An abnormal alarm system and control method for a high-temperature resistant safety valve

Abstract

This invention relates to the field of safety control technology, specifically to an abnormal alarm system and control method for a high-temperature resistant safety valve. The method includes: acquiring historical pressure change curves and a set of triggering causes; determining historical principal component vectors and historical eigenvalues; performing similarity clustering to obtain clusters; determining representative change curves for each triggering cause based on the same cluster; acquiring the current pressure change curve and determining the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling times; determining the cause of the anomaly based on the similarity between each current principal component vector and each representative change curve; determining the urgency level of the warning based on the frequency of the anomaly cause and the degree of pressure fluctuation anomaly; and triggering an abnormal alarm based on the urgency level of the warning and the pressure fluctuation trend. This invention enables detailed analysis of multiple triggering causes, timely detection and handling of anomalies, and improves the accuracy of safety valve alarm control.

Description

Technical Field

[0001] This invention relates to the field of safety control technology, specifically to an abnormal alarm system and control method for a high-temperature resistant safety valve. Background Technology

[0002] High-temperature resistant safety valves are safety pressure relief devices specifically designed for high-temperature environments. Through automatic opening and closing, they rapidly release pressure when the medium pressure exceeds a set value, preventing equipment or pipelines from malfunctioning due to overpressure. Once the pressure returns to normal, they automatically reseat and seal. Compared to ordinary safety valves, their materials and structural design are more resistant to material degradation and performance decline under high-temperature environments. They typically operate within a temperature range of 300℃ to 800℃ and are widely used in power energy, petrochemical, metallurgical, and building materials industries. As key equipment ensuring the safe operation of these systems, the proper functioning of high-temperature resistant safety valves is crucial.

[0003] Currently, most alarm systems only respond after the safety valve has tripped, which is a reactive measure. Pre-trip warnings are based on analysis of a single tripping cause. However, safety valve tripping can have multiple causes, and these causes can interfere with each other to some extent. As a result, the pressure fluctuations of these causes are inconsistent with any single tripping cause, making it impossible to match the warning cause in a timely manner based on analysis of a single tripping cause. Consequently, the accuracy of safety valve alarm control is insufficient. Summary of the Invention

[0004] To address the technical problem in related technologies where multiple tripping causes can lead to mutual interference, resulting in inconsistent pressure fluctuations with any single tripping cause, and making it impossible to timely match early warning causes through analysis of a single tripping cause, thus leading to insufficient accuracy in safety valve alarm control, this invention provides an abnormal alarm system and control method for high-temperature resistant safety valves. The specific technical solution adopted is as follows:

[0005] This invention proposes an abnormal alarm control method for a high-temperature resistant safety valve, the method comprising:

[0006] Obtain historical pressure change curves for different safety valve trips within a preset time period, as well as a set of tripping reasons for each trip;

[0007] Determine the historical principal component vectors and historical eigenvalues ​​of different historical pressure change curves; perform similarity clustering on all historical principal component vectors to obtain clusters; based on all historical principal component vectors and corresponding historical eigenvalues ​​within the same cluster, determine the representative change curve for each cause of jump;

[0008] Obtain the current pressure change curve for a preset time period prior to the current moment. Determine the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling moments in the current pressure change curve. Determine the current principal component vector and current eigenvalue of the current pressure change curve. Determine the cause of the anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve and the current eigenvalue.

[0009] Based on the frequency of occurrence of all abnormal causes in history and the degree of abnormal pressure fluctuations, the urgency level of the warning at the current moment is determined; based on the urgency level of the warning and the pressure fluctuation trend of the current pressure change curve, an abnormal alarm is issued.

[0010] Furthermore, the similarity clustering of all historical principal component vectors to obtain clusters includes:

[0011] Based on the dynamic time warping algorithm, the DTW value of any two historical principal component vectors is calculated, and the negative of the DTW value is normalized as the curve similarity.

[0012] Clustering is performed based on the curve similarity, and historical principal component vectors whose curve similarity is greater than a preset similarity threshold are regarded as the same cluster.

[0013] Furthermore, determining the representative change curve for each cause of a jump based on all historical principal component vectors within the same cluster and their corresponding historical eigenvalues ​​includes:

[0014] Within each cluster, the most frequent jump cause corresponding to the historical pressure change curve of the historical principal component vector is taken as the cluster cause of the cluster.

[0015] The historical principal component vectors within each cluster are weighted according to their corresponding historical feature values ​​to obtain the representative change curve of the cluster cause. By traversing all clusters, the representative change curve of each jump cause is obtained.

[0016] Furthermore, determining the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling times in the current pressure change curve includes:

[0017] Based on the discrete distribution characteristics of pressure values ​​at different sampling times in the current pressure change curve, the pressure dispersion index is determined.

[0018] The pressure difference between the next sampling time and the previous sampling time is calculated as a pressure increase index. The maximum value of the pressure increase index is normalized and used as a pressure change index.

[0019] The product of the pressure dispersion index and the pressure change index is normalized and used as the degree of pressure fluctuation anomaly.

[0020] Furthermore, determining the cause of the anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve, and the current eigenvalue, includes:

[0021] Calculate the Pearson correlation coefficient between each current principal component vector and each representative change curve, and normalize the Pearson correlation coefficient as the degree of correlation.

[0022] The jump cause corresponding to the representative change curve with the highest correlation value of each current principal component vector is taken as the principal component cause;

[0023] The principal component cause of the preset number of principal component vectors with the largest current eigenvalue is taken as the anomaly cause.

[0024] Furthermore, determining the current urgency level of the warning based on the frequency of occurrence of all abnormal causes in history and the degree of abnormal pressure fluctuations includes:

[0025] The frequency of all abnormal causes in history is normalized and used as a frequency index.

[0026] The product of the frequency index and the degree of pressure fluctuation anomaly is calculated and normalized to determine the urgency level of the warning at the current moment.

[0027] Furthermore, based on the aforementioned urgency level of the warning and the pressure fluctuation trend of the current pressure change curve, an abnormal alarm is triggered, including:

[0028] Determine the urgency of the trend based on the pressure trend of the current pressure change curve;

[0029] Calculate the product of the warning urgency level and the trend urgency level, and normalize it to obtain the anomaly coefficient;

[0030] An anomaly alarm is triggered when the anomaly coefficient exceeds a preset anomaly threshold.

[0031] Furthermore, based on the pressure trend of the current pressure change curve, the urgency of the trend is determined, including:

[0032] The pressure change curve is fitted with a straight line to obtain a fitted straight line. The slope of the fitted straight line is normalized and used as the degree of urgency of the trend.

[0033] Furthermore, the preset abnormal threshold is 0.7.

[0034] On the other hand, an abnormal alarm system for a high-temperature resistant safety valve is also provided. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the method as described in any of the foregoing.

[0035] The present invention has the following beneficial effects:

[0036] This invention employs principal component analysis on different historical pressure change curves to distinguish different triggering causes, performs similarity clustering, and obtains representative change curves based on the clustering results. Compared to directly obtaining representative change curves based on a single triggering cause analysis, this invention introduces mutual factor analysis between multiple triggering causes, making the representative change curves more accurately and objectively represent the actual triggering cause changes. Then, by combining the pressure value distribution of the current pressure change curve and the similarity between each current principal component vector and each representative change curve, the abnormal cause at the current moment is determined. Combining the frequency of the abnormal cause's occurrence in history and the degree of pressure fluctuation abnormality, the urgency of the warning is determined. Furthermore, by combining the pressure fluctuation trend, an abnormal alarm is implemented. Thus, this invention can perform detailed analysis of multiple triggering causes, combine current pressure changes and historical change data, provide early warning of safety valve anomalies at the current moment, promptly detect and handle anomalies, and improve the accuracy of safety valve alarm control. Attached Figure Description

[0037] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 A flowchart illustrating an abnormal alarm control method for a high-temperature safety valve according to an embodiment of the present invention;

[0039] Figure 2 This is a schematic diagram of a safety valve provided in one embodiment of the present invention. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an abnormal alarm system and control method for a high-temperature safety valve according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] The following description, in conjunction with the accompanying drawings, details the specific scheme of the abnormal alarm control method for a high-temperature safety valve provided by the present invention.

[0043] Please see Figure 1 The diagram illustrates a flowchart of an abnormal alarm control method for a high-temperature safety valve according to an embodiment of the present invention. The method includes:

[0044] S101: Obtain the historical pressure change curves for different safety valve trips within a preset time period, as well as the set of tripping reasons for each trip.

[0045] High-temperature resistant safety valves are safety pressure relief devices specifically designed for high-temperature environments. Through automatic opening and closing, they rapidly release pressure when the medium pressure exceeds a set value, preventing equipment or pipelines from malfunctioning due to overpressure. Once the pressure returns to normal, they automatically reseat and seal. Compared to ordinary safety valves, their materials and structural design are more resistant to material degradation and performance decline under high-temperature environments. They typically operate within a temperature range of 300℃ to 800℃ and are widely used in power energy, petrochemical, metallurgical, and building materials industries. As key equipment ensuring the safe operation of these systems, the proper functioning of high-temperature resistant safety valves is crucial.

[0046] Currently, most alarm systems only respond after the safety valve has tripped, which is a reactive measure. Pre-warning is based on analysis of a single tripping cause. However, safety valve tripping can have multiple causes, and these causes can interfere with each other to some extent. As a result, the pressure fluctuations of these causes are inconsistent with any single tripping cause, making it impossible to match the warning cause in a timely manner. Consequently, the accuracy of safety valve alarm control is insufficient.

[0047] In this embodiment of the invention, pressure changes during the historical safety valve tripping process are effectively analyzed to provide early warning, transforming passive pressure relief into proactive overpressure prevention. This avoids the impact of pressure fluctuations on the production process and improves production stability. (See also...) Figure 2 , Figure 2This is a schematic diagram of a safety valve provided in one embodiment of the present invention.

[0048] The preset time period is the pressure collection time period. In this embodiment of the invention, the time range of historical data is the number of times the safety valve trips in the past year, the reasons for tripping, and the pressure data in the ten minutes before each trip. The sampling time for the pressure data can be once every 10 seconds, that is, the preset time period is 10 minutes. The pressure data within 10 minutes is curve fitted according to the time sequence to obtain the historical pressure change curve. Curve fitting is well known to those skilled in the art and will not be further limited or described.

[0049] In the fractionation system of a large-scale oil refinery catalytic cracking unit, due to the process characteristics of the fractionation system, gas-liquid balance, heat transfer, and material flow are interconnected. Fluctuations in any parameter can cause pressure anomalies. In actual production, the tripping of high-temperature safety valves may be caused by the direct effect of a single strong interference factor, such as condenser cooling failure leading to the inability of the gas phase to condense and a sharp increase in pressure; or it may be caused by the superposition of multiple risk factors that exceed the system's pressure threshold. For example, while scale buildup on the trays hinders the flow of the liquid phase, the surge in reboiler heating exacerbates the gas phase load. Multiple effects work together to cause overpressure, causing the safety valve to trip. Therefore, in order to provide early warning of abnormalities in high-temperature safety valves, it is necessary to first analyze the causes of historical tripping.

[0050] Different triggers have different causes. A single trigger may be caused by a single cause or by multiple causes, resulting in a set of trigger causes for each trigger event. For example, the trigger cause for the xth historical trigger is {condenser scaling}; the trigger cause for the yth historical trigger is {condenser scaling, sudden increase in feed rate}; and the trigger cause for the zth historical trigger is {sudden increase in feed rate, water carried over to the feed}. For simplicity, each trigger cause will be represented by a letter, i.e., the set of trigger causes for the xth historical trigger is {a}; the set of trigger causes for the yth historical trigger is {a, b}; and the set of trigger causes for the zth historical trigger is {b, c}.

[0051] S102: Determine the historical principal component vectors and historical eigenvalues ​​of different historical pressure change curves; perform similarity clustering on all historical principal component vectors to obtain clusters; based on all historical principal component vectors and corresponding historical eigenvalues ​​within the same cluster, determine the representative change curve for each cause of jump.

[0052] Different triggering causes will result in different triggering pressure performance of the safety valve. Analysis can be carried out based on these differences. Triggering events induced by multiple triggering causes have different effects on each triggering event. Therefore, it is necessary to determine the degree of dominance of each triggering cause in multi-triggering events, so as to determine the representative change curve under each triggering cause in the dimension of a single triggering cause.

[0053] In this embodiment of the invention, principal component decomposition (PCD) can be used to decompose the stress change curve of a multi-cause event into multiple principal component vectors: PC1, PC2, ..., PCn, where n represents the number of principal component vectors obtained after decomposition. The eigenvalues ​​of each principal component vector are determined, thereby obtaining historical principal component vectors and historical eigenvalues ​​from the historical stress change curve. The principal component vectors and eigenvalues ​​obtained by PCD are well known to those skilled in the art. For the stress change curve of a single-cause event, it is directly denoted as a single principal component vector, and the historical eigenvalue is set to its maximum value; no further limitations or elaborations are made regarding this.

[0054] The principal component vectors are the corresponding sub-curves, which can characterize the principal component features, that is, the pressure change features corresponding to the jump cause. However, since the jump causes in real-world scenarios are complex and pressure changes are diverse, it is necessary to combine all pressure change features under the same jump cause for unified analysis. First, cluster analysis of the same jump cause is required.

[0055] Furthermore, in some embodiments of the present invention, similarity clustering is performed on all historical principal component vectors to obtain clusters, including: calculating the DTW value of any two historical principal component vectors based on the dynamic time warping algorithm, normalizing the negative of the DTW value as curve similarity; clustering is performed based on curve similarity, and historical principal component vectors with curve similarity greater than a preset similarity threshold are regarded as the same cluster.

[0056] The Dynamic Time Warping (DTW) algorithm is used to determine the similarity between two time series. The smaller the DTW value, the more similar the two time series are. In the scenario of this embodiment, the DTW value of any two historical principal component vectors is calculated, and the negative of the DTW value is normalized as the curve similarity. The larger the curve similarity value, the more consistent the numerical fluctuations between the two historical principal component vectors are.

[0057] In one embodiment of the present invention, the normalization process can be specifically, for example, maximum and minimum value normalization. Furthermore, the normalization in subsequent steps can all adopt maximum and minimum value normalization. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of the numerical values, which will not be elaborated further.

[0058] The preset similarity threshold is a threshold value for curve similarity. In this embodiment of the invention, the preset similarity threshold can be, for example, 0.75. That is, historical principal component vectors with curve similarity greater than 0.75 are regarded as the same cluster, and different clusters are obtained by traversing all historical principal component vectors.

[0059] Therefore, the fluctuation similarity characteristics within each cluster are relatively large, indicating the same initial cause. Based on this, weighted analysis can be performed to obtain representative change curves under each initial cause, and the weights can be determined using historical feature values.

[0060] Furthermore, in some embodiments of the present invention, the representative change curve for each jump cause is determined based on all historical principal component vectors within the same cluster and their corresponding historical feature values. This includes: within each cluster, the jump cause with the highest frequency corresponding to the historical pressure change curve of the historical principal component vector is taken as the cluster cause of the cluster; the historical principal component vectors within each cluster are weighted according to their corresponding historical feature values ​​to obtain the representative change curve of the cluster cause; and all clusters are traversed to obtain the representative change curve for each jump cause.

[0061] Among them, the historical pressure change curve of the historical principal component vector in each cluster corresponds to the most frequent jump cause, which is used as the cluster cause of the cluster. If two clusters have the same cluster cause at the same time, the two clusters are merged into a new cluster. That is, the principal component vectors contained in each cluster are similar, thus indicating that each cluster corresponds to a jump cause.

[0062] In this embodiment of the invention, the historical principal component vectors within each cluster are weighted according to their corresponding historical feature values ​​to obtain a representative change curve for the cluster cause. The historical feature values ​​can be normalized to their maximum and minimum values ​​and used as the weights of the corresponding historical principal component vectors. The historical principal component vector of a single jump cause's historical pressure change curve is itself, with a weight of 1. Thus, by integrating all historical principal component vectors within the same cluster and performing unified feature analysis and extraction, a representative change curve is obtained. This representative change curve effectively characterizes the overall pressure change features of the corresponding jump cause.

[0063] Based on the historical feature values ​​of all historical principal component vectors contained in each cluster, they are weighted and fused to form a comprehensive pressure change curve, which is used to represent the pressure change curve of the safety valve in the ten minutes before a certain cause triggers an abnormal alarm. This curve is recorded as the representative pressure change curve of that historical cause.

[0064] S103: Obtain the current pressure change curve for a preset time period prior to the current moment; determine the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling moments in the current pressure change curve; determine the current principal component vector and current eigenvalue of the current pressure change curve; and determine the cause of the anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve and the current eigenvalue.

[0065] Under normal circumstances, the pressure at the top of the fractionation column in the fractionation system remains in dynamic equilibrium within the design range, but small, regular fluctuations are allowed. When abnormalities occur in materials, heat, or operation, this equilibrium will be broken, and the pressure will change significantly. Therefore, it is necessary to first determine the pressure change characteristics of a period of time before the current moment (similar to historical data acquisition, the pressure value of the previous ten minutes is used).

[0066] After analyzing all historical data, it is necessary to conduct anomaly analysis by considering the pressure changes under the current conditions. First, it is necessary to perform fluctuation analysis on the pressure changes over a preset time period prior to the current moment to obtain the degree of pressure fluctuation anomaly. The degree of pressure fluctuation anomaly characterizes the abnormality index of the pressure fluctuation.

[0067] Furthermore, in some embodiments of the present invention, determining the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling times in the current pressure change curve includes: determining a pressure dispersion index based on the pressure value dispersion distribution characteristics at different sampling times in the current pressure change curve; calculating the pressure value difference between the next sampling time and the previous sampling time as a pressure increase index; normalizing the maximum value of the pressure increase index as a pressure change index; and normalizing the product of the pressure dispersion index and the pressure change index as the degree of pressure fluctuation anomaly.

[0068] The pressure dispersion index represents the dispersion characteristics of the pressure value distribution itself. In this embodiment of the invention, existing dispersion calculation methods, such as variance and standard deviation, can be used for dispersion characteristic analysis without limitation. Specifically, the variance of the pressure values ​​at different sampling times in the current pressure change curve can be calculated as the pressure dispersion index.

[0069] As the pressure value gradually increases, it will gradually approach the start-up process. Therefore, in this embodiment of the invention, the pressure value difference between the next sampling time and the previous sampling time is calculated as the pressure increase index. The maximum value of the pressure increase index is normalized and used as the pressure change index. The larger the value of the pressure change index, the greater the instantaneous pressure increase. The product of the pressure dispersion index and the pressure change index is normalized and used as the degree of pressure fluctuation anomaly.

[0070] The larger the numerical value of the abnormal pressure fluctuation, the more abnormal the numerical distribution and instantaneous increase, and the more likely it is to trigger the safety valve to trip.

[0071] Furthermore, in some embodiments of the present invention, determining the cause of anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve and the current eigenvalue includes: calculating the Pearson correlation coefficient between each current principal component vector and each representative change curve, and normalizing the Pearson correlation coefficient as the degree of correlation; taking the jump cause corresponding to the representative change curve with the largest correlation value for each current principal component vector as the principal component cause; and taking the principal component causes of a preset number of principal component vectors with the largest current eigenvalues ​​as the anomaly causes.

[0072] The Pearson correlation coefficient is a well-known correlation calculation method among those skilled in the art. The Pearson correlation coefficient is normalized to represent the degree of correlation; that is, the higher the degree of correlation, the more significant the correlation between the current principal component vector and the representative change curve. The representative change curve with the highest degree of correlation corresponds to the triggering cause, which is then used as the reason for the safety valve to trip at the current moment.

[0073] Since each principal component vector corresponds to a principal component cause, in order to perform multi-cause analysis at the current moment, the principal component causes of the principal component vectors with the largest current eigenvalues ​​are selected as abnormal causes. Specifically, the selected number can be, for example, 3, meaning that the principal component causes of the 3 principal component vectors with the largest current eigenvalues ​​are selected as abnormal causes.

[0074] S104: Determine the current urgency level of the warning based on the frequency of occurrence of all abnormal causes in history and the degree of abnormal pressure fluctuation; issue an abnormal alarm based on the urgency level of the warning and the pressure fluctuation trend of the current pressure change curve.

[0075] Since the abnormal cause does not indicate a current risk of triggering a jump, further analysis of the current urgency is still needed, taking into account historical data and the degree of abnormal pressure fluctuations.

[0076] Furthermore, in some embodiments of the present invention, the urgency level of the warning at the current moment is determined based on the frequency of occurrence of all abnormal causes in history and the degree of pressure fluctuation abnormality, including: normalizing the frequency of occurrence of all abnormal causes in history as a frequency index; calculating the product of the frequency index and the degree of pressure fluctuation abnormality, and normalizing it as the urgency level of the warning at the current moment.

[0077] The higher the frequency of abnormal causes, the more serious the corresponding abnormal situation. Therefore, in this embodiment of the invention, the frequency of abnormal causes is used as a feature of emergency analysis. The frequency index is obtained by calculating the proportion of the frequency of all abnormal causes in the history to the total frequency of all jump causes.

[0078] The numerical value of the pressure fluctuation anomaly characterizes the degree of abnormality in pressure fluctuations. Therefore, the product of the calculation frequency index and the pressure fluctuation anomaly degree is calculated, normalized, and used as the current warning urgency level. The higher the warning urgency level value, the more likely the abnormal situation is to occur, and the more abnormal the pressure change.

[0079] Furthermore, in some embodiments of the present invention, an abnormal alarm is triggered based on the urgency of the warning and the pressure fluctuation trend of the current pressure change curve, including: determining the urgency of the trend based on the pressure trend of the current pressure change curve; calculating the product of the urgency of the warning and the urgency of the trend, and normalizing it as an abnormal coefficient; and triggering an abnormal alarm when the abnormal coefficient is greater than a preset abnormal threshold.

[0080] At this moment, not only is numerical dispersion analysis necessary, but also analysis of the overall numerical change trend is required to facilitate early warning based on the trend. Based on the pressure trend of the current pressure change curve, the urgency of the trend is determined, including: fitting a straight line to the pressure change curve to obtain a fitted straight line, and normalizing the slope of the fitted straight line as the degree of trend urgency.

[0081] The specific trend is the overall numerical change trend, that is, the trend of increasing or decreasing values. It is characterized by the slope of the fitted straight line. The greater the urgency of the trend, the greater the slope, and the more obvious the upward trend of pressure.

[0082] The product of the urgency level of the warning and the urgency level of the trend is calculated, normalized, and used as the anomaly coefficient. The anomaly coefficient represents a multi-dimensional analysis that combines the current pressure distribution, pressure trend, and pressure fluctuations under different triggering causes, enabling accurate anomaly alarms.

[0083] Specifically, when the abnormality coefficient is greater than the preset abnormality threshold, an abnormality alarm is triggered. The preset abnormality threshold is the threshold value of the abnormality coefficient. In this embodiment of the invention, the preset abnormality threshold is 0.7. That is, when the abnormality coefficient is greater than 0.7, it means that the more abnormal the pressure at the current moment, the more likely it is to exceed the safety threshold and trigger an abnormal alarm of the safety valve. The more urgent the warning, the more likely an abnormality alarm will be triggered.

[0084] This invention employs principal component analysis on different historical pressure change curves to distinguish different triggering causes, performs similarity clustering, and obtains representative change curves based on the clustering results. Compared to directly obtaining representative change curves based on a single triggering cause analysis, this invention introduces mutual factor analysis between multiple triggering causes, making the representative change curves more accurately and objectively represent the actual triggering cause changes. Then, by combining the pressure value distribution of the current pressure change curve and the similarity between each current principal component vector and each representative change curve, the abnormal cause at the current moment is determined. Combining the frequency of the abnormal cause's occurrence in history and the degree of pressure fluctuation abnormality, the urgency of the warning is determined. Furthermore, by combining the pressure fluctuation trend, an abnormal alarm is implemented. Thus, this invention can perform detailed analysis of multiple triggering causes, combining current pressure changes and historical change data to provide early warning of safety valve anomalies at the current moment, enabling timely detection and handling of anomalies, and improving the accuracy of safety valve alarms.

[0085] On the other hand, an abnormal alarm system for a high-temperature resistant safety valve is also provided. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the method as described in any of the foregoing.

[0086] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0087] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for abnormal alarm control of a high-temperature resistant safety valve, characterized in that, The method includes: Obtain historical pressure change curves for different safety valve trips within a preset time period, as well as a set of tripping reasons for each trip; Determine the historical principal component vectors and historical eigenvalues ​​of different historical pressure change curves; perform similarity clustering on all historical principal component vectors to obtain clusters; based on all historical principal component vectors and corresponding historical eigenvalues ​​within the same cluster, determine the representative change curve for each cause of jump; Obtain the current pressure change curve for a preset time period prior to the current moment. Determine the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling moments in the current pressure change curve. Determine the current principal component vector and current eigenvalue of the current pressure change curve. Determine the cause of the anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve and the current eigenvalue. Based on the frequency of occurrence of all abnormal causes in history and the degree of abnormal pressure fluctuations, the urgency level of the warning at the current moment is determined; based on the urgency level of the warning and the pressure fluctuation trend of the current pressure change curve, an abnormal alarm is issued.

2. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, The similarity clustering of all historical principal component vectors yields clusters, including: Based on the dynamic time warping algorithm, the DTW value of any two historical principal component vectors is calculated, and the negative of the DTW value is normalized as the curve similarity. Clustering is performed based on the curve similarity, and historical principal component vectors whose curve similarity is greater than a preset similarity threshold are regarded as the same cluster.

3. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, The step of determining the representative change curve for each cause of a jump based on all historical principal component vectors within the same cluster and their corresponding historical eigenvalues ​​includes: Within each cluster, the most frequent jump cause corresponding to the historical pressure change curve of the historical principal component vector is taken as the cluster cause of the cluster. The historical principal component vectors within each cluster are weighted according to their corresponding historical feature values ​​to obtain the representative change curve of the cluster cause. By traversing all clusters, the representative change curve of each jump cause is obtained.

4. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, The step of determining the degree of pressure fluctuation anomaly based on the pressure value distribution at different sampling times in the current pressure change curve includes: Based on the discrete distribution characteristics of pressure values ​​at different sampling times in the current pressure change curve, the pressure dispersion index is determined. The pressure difference between the next sampling time and the previous sampling time is calculated as a pressure increase index. The maximum value of the pressure increase index is normalized and used as a pressure change index. The product of the pressure dispersion index and the pressure change index is normalized and used as the degree of pressure fluctuation anomaly.

5. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, The step of determining the cause of the anomaly at the current moment based on the similarity between each current principal component vector and each representative change curve, and the current eigenvalue, includes: Calculate the Pearson correlation coefficient between each current principal component vector and each representative change curve, and normalize the Pearson correlation coefficient as the degree of correlation. The jump cause corresponding to the representative change curve with the highest correlation value of each current principal component vector is taken as the principal component cause; The principal component cause of the preset number of principal component vectors with the largest current eigenvalue is taken as the anomaly cause.

6. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, The determination of the current alert urgency level based on the frequency of occurrence of all abnormal causes in history and the degree of abnormal pressure fluctuations includes: The frequency of all abnormal causes in history is normalized and used as a frequency index. The product of the frequency index and the degree of pressure fluctuation anomaly is calculated and normalized to determine the urgency level of the warning at the current moment.

7. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 1, characterized in that, Based on the aforementioned urgency level of the warning and the pressure fluctuation trend of the current pressure change curve, an abnormal alarm will be triggered, including: Determine the urgency of the trend based on the pressure trend of the current pressure change curve; Calculate the product of the warning urgency level and the trend urgency level, and normalize it to obtain the anomaly coefficient; An anomaly alarm is triggered when the anomaly coefficient exceeds a preset anomaly threshold.

8. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 7, characterized in that, Based on the pressure trend of the current pressure change curve, determine the urgency of the trend, including: The pressure change curve is fitted with a straight line to obtain a fitted straight line. The slope of the fitted straight line is normalized and used as the degree of urgency of the trend.

9. The abnormal alarm control method for a high-temperature resistant safety valve as described in claim 7, characterized in that, The preset abnormal threshold is 0.

7.

10. An abnormal alarm system for a high-temperature resistant safety valve, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 9.

Images