scaling abnormality level, Increase by 0.1, the aforementioned The interval is a closed interval from 1 to 1.3; Where T is the decay coefficient of the reactor operating time, and the expression for T is: ； Where T is a double-closed interval from 0.8 to 1; For the second dynamic mapping relationship, the measured values ​​of scaling parameters are set as the baseline value of 0, and the weights of each dimension of the scaling parameters are... Set the corresponding second benchmark weight. When the scaling parameter is a single dimension, the weight is 1. Calculate according to the health assessment value calculation logic to obtain H. The health assessment value H indicates that the lower the value, the worse the operating condition of the reactor.

9. The petrochemical equipment operation anomaly early warning system as described in claim 8, characterized in that, The health assessment value H is divided into four intervals: the first interval, the second interval, the third interval, and the fourth interval. The first interval is where the H value belongs to the double-closed interval from the first value to the second value, corresponding to the first level of abnormal warning level; The second interval is where the H value falls within the double-closed interval between the second and third values, corresponding to a level two anomaly warning level; The third interval is where the H value belongs to the double-closed interval between the third and fourth values, corresponding to the third level of abnormal warning. The fourth interval is where the H value belongs to the double-closed interval between the fourth and fifth values, corresponding to the fourth level of abnormal warning level; If it is a Level 1 warning, a red audible and visual alarm signal will be generated to trigger the emergency shutdown preparation procedure of the reactor in advance, and an emergency maintenance SMS instruction with fault location and trend prediction will be sent to the operation and maintenance personnel. If it is a Level II warning, a yellow audible and visual alarm signal will be generated, triggering the reactor load reduction plan in advance, and sending a maintenance reminder SMS instruction with abnormal parameter tracing and development trend analysis to the operation and maintenance personnel; If it is a Level 3 warning, a blue audible and visual alert signal will be generated, and an abnormal warning email and planned inspection suggestions will be sent to the operation and maintenance personnel in advance, prompting them to pay attention to the operating status of the reactor and start preventive data monitoring. If there is no abnormality level, no abnormality warning signal will be generated. Only the health data of the reactor will be recorded for historical analysis and early warning model optimization and iteration.

Images