A method, device and system for detecting sealing performance of reactor

Abstract

The present application relates to the technical field of sealing performance detection, and specifically to a method, device and system for detecting the sealing performance of a reactor, the method comprising: real-time acquisition of pressure data of multiple reactors to be detected for sealing performance; based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range, determining the fluctuation characteristic value of the pressure data of each reactor within each preset time range, and determining the pressure deviation value of the pressure data of each reactor within each preset time range; multiple sampling of the total sample composed of all the pressure deviation values ​​of all reactors, and constructing isolated trees respectively, determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree; and then determining the total path length of each pressure deviation value in the total sample in the isolated forest; using the isolated forest algorithm to detect the sealing performance of each reactor. The present application aims to improve the detection accuracy of the sealing performance of the reactor.

Description

Technical Field

[0001] The present application relates to the technical field of sealing performance detection, and in particular to a method, device and system for detecting the sealing performance of a reactor. Background Art

[0002] Reactors are comprehensive chemical reaction vessels. Depending on the reaction conditions, toxic, highly corrosive, flammable, or explosive media may be produced in the reactors. In the chemical industry, in order to ensure production safety, there are high requirements for the sealing performance of reactors containing dangerous gases. Before production, the reactors are tested for sealing performance to identify reactors with sealing problems so that production personnel can take targeted measures in advance to prevent leakage failures and reduce losses caused by leakage failures.

[0003] The pressure detection method is a commonly used method for detecting the sealing performance of reactors. It directly measures the pressure changes caused by the leakage to determine whether the reactor has a leakage fault. However, the pressure detection method is greatly affected by the detection environment and it is difficult to capture the pressure changes when the reactor has a slight leakage, resulting in a decrease in the detection accuracy of the reactor's sealing performance. How to eliminate the disturbance of environmental factors to the detection and improve the accuracy of the reactor's sealing performance detection is a problem that needs to be solved urgently. Summary of the invention

[0004] In view of the above, it is necessary to provide a method, device and system for detecting the sealing performance of a reactor, which can improve the detection accuracy of the sealing performance of the reactor compared with the traditional method for detecting the sealing performance of the reactor:

[0005] In a first aspect, an embodiment of the present application provides a method for detecting the sealing performance of a reactor, the method comprising the following steps:

[0006] Real-time collection of pressure data of multiple reactors to be tested for sealing performance;

[0007] Based on the change trend and distribution dispersion of the pressure data of each reactor within each preset time range, determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range;

[0008] Based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range;

[0009] Sampling the total sample consisting of all the pressure deviation values ​​of all the reactors for multiple times, and constructing isolated trees respectively, and determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the occurrence frequency of the pressure deviation values ​​corresponding to the time range;

[0010] Based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest;

[0011] Based on the total path length, the isolation forest algorithm is used to detect the sealing performance of each reactor.

[0012] In one embodiment, the process of determining the fluctuation characteristic value is:

[0013] Arrange the pressure data of each reactor within each preset time range in time sequence to form each pressure sequence of each reactor;

[0014] Obtain the fitting straight line and its slope of all pressure data in each pressure sequence of each reactor;

[0015] Based on the slope and the difference between all the pressure data in each pressure sequence and the corresponding fitting value on the fitting straight line, the fluctuation characteristic value of the pressure data of each reactor within each preset time range is determined.

[0016] In one embodiment, the expression of the fluctuation characteristic value is:

[0017] ; In the formula, represents the fluctuation characteristic value of the pressure data of the ith reactor within the oth preset time range; represents the slope of the fitted straight line of the jth pressure sequence of the i-th reactor; exp() represents an exponential function with a natural constant as the base; R represents the number of pressure data in the pressure sequence; , They respectively represent the rth pressure data and the fitting value of the rth pressure data in the jth pressure sequence of the ith reactor on the fitting straight line.

[0018] In one embodiment, the expression of the pressure deviation value is:

[0019] ; In the formula, Indicates the pressure deviation value of the pressure data of the ith reactor within the oth preset time range; , represent the fluctuation characteristic values ​​of the pressure data of the ith and nth reactors within the oth preset time range respectively; N represents the number of reactors to be tested for sealing performance; , They represent the j-th pressure sequence of the ith and n-th reactors respectively; D() represents the distance function; exp() represents the exponential function with a natural constant as the base.

[0020] In one embodiment, the expression of the discreteness is:

[0021] ; In the formula, Indicates the discreteness of the pressure deviation value distribution in the kth isolated tree; represents the discrete degree of the pressure deviation value in the kth isolated tree; L represents the number of pressure sequences of each reactor during the sealing performance test; j represents the sequence number of the pressure sequence corresponding to the pressure deviation value in the kth isolated tree; It represents the probability of occurrence of the pressure deviation value corresponding to the pressure sequence number j in the kth isolated tree; log represents the logarithmic function with base 2.

[0022] In one embodiment, the total path length is negatively correlated with the discreteness of each pressure deviation value in each isolated tree in the total sample, and is positively correlated with the path length of each pressure deviation value in each isolated tree.

[0023] In one embodiment, the total path length is expressed as: ; In the formula, It represents the total path length of the mth pressure deviation value in the total samples in the isolation forest; It indicates the number of times the mth pressure deviation value in the total samples is extracted to construct an isolated tree; norm() indicates the normalization function; represents the discreteness of the mth pressure deviation value in the total samples in the ath isolated tree; ε is a value preset to be greater than 0; It represents the path length of the mth pressure deviation value in the total samples in the ath isolated tree.

[0024] In one embodiment, the process of using the isolation forest algorithm to detect the sealing performance of each reactor is as follows:

[0025] The total path length of each pressure deviation value of each reactor to be tested for sealing performance in the isolation forest is used to replace the average path length of each pressure deviation value in the isolation forest algorithm, and the isolation forest algorithm is used to output the abnormal score of each pressure deviation value of each reactor to be tested for sealing performance;

[0026] Acquire multiple reactors with good sealing performance, output the abnormality scores according to the same acquisition method as the abnormality scores of each reactor whose sealing performance is to be tested, and use the maximum value as the sealing performance threshold;

[0027] If the abnormal score of any reactor to be tested for sealing performance is greater than the sealing performance threshold, it is determined that any reactor to be tested for sealing performance has a leakage problem; otherwise, it is determined that the sealing performance is good.

[0028] In a second aspect, the present application also provides a reactor sealing performance detection device, the device comprising:

[0029] A pressure acquisition module, used for real-time acquisition of pressure data of multiple reactors to be tested for sealing performance;

[0030] The pressure anomaly analysis module is used to determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range;

[0031] Based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range;

[0032] Sampling the total sample consisting of all the pressure deviation values ​​of all the reactors for multiple times, and constructing isolated trees respectively, and determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the occurrence frequency of the pressure deviation values ​​corresponding to the time range;

[0033] Based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest;

[0034] The sealing performance detection module is used to perform sealing performance detection on the reactor based on the total path length by using an isolation forest algorithm.

[0035] In a third aspect, an embodiment of the present application further provides a reactor sealing performance detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned reactor sealing performance detection methods when executing the computer program.

[0036] This application has at least the following beneficial effects:

[0037] The present application determines the fluctuation characteristic value of the pressure data of each reactor within each preset time range based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range; its beneficial effect is to utilize the correlation between the fluctuation characteristics of the pressure in the reactor and the probability of its existence of sealing performance problems, while obtaining the influence of temperature changes on the pressure in the reactor, enhance the feature extraction effect of the pressure data fluctuation caused by slight leakage, and improve the accuracy of subsequent testing of the sealing performance of the reactor;

[0038] Further, based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, the pressure deviation value of the pressure data of each reactor within each preset time range is determined; its beneficial effect is to utilize the characteristic that the detection environment of the chemical workshop has a similar impact on the pressure in different reactors, and by comparing the difference in pressure sequences between different reactors in the same detection period, obtain the pressure deviation value reflecting the degree of deviation of the pressure data of each reactor relative to the pressure data of all other reactors, eliminate the disturbance of temperature change on the detection, and improve the detection accuracy of the pressure change in the reactor caused by early slight leakage;

[0039] Furthermore, based on the degree of discreteness of the pressure deviation value in each isolated tree and the uniformity of the frequency of occurrence of the pressure deviation value in the corresponding time range, the discreteness of the distribution of the pressure deviation value in each isolated tree is determined; the total path length of each pressure deviation value in the total sample in the isolated forest is further determined; and the abnormal score of each pressure deviation value is calculated using the isolated forest algorithm; the sealing performance of the reactor is tested. Through the discreteness of the isolated tree, the weight of each isolated tree in calculating the total path length of the pressure deviation value is adjusted. For isolated trees with larger discreteness, the path length of the pressure deviation value in it is reduced, so as to improve the role of isolated trees with larger discreteness in identifying abnormal pressure deviation values, thereby increasing the abnormal score of abnormal pressure deviation values ​​output by the isolated forest algorithm, and improving the detection accuracy of the sealing performance of the reactor. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 A flowchart of a method for detecting the sealing performance of a reactor provided in one embodiment of the present application;

[0042] Figure 2 It is a schematic diagram of the process of determining the fluctuation characteristic value;

[0043] Figure 3 Schematic diagram of the process of obtaining the total path length. DETAILED DESCRIPTION

[0044] In the description of the embodiments of the present application, words such as "exemplary", "or", "for example" and the like are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "or", "for example" and the like is intended to present related concepts in a concrete manner.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the present application. The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. It should be understood that, unless otherwise specified, " / " means or.

[0046] It should also be noted that the terms "first" and "second" in the present application are used to distinguish similar objects rather than to describe a specific order or sequence.

[0047] The specific scheme of a reactor sealing performance detection method, device and system provided by the present application is described in detail below with reference to the accompanying drawings.

[0048] See also Figure 1 , which shows a flowchart of a method for detecting the sealing performance of a reactor provided by an embodiment of the present application, the method comprising the following steps:

[0049] Step 1: real-time acquisition of pressure data of multiple reactors whose sealing performance is to be tested.

[0050] Connect the stabilized gas source to the inlet valves of N reactors to be tested for sealing performance in the chemical workshop through a conduit, and tighten the exhaust valve of the reactor. Open the main valve and pressure-dividing valve of the stabilized gas source, first adjust the pressure of the pressure-dividing valve to the test pressure value Pt required for the test, and then open the reactor inlet valve to allow the gas to slowly fill the reactor. The gas medium of the stabilized gas source can be air or nitrogen, and oxygen or other gases are prohibited.

[0051] In this embodiment, the values ​​of N and Pt are 10 and 1.5 MPa respectively. The values ​​of N and Pt are preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0052] After the reactor air inlet valve is opened for a preset time, the pressure in the pressure-stabilizing gas source and the reactor has been relatively stable, and the reactor air inlet valve and the pressure-stabilizing gas source pressure-dividing valve are closed in sequence. At this time, the pressure data in each reactor is collected in real time through the digital pressure gauge installed on the reactor, and the pressure data of each reactor within each preset time range is arranged in time sequence to form each pressure sequence of each reactor, and all pressure sequences of each reactor are numbered in time sequence.

[0053] In this embodiment, the preset duration, the frequency of collecting pressure data by the digital pressure gauge, and the length of the preset time range are 5 minutes, 1 Hz, and 30 minutes, respectively. The values ​​of the preset duration, the collection frequency, and the length of the preset time range are all preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0054] Step 2: Based on the distribution of pressure data within each preset time range, determine the fluctuation characteristic value of the pressure data within each preset time range; and determine the pressure deviation value based on the difference in pressure data between different reactors within the same time range; sample the pressure deviation values ​​to construct an isolated tree, calculate the discreteness of the distribution of pressure deviation values ​​within the isolated tree, and the total path length of each pressure deviation value in the isolated forest.

[0055] Due to the wear and aging of the reactor seals during the production process, the sealing performance of the reactor continues to decline, and its sealing performance needs to be tested regularly. The pressure detection method directly measures the pressure changes caused by the leakage to determine whether the reactor has a leakage fault. It is a commonly used method for testing the sealing performance of the reactor. However, the pressure detection method is greatly affected by the detection environment and it is difficult to capture the pressure changes of the reactor when a slight leak occurs, resulting in a decrease in the accuracy of the detection of the reactor's sealing performance.

[0056] Step 2.1, based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range, determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range.

[0057] Considering the large volume of the reactor in the chemical production workshop, its internal pressure is greatly affected by temperature. When the temperature in the detection environment changes, the pressure data in the reactor will fluctuate, and because the change of the ambient temperature is continuous, the reactor is affected by the temperature and fluctuates slowly. In the early stage of the sealing performance problem of the reactor, the leakage is very weak and usually occurs intermittently. Its impact on the pressure in the reactor is relatively direct, causing the pressure data of the reactor to fluctuate for a short time. At the same time, the pressure changes caused by the leakage of the reactor will continue to accumulate, resulting in an overall decreasing trend of its pressure data.

[0058] In order to obtain the fluctuation information of the pressure in the reactor, the fitting straight line of all the pressure data in each pressure sequence is obtained, and the slope of the fitting straight line is calculated. The calculation of the slope is a well-known technology and will not be described in detail in this application.

[0059] In this embodiment, the least squares straight line fitting method is used to obtain the fitting straight line. As other implementation methods, based on the fitting straight line of the pressure data, the implementer can use other existing technologies to obtain the fitting straight line, such as linear regression analysis, weighted least squares method, etc., and this application does not make any special restrictions.

[0060] Based on the slope of the fitting straight line of each pressure sequence of each reactor and the difference between all pressure data in each pressure sequence and the corresponding fitting value on the fitting straight line, the fluctuation characteristic value of the pressure data of each reactor within the preset time range corresponding to each pressure sequence is determined, and the expression is:

[0061] ; In the formula, represents the fluctuation characteristic value of the pressure data of the ith reactor within the oth preset time range; represents the slope of the fitted straight line of the jth pressure sequence of the ith reactor; exp() represents an exponential function with a natural constant as the base, the purpose of which is to reflect the negative correlation between the slope and the fluctuation characteristic value; R represents the number of pressure data in the pressure sequence; , They respectively represent the rth pressure data and the fitting value of the rth pressure data in the jth pressure sequence of the ith reactor on the fitting straight line.

[0062] It should be noted that: on the one hand, under the influence of the same detection temperature change, the more obvious the overall downward trend of the pressure data in the pressure sequence of the reactor is, the greater the probability that the reactor has sealing performance problems. When the slope of the fitting line is negative, the smaller the slope of the pressure sequence fitting line, the greater the degree of fluctuation, and the larger the fluctuation characteristic value; when the slope of the fitting line is not negative, the greater the slope of the pressure sequence fitting line, indicating that the pressure in the reactor increases due to the influence of temperature, and the smaller the fluctuation characteristic value;

[0063] On the other hand, the greater the difference between the pressure data and the fitting value in the pressure sequence of the reactor, the greater the amplitude of the short-term fluctuation in the pressure data, the greater the possibility of a slight leak in the reactor, and the greater the fluctuation characteristic value. By using the correlation between the fluctuation characteristics of the pressure in the reactor and the probability of its sealing performance problem, while obtaining the impact of temperature changes on the pressure in the reactor, the feature extraction effect of the pressure data fluctuation caused by slight leaks is enhanced, thereby improving the accuracy of subsequent testing of the reactor's sealing performance. The flow chart for determining the fluctuation characteristic value is shown in the figure below. Figure 2 shown.

[0064] Step 2.3, based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range.

[0065] Compared with the pressure change in the reactor caused by temperature change, the pressure change caused by a slight leakage in the reactor is usually weaker. When the temperature change is large, it is difficult to distinguish the small fluctuations in pressure data caused by a slight leakage. Therefore, based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, the pressure deviation value of the pressure data of each reactor within each preset time range is determined, reflecting the degree of deviation of the pressure data of each reactor relative to the pressure data of all other reactors within each preset time range. The expression is:

[0066] ; In the formula, Indicates the pressure deviation value of the pressure data of the ith reactor within the oth preset time range; , represent the fluctuation characteristic values ​​of the pressure data of the ith and nth reactors within the oth preset time range respectively; N represents the number of reactors to be tested for sealing performance; , denote the jth pressure sequence of the ith and nth reactors respectively; D() denotes the distance function; exp() denotes an exponential function with a natural constant as the base, the purpose of which is to map the differences between all fluctuation eigenvalues ​​to positive values.

[0067] In this embodiment, the distance between pressure sequences is the DTW distance. As other implementation methods, on the basis of being able to measure the distance between pressure sequences, the implementer may use other existing technologies for measurement, such as Euclidean distance, Manhattan distance, etc., and this application does not impose any special restrictions.

[0068] It should be noted that: the larger the fluctuation characteristic value corresponding to the pressure sequence of the reactor, the greater the fluctuation degree of the pressure data in the pressure sequence, the greater the probability that the reactor has sealing performance problems, and the greater the pressure deviation value. Taking into account the same testing environment when the sealing performance of N reactors in the chemical workshop is tested at the same time, the impact of temperature changes on the pressure data of different reactors is similar. The greater the distance between the pressure sequence of the ith reactor and all other reactors, the greater the possibility that the ith reactor will cause pressure fluctuations due to slight leaks in addition to the influence of temperature, the greater the probability that it has sealing performance problems, and the greater the pressure deviation value. At the same time, since the pressure data changes caused by slight leaks in the reactor are relatively weak, the difference between the fluctuation characteristic values ​​of the remaining reactors and the ith reactor in the same time range is used as the weight. When When the value is larger, the possibility of sealing performance problems of the i-th reactor is smaller than that of the n-th reactor, so Assign smaller weight; when The smaller the value, the greater the possibility that the sealing performance of the i-th reactor will have problems compared with the n-th reactor. The greater the weight assigned, the more prominent the difference between the pressure in the ith reactor and the nth reactor at this time is, the difference between the pressure deviation value with slight fluctuations in the pressure data and the other pressure deviation values ​​is increased, and the obtained pressure deviation value is larger. Taking advantage of the fact that the detection environment of the chemical workshop has a similar impact on the pressure in different reactors, by comparing the difference in the pressure sequence between different reactors in the same detection period, the pressure deviation value reflecting the degree of deviation of the pressure data of each reactor relative to the pressure data of all other reactors is obtained, eliminating the disturbance of temperature change on the detection, and improving the detection accuracy of the pressure change in the reactor caused by early slight leakage.

[0069] Step 2.4, perform multiple sampling on the total sample consisting of all the pressure deviation values ​​of all reactors, and construct isolated trees respectively, and determine the discreteness of the distribution of pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the frequency of occurrence of the pressure deviation values ​​in the corresponding time range.

[0070] All the pressure deviation values ​​of all reactors in the chemical workshop are combined into a total sample, and the pressure deviation values ​​in the total sample are analyzed using the isolation forest algorithm. The number of isolated trees is set to 100, the number of subsamples extracted each time is 256, the maximum depth of the isolated tree is 8, and the path length of each pressure deviation value in each isolated tree is output.

[0071] It should be noted that: the number of isolated trees is 100, the number of subsamples extracted each time is 256, and the maximum depth of the isolated tree is 8. These are only an embodiment of the present application, and the implementer may set them according to actual conditions.

[0072] Since the leakage of the reactor is usually intermittent in the early stage of sealing performance problems, and the impact of the detection environment on the pressure data is also time-varying, the pressure deviation values ​​of different reactors within different preset time ranges vary greatly, so there are differences in the distribution of pressure deviation values ​​in the randomly selected sub-samples.

[0073] Based on the discrete degree of the pressure deviation value in each isolated tree and the uniformity of the frequency of occurrence of the pressure deviation value corresponding to the time range, the discrete degree of the distribution of the pressure deviation value in each isolated tree is determined, and the expression is:

[0074] ; In the formula, Indicates the discreteness of the pressure deviation value distribution in the kth isolated tree; represents the discrete degree of the pressure deviation value in the kth isolated tree; L represents the number of pressure sequences of each reactor during the sealing performance test; j represents the sequence number of the pressure sequence corresponding to the pressure deviation value in the kth isolated tree; It represents the probability of occurrence of the pressure deviation value corresponding to the pressure sequence number j in the kth isolated tree; log represents the logarithmic function with base 2.

[0075] In this embodiment, the degree of dispersion is the standard deviation. As other implementation methods, based on the degree of uneven distribution of measurable pressure deviation values, the implementer may use other existing technologies for measurement, such as variance, coefficient of variation, etc. This application does not impose any special restrictions.

[0076] It should be noted that the more dispersed the size distribution of the pressure deviation value in the isolated tree is, the greater its discreteness is, and the greater the discreteness is obtained. In addition, the pressure deviation values ​​corresponding to the pressure sequences with different serial numbers in the isolated tree, when the serial number distribution of the pressure deviation values ​​corresponding to the pressure sequence in the isolated tree is more uniform, the calculated entropy is greater, indicating that the discreteness of the distribution of the pressure deviation values ​​in the isolated tree is greater. At the same time, as the leakage continues to occur, the deviation degree of the pressure data of the reactor with sealing performance problems continues to accumulate. The larger the serial number of the corresponding pressure sequence, the higher the accuracy of the pressure deviation value in indicating the leakage of the reactor. Therefore, the serial number of the pressure sequence corresponding to the pressure deviation value is used as the weight to calculate the discreteness, so that the discreteness of the isolated tree containing the pressure deviation value corresponding to the larger pressure sequence number is greater.

[0077] Step 2.5, based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest.

[0078] The expression for calculating the total path length of each pressure deviation value in the total sample in the isolation forest is:

[0079] ; In the formula, It represents the total path length of the mth pressure deviation value in the total samples in the isolation forest; It indicates the number of times the mth pressure deviation value in the total samples is extracted to construct an isolated tree; norm() indicates the normalization function; represents the discreteness of the mth pressure deviation value in the total sample in the ath isolated tree; ε is a value preset to be greater than 0, in order to avoid the denominator being 0. The value of ε is preset manually and can be set by the implementer. In this embodiment, the value of ε is 0.1; It represents the path length of the mth pressure deviation value in the total samples in the ath isolated tree.

[0080] In this embodiment, the normalization function is the Min-Max normalization method. As another implementation method, Based on the normalization process, implementers can use other existing technologies to Normalization processing is performed, such as Z-Score normalization method, decimal calibration normalization method, etc., and this application does not make any special restrictions.

[0081] It should be noted that the pressure data changes caused by slight leakage in the reactor are difficult to capture. The discreteness of the isolated tree is used to adjust the weight of each isolated tree in calculating the total path length of the pressure deviation value. For isolated trees with larger discreteness, the path length of the pressure deviation value in it is reduced, so as to improve the role of isolated trees with larger discreteness in identifying abnormal pressure deviation values, thereby increasing the abnormal score of abnormal pressure deviation values ​​output by the isolation forest algorithm and improving the detection accuracy of the reactor sealing performance. The schematic diagram of the total path length acquisition process is shown in the figure. Figure 3 shown.

[0082] Step 3: Based on the total path length, an isolation forest algorithm is used to detect the sealing performance of the reactor.

[0083] The total path length of each pressure deviation value in the isolation forest replaces the average path length of each pressure deviation value in the isolation forest algorithm, and the isolation forest algorithm is used to output the abnormal score of each pressure deviation value, that is, the abnormal score obtained when the sealing performance of N reactors to be tested is tested. Among them, the isolation forest algorithm is a well-known technology and will not be described in detail in this application.

[0084] Obtain N reactors with good sealing performance, and perform a preset number of consecutive sealing performance tests on the N reactors with good sealing performance according to the same method for obtaining the abnormal scores of each reactor to be tested for sealing performance, and output the abnormal scores, and use the maximum value of the abnormal scores as the sealing performance threshold. When performing sealing performance tests on the reactor to be tested for sealing performance and the reactor with good sealing performance, perform a sealing performance test every preset time length.

[0085] In this embodiment, the values ​​of the preset number and the preset time length are 50 and 30 minutes respectively. The values ​​of the preset number and the preset time length are preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0086] In the process of testing the sealing performance of N reactors to be tested in a chemical workshop, if the abnormal score of the pressure deviation value of any reactor output by the isolation forest algorithm is greater than the sealing performance threshold, it is determined that any reactor has a leakage problem, and the buzzer sends an alarm signal to remind the workshop staff to take targeted maintenance measures for the reactor with leakage problem to prevent the reactor from leaking; otherwise, it is determined that the sealing performance of any reactor is good.

[0087] Based on the same inventive concept as the above method, the embodiment of the present application also provides a reactor sealing performance detection device, comprising:

[0088] A pressure acquisition module, used for real-time acquisition of pressure data of multiple reactors to be tested for sealing performance;

[0089] The pressure anomaly analysis module is used to determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range;

[0090] Based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range;

[0091] Sampling the total sample consisting of all the pressure deviation values ​​of all the reactors for multiple times, and constructing isolated trees respectively, and determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the occurrence frequency of the pressure deviation values ​​corresponding to the time range;

[0092] Based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest;

[0093] The sealing performance detection module is used to perform sealing performance detection on the reactor based on the total path length by using an isolation forest algorithm.

[0094] Based on the same inventive concept as the above method, an embodiment of the present application also provides a reactor sealing performance detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned reactor sealing performance detection methods are implemented.

[0095] In summary, the present application determines the fluctuation characteristic value of the pressure data of each reactor within each preset time range based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range; its beneficial effect is to utilize the correlation between the fluctuation characteristic of the pressure in the reactor and the probability of its existence of sealing performance problems, and at the same time as obtaining the influence of temperature change on the pressure in the reactor, enhance the feature extraction effect of the pressure data fluctuation caused by slight leakage, and improve the accuracy of subsequent testing of the sealing performance of the reactor;

[0096] Further, based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, the pressure deviation value of the pressure data of each reactor within each preset time range is determined; its beneficial effect is to utilize the characteristic that the detection environment of the chemical workshop has a similar impact on the pressure in different reactors, and by comparing the difference in pressure sequences between different reactors in the same detection period, obtain the pressure deviation value reflecting the degree of deviation of the pressure data of each reactor relative to the pressure data of all other reactors, eliminate the disturbance of temperature change on the detection, and improve the detection accuracy of the pressure change in the reactor caused by early slight leakage;

[0097] Furthermore, based on the degree of discreteness of the pressure deviation value in each isolated tree and the uniformity of the frequency of occurrence of the pressure deviation value in the corresponding time range, the discreteness of the distribution of the pressure deviation value in each isolated tree is determined; the total path length of each pressure deviation value in the total sample in the isolated forest is further determined; and the abnormal score of each pressure deviation value is calculated using the isolated forest algorithm; the sealing performance of the reactor is tested. Through the discreteness of the isolated tree, the weight of each isolated tree in calculating the total path length of the pressure deviation value is adjusted. For isolated trees with larger discreteness, the path length of the pressure deviation value in it is reduced, so as to improve the role of isolated trees with larger discreteness in identifying abnormal pressure deviation values, thereby increasing the abnormal score of abnormal pressure deviation values ​​output by the isolated forest algorithm, and improving the detection accuracy of the sealing performance of the reactor.

[0098] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

[0099] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the basic features of the present application. Therefore, no matter from which point of view, the above embodiments of the present application should be regarded as exemplary and non-restrictive.

Claims

1. A method for detecting the sealing performance of a reactor, characterized in that: The method comprises the following steps: Real-time collection of pressure data of multiple reactors to be tested for sealing performance; Based on the change trend and distribution dispersion of the pressure data of each reactor within each preset time range, determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range; Based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range; Sampling the total sample consisting of all the pressure deviation values ​​of all the reactors for multiple times, and constructing isolated trees respectively, and determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the occurrence frequency of the pressure deviation values ​​corresponding to the time range; Based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest; Based on the total path length, the isolation forest algorithm is used to detect the sealing performance of each reactor.

2. A method for detecting the sealing performance of a reactor as claimed in claim 1, characterized in that: The determination process of the fluctuation characteristic value is as follows: Arrange the pressure data of each reactor within each preset time range in time sequence to form each pressure sequence of each reactor; Obtain the fitting straight line and its slope of all pressure data in each pressure sequence of each reactor; Based on the slope and the difference between all the pressure data in each pressure sequence and the corresponding fitting value on the fitting straight line, the fluctuation characteristic value of the pressure data of each reactor within each preset time range is determined.

3. A method for detecting the sealing performance of a reactor as claimed in claim 2, characterized in that: The expression of the fluctuation characteristic value is: ; In the formula, represents the fluctuation characteristic value of the pressure data of the ith reactor within the oth preset time range; represents the slope of the fitted straight line of the jth pressure sequence of the i-th reactor; exp() represents an exponential function with a natural constant as the base; R represents the number of pressure data in the pressure sequence; , They respectively represent the rth pressure data and the fitting value of the rth pressure data in the jth pressure sequence of the ith reactor on the fitting straight line.

4. A method for detecting the sealing performance of a reactor as claimed in claim 2, characterized in that: The expression of the pressure deviation value is: ; In the formula, Indicates the pressure deviation value of the pressure data of the ith reactor within the oth preset time range; , Respectively represent the fluctuation characteristic values ​​of the pressure data of the ith and nth reactors within the oth preset time range; N represents the number of reactors to be tested for sealing performance; , They represent the j-th pressure sequence of the ith and n-th reactors respectively; D() represents the distance function; exp() represents the exponential function with a natural constant as the base.

5. A method for detecting the sealing performance of a reactor as claimed in claim 2, characterized in that: The expression of the discreteness is: ; In the formula, Indicates the discreteness of the pressure deviation value distribution in the kth isolated tree; represents the discrete degree of the pressure deviation value in the kth isolated tree; L represents the number of pressure sequences of each reactor during the sealing performance test; j represents the sequence number of the pressure sequence corresponding to the pressure deviation value in the kth isolated tree; It represents the probability of occurrence of the pressure deviation value corresponding to the pressure sequence number j in the kth isolated tree; log represents the logarithmic function with base 2.

6. A method for detecting the sealing performance of a reactor as claimed in claim 1, characterized in that: The total path length is negatively correlated with the discreteness of each pressure deviation value in each isolated tree in the total sample, and is positively correlated with the path length of each pressure deviation value in each isolated tree.

7. A method for detecting the sealing performance of a reactor as claimed in claim 6, characterized in that: The total path length is expressed as: ; In the formula, It represents the total path length of the mth pressure deviation value in the total samples in the isolation forest; It indicates the number of times the mth pressure deviation value in the total samples is extracted to construct an isolated tree; norm() indicates the normalization function; Indicates the discreteness of the mth pressure deviation value in the total samples within the ath isolated tree; ε is a value preset to be greater than 0; It represents the path length of the mth pressure deviation value in the total samples in the ath isolated tree.

8. A method for detecting the sealing performance of a reactor as claimed in claim 1, characterized in that: The process of using the isolation forest algorithm to detect the sealing performance of each reactor is as follows: The total path length of each pressure deviation value of each reactor to be tested for sealing performance in the isolation forest is used to replace the average path length of each pressure deviation value in the isolation forest algorithm, and the isolation forest algorithm is used to output the abnormal score of each pressure deviation value of each reactor to be tested for sealing performance; Acquire multiple reactors with good sealing performance, output the abnormality scores according to the same acquisition method as the abnormality scores of each reactor whose sealing performance is to be tested, and use the maximum value as the sealing performance threshold; If the abnormal score of any reactor to be tested for sealing performance is greater than the sealing performance threshold, it is determined that any reactor to be tested for sealing performance has a leakage problem; otherwise, it is determined that the sealing performance is good.

9. A reactor sealing performance detection device, characterized in that: The device comprises: A pressure acquisition module, used for real-time acquisition of pressure data of multiple reactors to be tested for sealing performance; The pressure anomaly analysis module is used to determine the fluctuation characteristic value of the pressure data of each reactor within each preset time range based on the change trend and distribution discreteness of the pressure data of each reactor within each preset time range; Based on the fluctuation characteristic value, combined with the difference in pressure data between each reactor and all other reactors within each preset time range, and the difference in the fluctuation characteristic value between them, determine the pressure deviation value of the pressure data of each reactor within each preset time range; Sampling the total sample consisting of all the pressure deviation values ​​of all the reactors for multiple times, and constructing isolated trees respectively, and determining the discreteness of the distribution of the pressure deviation values ​​in each isolated tree based on the discreteness of the pressure deviation values ​​in each isolated tree and the uniformity of the occurrence frequency of the pressure deviation values ​​corresponding to the time range; Based on the path length of each pressure deviation value in the total sample in all isolated trees and the dispersion, determine the total path length of each pressure deviation value in the total sample in the isolated forest; The sealing performance detection module is used to perform sealing performance detection on the reactor based on the total path length by using an isolation forest algorithm.

10. A reactor sealing performance detection 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, the steps of a method for detecting the sealing performance of a reactor as described in any one of claims 1 to 8 are implemented.

Images