Dangerous gas leakage detection method, detection system and equipment

Abstract

The invention relates to the technical field of gas detection, in particular to a hazardous gas leakage detection method, system and equipment, and the method comprises the steps: collecting gas concentration detection data and parameters in a working space; according to the parameter change condition of each detection point, calculating a diffusion characteristic coefficient for measuring the position condition between the detection point and the leakage point in the whole space and the corresponding diffusion change condition; establishing a function relationship based on the time sequence change characteristics of each detection point parameter, and calculating a drift disturbance coefficient in combination with the change condition between the detection data and the function data to measure the change condition of drift disturbance suffered by each detection point sensor; and finally, dynamically adjusting the threshold value of each detection point by combining the two indexes, and carrying out hazardous gas leakage detection according to the adjusted threshold value. The invention aims to improve the sensitivity and precision of gas leakage detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gas detection, in particular to a dangerous gas leakage detection method, a detection system and equipment. BACKGROUND

[0002] With the continuous progress of sensor technology, Internet of Things and intelligent algorithms, various gas detection systems have been widely applied to industrial sites. By monitoring the concentration of flammable gas and toxic gas in the environment in real time, the early warning and prevention of leakage risk are realized. However, in the face of complex and variable industrial environments and increasingly improved safety standards, traditional detection methods gradually reveal limitations in sensitivity, accuracy and system robustness.

[0003] The existing dangerous gas leakage detection technology generally relies on a fixed threshold alarm mechanism, that is, an alarm is triggered when the gas concentration detected exceeds the preset threshold. However, this method has significant defects: first, the gas has dynamic diffusion characteristics in space, and the concentration is low and unevenly distributed in the early stage of leakage, which often makes it difficult to reach the alarm threshold, resulting in poor system sensitivity and missing the best disposal opportunity; even if the threshold is lowered to improve sensitivity, it can only be effective at detection points near the leakage point, and may increase false alarms due to environmental fluctuations and other factors. Second, gas concentration detection sensors (such as gas detection sensors using TDLAS technology) inevitably have problems such as aging and drift in long-term operation, resulting in a decrease in measurement accuracy. The existing system uses periodic manual calibration to compensate for errors, which not only affects the continuous operation of the system, but also makes it difficult to accurately determine the calibration period, and cannot achieve real-time dynamic correction, further exacerbating the uncertainty of the detection results. SUMMARY

[0004] In view of the above, it is necessary to provide a dangerous gas leakage detection method, a detection system and equipment to solve the above problems.

[0005] According to an aspect of the present application, a dangerous gas leakage detection method is provided, the method comprising: presetting detection points in the working space, collecting various parameter data of each detection point at each time, including gas concentration data, wind direction data; determining a suspected leakage point based on the correlation measure of the gas concentration data between different detection points at each time and historical time; analyzing the difference characteristics of the local gas concentration change between each detection point and the suspected leakage point in the working space, and combining the overall distribution of the correlation measure of each detection point and the remaining detection points to determine the diffusion characteristic coefficient of each detection point at each time; classify the diffusion characteristic coefficients of each detection point at each time and historical time, establish a multivariate function for each category based on the data mapping relationship between the gas concentration at the corresponding time of each category and the remaining parameter data except the wind direction data, determine the drift disturbance coefficient of each detection point at each time by analyzing the change trend characteristics and distribution characteristics of the difference between the multivariate function and the gas concentration; obtain a fixed threshold of historical gas leakage detection, adjust the fixed threshold based on the diffusion characteristic coefficient and the drift disturbance coefficient of each detection point at each time, and obtain the adjusted threshold of each detection point at each time; determine the gas leakage factor at each time based on the numerical relationship between the gas concentration and the adjusted threshold of each detection point at each time, and determine whether the gas leaks.

[0006] The determination of the suspected leakage point is specifically: The gas concentration data of each detection point at each time and historical time is formed into a gas concentration sequence of each detection point, the average of the correlation measure between each detection point and all other detection points is calculated, and the detection point with the largest average is marked as the suspected leakage point.

[0007] The acquisition method of the diffusion characteristic coefficient of each detection point at each time is: The collected parameter data also includes wind direction data; For each time, the wind direction path is determined through the position of the suspected leakage point and the wind direction data, the gas concentration data of the nearest preset number of detection points to the wind direction path at each time is linearly fitted, and the absolute value of the slope of the fitted straight line is taken as the change measure at each time; the average of the correlation measure between all two-by-two combinations of detection points at each time is calculated; based on the average and the change measure, the global diffusion score at each time is determined; analyze the gas concentration difference and distance distribution between each detection point and its adjacent detection points along the direction of the wind direction path at each time to determine the local change coefficient of each detection point at each time; determine the local diffusion score of each monitoring point based on the correlation measure between each detection point and the suspected leakage point at each time and the difference in local change coefficient; The ratio between the global diffusion score at each time and the local diffusion score of each detection point at each time is taken as the diffusion characteristic coefficient of each detection point at each time.

[0008] The determination of the local change coefficient of each detection point at each time is specifically: For each detection point, the nearest preset number of detection points to it along the direction of the wind direction path are obtained, denoted as adjacent detection points; the distance between each detection point and each adjacent detection point is calculated; The difference between the gas concentration of each detection point and each adjacent detection point at each time is calculated, and the distance is divided to obtain a distribution change factor of each adjacent detection point. The distribution change factors of all adjacent detection points of each detection point are positively fused to obtain a local change coefficient of each detection point at each time.

[0009] The local diffusion score of each monitoring point is determined, and specifically: For each time, the difference between the local change coefficient between each detection point and the suspected leakage point is calculated, denoted as a change difference. The local diffusion score of each detection point is determined by positively fusing the negative correlation mapping result of the absolute value of the correlation measure between each detection point and the suspected leakage point and the change difference.

[0010] The drift disturbance coefficient of each detection point at each time is determined, and specifically: The residual between the gas concentration of each detection point at each time and the corresponding multivariate function value is obtained, and the sequence composed of the residuals obtained at all times is taken as a residual sequence. The specific formula of the drift disturbance coefficient is: : is the drift disturbance coefficient of the i-th detection point at time t, is the slope of the fitting straight line of the residual sequence of the i-th detection point at time t, is the median of the slopes of the fitting straight lines of the residual sequences of all detection points at time t, is the absolute deviation of the median of the slopes of the fitting straight lines of the residual sequences of all detection points at time t, is the absolute deviation of the median of the slopes of the fitting straight lines of the residual sequences of all detection points at time t, is a sign function.

[0011] The adjusted threshold value of each detection point at each time is obtained, and specifically: The numerical value proportion of the diffusion feature coefficient of each detection point at each time among all diffusion feature coefficients is calculated. The difference between the natural number 1 and the numerical value proportion is calculated, and the normalized value of the drift disturbance coefficient of each detection point at each time is positively fused. The sum of the natural number 1 and the normalized value is calculated, and the minimum value of the positively fused result and the sum is taken as the threshold adjustment parameter of each detection point at each time. The product of the threshold adjustment parameter of each detection point at each time and the fixed threshold value is taken as the adjusted threshold value of each detection point at each time.

[0012] The gas leakage factor at each time is determined to determine whether the gas is leaking, and specifically: For each time point, when the gas concentration of each detection point is greater than or equal to the adjusted threshold value, the threshold value coefficient is recorded as 1; otherwise, the threshold value coefficient is recorded as 0; The proportion of the absolute value of the difference between the adjusted threshold value of each detection point and the fixed threshold value in all detection points is calculated as the weight of the threshold value coefficient; the threshold value coefficients of all detection points at each time point are summed to obtain the gas leakage factor at each time point; When the gas leakage factor is greater than the preset leakage threshold value, it is determined that the gas leaks; otherwise, it is determined that the gas does not leak.

[0013] According to another aspect of the present application, a dangerous gas leakage detection device is provided, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the method according to any one of the above aspects when executing the computer program.

[0014] According to still another aspect of the present application, a dangerous gas leakage detection system is provided, which stores a computer program, and the computer program is executed by a processor to implement the method according to any one of the above aspects.

[0015] The present application has at least the following beneficial effects: The present application firstly presets detection points in the working space and collects gas concentration data and parameter data at each time to ensure comprehensive data coverage, which can effectively monitor the change of gas concentration, provide original data basis for subsequent analysis, and help accurately detect gas leakage. Further, by measuring the correlation of gas concentration data at each time and at different detection points at historical time, a suspected leakage point is determined, and an area with abnormal gas concentration change in the space is found, thereby narrowing the search range of the leakage point, improving the detection accuracy, and avoiding invalid detection of the entire space. According to the gas concentration difference characteristics between the suspected leakage point and each detection point, the diffusion characteristic coefficient is calculated, which provides characteristic information of gas diffusion, so that the diffusion path of gas leakage can be accurately analyzed under different conditions, providing a theoretical basis for subsequent drift disturbance coefficient calculation and threshold adjustment, and enhancing the accuracy of the early warning system. Then, by classifying the diffusion characteristic coefficient of each detection point at each time, a multivariate function is established for analyzing the trend of gas concentration change, which helps to accurately predict the trend of gas concentration change at different detection points, especially the slight change when there is no obvious leakage, thereby improving the response speed and accuracy of the detection system. The trend of change of the multivariate function and the gas concentration difference is analyzed to determine the drift disturbance coefficient of each detection point, which quantifies the single error caused by sensor aging and drift, and helps to improve the accuracy of subsequent threshold adjustment, thereby reducing the false alarm and false alarm probability to a certain extent. According to the drift disturbance coefficient, the threshold of each detection point is adjusted to realize dynamic adjustment, avoid false alarms caused by environmental fluctuations, improve the robustness of the detection system, and ensure high-precision detection in complex environments. By the relationship between the gas concentration of each detection point and the adjusted threshold, it is determined whether the gas leaks or not. In this way, the gas diffusion movement in the space and the single error caused by aging and drift are quantified, and after dynamic threshold adjustment, the detection sensitivity and accuracy are improved, and the false alarm probability is further reduced by combining the multi-point detection results for comprehensive judgment of dangerous gas leakage, thereby realizing a dangerous gas leakage detection method with higher sensitivity and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A step flow chart of a dangerous gas leakage detection method provided by the present application; Figure 2 An acquisition schematic diagram of a diffusion characteristic coefficient provided by the present application. DETAILED DESCRIPTION

[0017] In the description of the present embodiments, the words "example" and "exemplary" are used to mean serving as an example, instance, or illustration. Any embodiment or design described herein as "example" or "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the words "example" and "exemplary" is intended to present concepts in a concrete manner. As used in this application, the term "if' can be construed to mean "when" or "if," depending on the context. That is, if a given event occurs, then a given result can occur. Alternatively, if a given event does not occur, then a given result can not occur.

[0018] 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 application belongs. The terminology used in the description of the application herein is for describing particular embodiments only and is not intended to be limiting of the application. Unless otherwise defined, all terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0019] In addition, it should be pointed out that the terms "first", "second" in the present application and the drawings are used to distinguish similar objects, and are not used to describe a specific order or sequence. The method disclosed in the present embodiments or the method shown in the flowchart includes one or more steps for implementing the method, and the execution order of the steps can be interchanged with each other without departing from the scope of the present application, and some steps can also be deleted.

[0020] Please refer to Figure 1 , which shows a step flowchart of a dangerous gas leakage detection method provided by an embodiment of the present application, and the method includes the following steps: Step one: presetting detection points in the working space, collecting various parameter data of each detection point at each time, including gas concentration data and wind direction data.

[0021] In order to improve the sensitivity of dangerous gas leakage detection, real-time detection is performed at different positions in the working space. Specifically, a plurality of detection points are uniformly arranged in the entire space , in the present embodiment , a DFB laser is used to emit laser, the laser passes through the area to be detected, a detector is used to receive the laser signal, and TDLAS (Tunable Diode Laser Absorption Spectroscopy) detection is performed at each detection point. In the present embodiment, the modulation frequency is 1 kHz, the tuning coefficient of the laser is 0.1 nm / mA, the temperature tuning coefficient is , and the scanning period is 100 ms. The laser scanning band needs to cover the central absorption wavelength of the corresponding dangerous gas. According to the obtained laser signal, the dangerous gas absorption characteristics are extracted by using the Beer-Lambert law and the direct absorption method, that is, the gas concentration data of each position is obtained. TDLAS, the Beer-Lambert law and the direct absorption method are well-known technologies, and will not be described in detail. In order to ensure the correspondence of the data, the gas concentration obtained by each scanning within 1 second is averaged as the gas concentration data of the 1 second, that is, the sampling period of the gas concentration data is 1 Hz.

[0022] Then the temperature sensor, humidity sensor, air pressure sensor, wind speed and direction instrument synchronously collect temperature data, humidity data, air pressure data, and wind direction data in the space. The collection frequency of the above types of data is 1 Hz. The collected data, except for the wind direction data, are normalized respectively. The normalization method can include, but is not limited to, maximum normalization and maximum-minimum normalization. In this embodiment, the maximum-minimum normalization method is adopted.

[0023] Step two: based on the correlation measure of the gas concentration data between different detection points at each time and historical time, determine the suspected leakage point; analyze the difference characteristics of the local gas concentration change between each detection point and the suspected leakage point in the working space, and combine the overall distribution of the correlation measure of each detection point and the remaining detection points to determine the diffusion characteristic coefficient of each detection point at each time.

[0024] Since there is always a certain blind area in the layout of sensors in the entire space, and the diffusion movement of the gas makes the change of the gas concentration in the space often smooth and increasing, it is difficult to discover the gas leakage in time. Therefore, in order to improve the quality of gas leakage detection, it is necessary to first determine the diffusion change of the gas in the space.

[0025] Under normal circumstances, in the absence of gas leakage, the total amount of gas in the entire space is relatively stable or gradually decreases. At this time, the change of the gas concentration detected at different positions is mainly due to the movement of the gas under the influence of environmental parameters such as temperature and humidity, so there is no significant correlation between the gas concentrations of the detection points in the space. Only under the influence of the ventilation system, the concentration of each detection point gradually increases along the wind direction, but since the total amount of gas is relatively small, the difference between the gas concentrations of each detection point is still relatively small.

[0026] When gas leakage occurs and cannot be completely discharged by the ventilation system, due to the continuous input of gas at the leakage point, the total amount of gas in the space is constantly changing, and the concentration change of each point also has a certain correlation with the leakage point. Specifically, due to the diffusion movement of the gas in the space, the farther the distance between the detection point and the leakage point, the larger the corresponding space change range, and thus the lower the correlation between the gas concentration change at the corresponding detection point and the leakage point. Secondly, under the influence of the ventilation system, the gas concentration of each detection point shows a gradually increasing trend along the wind direction, and the difference between the gas concentrations of each point is large. Since the distribution of the gas concentration is nonlinear, the farther the distance from the leakage point, the greater the difference in the distribution change of the gas concentration.

[0027] For example, at time t, the gas concentration of each detection point before t is denoted as C(t-1), and the gas concentration of each detection point at t is denoted as C(t). The correlation measure between the gas concentration of each detection point and the gas concentration of the remaining detection points at t is denoted as R(t). ​The gas concentration data at the time points form a gas concentration sequence of each detection point in the collection order, and then the correlation measure between the gas concentration sequences of any two detection points in all detection points is calculated respectively, the average of the correlation measure between each detection point and all other detection points is calculated, and the detection point with the maximum average is marked as a suspected leakage point. In different processing methods, the correlation measure can be calculated by the Pearson correlation coefficient or the Spearman correlation coefficient, and the Pearson correlation coefficient is selected in this embodiment.

[0028] According to The wind direction path is determined according to the wind direction at the time points and the position of the suspected leakage point. Specifically, the straight line equation can be determined by the wind direction angle and the suspected leakage point position using the point-slope method, and the straight line can be used as the wind direction path. The detection point closest to the wind direction path is taken as the wind direction detection point. In this embodiment The number of detection points is taken as It should be noted that The number of detection points can be set by itself. When the number of detection points is small, the proportion can be appropriately increased and the rounding operation can be performed to ensure that the final obtained is an integer. The point-slope straight line equation is a known technology and will not be described in detail. Finally, the gas concentration data of the wind direction detection point is formed into a wind direction concentration sequence along the direction of the wind direction path.

[0029] Based on the above analysis, the diffusion characteristic coefficient is calculated to measure the position relationship between the detection points and the leakage point and the corresponding diffusion change in the whole space.

[0030] First, for each time point, the absolute value of the slope after linear fitting of the wind direction concentration sub-sequence is obtained, which is denoted as the change measure of each time point. The average of the correlation measure between all two-by-two combinations of detection points at each time point is calculated. Based on the average and the change measure, the global diffusion score of each time point is determined. In this embodiment, the average is denoted as Q, the change measure is denoted as X, and the global diffusion score is calculated by the formula ; wherein e represents the natural constant, and arctan() represents the inverse tangent function.

[0031] For each detection point, the preset number of detection points closest to it in the direction of the wind direction path are obtained, denoted as adjacent detection points; the distance between each detection point and each adjacent detection point thereof is calculated; the difference in gas concentration between each detection point and each adjacent detection point thereof at each time is calculated, and then divided by the distance, to obtain a distribution change factor of each adjacent detection point; and the distribution change factors of all adjacent detection points of each detection point are positively fused to obtain a local change coefficient of each detection point at each time. In this embodiment, the preset number is 2, and the implementer can adjust it according to the actual situation, which is not limited in the present application; the distance between detection points is calculated by the Euclidean distance; the difference in gas concentration is calculated by the absolute value of the difference between gas concentrations; and the positive fusion of multiple variables is specifically as follows: the distribution change factors of two adjacent detection points are respectively denoted as 、 , and the local change coefficient is calculated by the formula .

[0032] Further, for each time, the difference between the local change coefficients between each detection point and the suspected leakage point is calculated, denoted as a change difference; and the local diffusion score of each detection point is determined by positively fusing the change difference and the negative correlation mapping result of the absolute value of the correlation measure between each detection point and the suspected leakage point. In this embodiment, the difference between data is calculated by the absolute value of the difference; the change difference is denoted as , the correlation measure is denoted as P, and the local diffusion score is obtained by the formula ; wherein e represents a natural constant.

[0033] The ratio between the global diffusion score at each time and the local diffusion score of each detection point at each time is taken as the diffusion feature coefficient of each detection point at each time. It should be noted that if the denominator is 0, a preset value needs to be added to the denominator, and in this embodiment, the preset value is 1. The acquisition diagram of the diffusion feature coefficient is shown in Figure 2 .

[0034] It should be understood that, normally, the gas concentration between each detection point in the space has no correlation, and the change trend and difference of the gas concentration along the wind direction are small, so that the global diffusion score is small. Since the dimension of the global diffusion score is much smaller than that of the local diffusion score, the global diffusion distribution score plays a leading role at this time, and the corresponding diffusion characteristic coefficient is small, indicating that there is almost no gas leakage. Conversely, when there is gas leakage, the global diffusion score is larger, which indicates that the change of the gas concentration of each detection point is more consistent with the gas leakage. Therefore, the position of the leakage point can be further determined according to the corresponding data distribution characteristics. When the distance from the corresponding detection point to the leakage point is closer, the similarity between the local distribution change of the gas concentration around the corresponding detection point and the leakage point is higher, and the correlation between the change of the gas concentration and the leakage point is larger. At this time, the diffusion characteristic coefficient of the corresponding detection point is larger.

[0035] Step three: classify the diffusion characteristic coefficients of each detection point at each time and historical time, and establish a multivariate function for each category based on the data mapping relationship between the gas concentration at the corresponding time and the remaining parameter data except the wind direction data; analyze the change trend characteristics and distribution characteristics of the difference between the multivariate function and the gas concentration, and determine the drift disturbance coefficient of each detection point at each time.

[0036] The diffusion characteristic coefficient measures the distribution of the gas concentration between different detection points in the space, that is, the spatial characteristics of the change of the gas concentration. However, relying only on this spatial characteristic cannot guarantee the accuracy of long-term real-time gas leakage detection. Since the sensors used for gas concentration detection have problems such as aging and drift, and although all sensors have such changes, the change speed between different sensors also has certain differences, which will cause the detection error to become larger and larger. Therefore, the time characteristics need to be further considered to appropriately eliminate the differences between the sensors and determine the drift disturbance of each detection point.

[0037] First, when the sensor has no aging, drift problem or has little effect, the change of the gas concentration in the space has a high correlation with the fluctuation of the environmental parameters, that is, under the condition that the gas diffusion motion characteristics are similar, the change of the gas concentration has high predictability. Second, for sensors at different detection points, the influence of environmental parameters is universal, that is, when the environment fluctuates, it will cause the change of the gas diffusion motion in the space, thereby affecting all sensors at all detection points. However, the error caused by the aging and drift of the sensor is a single error of the sensor itself and does not affect all detection points. Therefore, the data fluctuation caused by environmental changes and the data fluctuation caused by aging and drift can be decoupled according to these characteristics.

[0038] At time t, the first detection point is the second detection point is Taking the first testing point as an example, the first... Before each testing point Using the diffusion characteristic coefficient at time step *i* as input, K-means clustering is employed to output the clustering results. K-means clustering is a well-known technique and will not be elaborated upon further. Within each cluster, the gas concentration data, temperature, humidity, and air pressure data corresponding to each element at each time step are used as input. Gas concentration data is used as the dependent variable, and temperature, humidity, and air pressure data are used as independent variables. A multivariate function is fitted to output the multivariate function for each cluster. Then, the multivariate function is used to calculate the *i*th time step *j*. Each testing point is located at Time and before The residuals at each time point, i.e., the differences between the actual values ​​and the values ​​of the multivariate function, are arranged in chronological order to form a residual sequence. A linear fit is then performed on the residual sequence to output the fitting equation. In this embodiment... We take 3600, which is one hour. For periods less than one hour, we fit the data based on the existing time length. Multivariate function fitting and linear fitting are well-known techniques and will not be elaborated further. We then process each detection point according to the above method.

[0039] Therefore, it is necessary to combine the temporal variation characteristics of the residual sequence elements at each detection point at each time step to calculate the drift perturbation coefficient, which is used to measure the drift perturbation changes experienced by the sensor at each detection point. The specific formula is as follows: : yes Time of the first The drift perturbation coefficient at each detection point yes Time of the first The slope of the fitted line of the residual sequence at each detection point. yes The median of the slope of the fitted straight line of the residual sequence of all detection points at time t. yes The median absolute deviation of the slope of the fitted straight line of the residual sequence at all detection points at time t is given by the time interval t. It is a symbolic function. In this embodiment, , yes Time of the first The slope of the fitted straight line of the residual sequence at each detection point. It is a median function.

[0040] It is understandable that environmental fluctuations have a global impact on sensor detection, meaning that the fluctuations in data collected by all sensors are relatively similar. Therefore, the median can be compared with the fluctuations of each sensor individually. Using the median can avoid the influence of individual sensor anomalies on the overall sensor performance. Reflects the trend characteristics of the overall sensor data collection changes, and the single sensor Reflects the trend characteristics of the overall sensor data collection changes, and the single sensor

[0041] Step four: obtain the fixed threshold value of historical gas leakage detection, adjust the fixed threshold value based on the diffusion characteristic coefficient and the drift disturbance coefficient of each detection point at each time, and obtain the adjusted threshold value of each detection point at each time.

[0042] The prior art often uses a fixed threshold value when detecting gas leakage, but according to the above analysis, the sensitivity of the fixed threshold value method is poor, and the gas leakage is often detected throughout the space, that is, the fixed threshold value is often large with respect to early detection of leakage (in the absence of sensor errors), and it is also difficult to locate the leakage point position in the early stage. If the threshold value is directly reduced as a whole, the false positive probability will increase due to the influence of factors such as environment and aging on the sensor. Since the above steps have quantified the influence of gas diffusion distribution in space and aging and drift, it is necessary to further adjust the threshold value distribution at different positions according to these quantification results, so as to improve the sensitivity while avoiding a significant increase in false positive probability.

[0043] The diffusion characteristic coefficient of each detection point is calculated at each time, and the numerical value of all diffusion characteristic coefficients is calculated. The difference between the natural number 1 and the numerical value is calculated, and the normalized value of the drift disturbance coefficient of each detection point at each time is positively fused. The sum of the natural number 1 and the normalized value is calculated, and the minimum value of the obtained positive fusion result and the sum value is taken as the threshold adjustment parameter of each detection point at each time. In this embodiment, the calculation method of multiplication is used for positive fusion of multiple variables; the maximum normalization method is used for normalization. It should be noted that the maximum and minimum normalization cannot be used, because the maximum and minimum normalization will make the drift disturbance coefficient constant positive, so that the drift direction of the sensor cannot be determined.

[0044] Further, the product of the threshold adjustment parameter of each detection point at each time and the fixed threshold value is calculated as the adjusted threshold value of each detection point at each time. The fixed threshold value can be the fixed threshold value in the historical gas leakage detection scheme, and will not be described in detail.

[0045] It can be understood that the greater the gas leakage probability and the closer the distance to the leakage point, the greater the corresponding diffusion characteristic coefficient, and the relative change of the gas concentration at these detection points in the early stage is relatively obvious. If the abnormal gas concentration can be detected at these positions, the gas leakage can be determined earlier. Secondly, when the sensor occurs positive drift, the greater the drift, the more it should be avoided to occur false positives. When negative drift occurs, the greater the drift, the more it should be avoided to occur false negatives.

[0046] Step five: based on the numerical relationship between the gas concentration of each detection point at each time and the adjusted threshold value, determine the gas leakage factor at each time, and judge whether the gas leaks.

[0047] When detecting the dangerous gas leakage in the operation process, first, the dynamic adjustment of the threshold value is performed, then the adjusted threshold value is compared with the gas concentration, and finally the gas leakage detection judgment is performed according to the weighted fusion result of all detection points. The specific weighted fusion method is: is the gas leakage factor at each time, is the change value between the threshold value of the i-th detection point and the fixed threshold value, is the sum of the absolute values of the change values of the threshold values of all detection points and the fixed threshold value, is the threshold coefficient of the i-th detection point, when the gas concentration is greater than or equal to the adjusted threshold value, takes 1, otherwise takes 0; represents the number of detection points. According to the above method, the gas leakage factor is calculated,

[0048] when the gas leakage factor is greater than the leakage threshold value , there is a dangerous gas leakage; otherwise, there is no dangerous gas leakage. It should be noted that the value interval is , in the embodiment the value is 0.7, the smaller the value , the higher the sensitivity, but the false positive probability is also greater; the implementer can adjust the value of according to the actual situation. Based on the same concept as the method embodiment of the present application, a dangerous gas leakage detection device is provided, which 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, the steps of the method described in any one of the above embodiments are implemented.

[0049]

[0050] ​​Based on the same concept as the method embodiments of the present application, a dangerous gas leakage detection system is provided. The system stores a computer program. When the computer program is executed by a processor, the method described above is implemented.

[0051] It should be noted that the flowcharts and block diagrams in the accompanying drawings show the architectural, functional and operational logic of possible implementations of the system, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a portion of code that contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions noted in the blocks can occur in different orders than those noted in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks can also occur in different orders than those disclosed in the descriptions, 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 reverse order, depending on the functions involved. Each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0052] The above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for detecting hazardous gas leaks, characterized in that, The method includes the following steps: Pre-set detection points in the work area and collect various parameter data at each detection point at each time, including gas concentration data and wind direction data; Based on the correlation measurement of gas concentration data between different detection points at various times and historical times, suspected leak points are identified; the differences in local gas concentration changes between each detection point and the suspected leak points in the work space are analyzed, and the overall distribution of the correlation measurement between each detection point and the other detection points is combined to determine the diffusion characteristic coefficient of each detection point at each time. The diffusion characteristic coefficients of each detection point at each time and historical time are classified. Based on the data mapping relationship between the gas concentration at the corresponding time of each category and the other parameter data except for wind direction data, a multivariate function is established for each category. The changing trend and distribution characteristics of the difference between the multivariate function and the gas concentration are analyzed to determine the drift disturbance coefficient of each detection point at each time. A fixed threshold for historical gas leak detection is obtained. Based on the diffusion characteristic coefficient and drift disturbance coefficient at each detection point at each time, the fixed threshold is adjusted to obtain the adjusted threshold at each detection point at each time. Based on the numerical relationship between the gas concentration at each detection point and the adjusted threshold at each time, the gas leakage factor at each time is determined to determine whether there is a gas leak.

2. The method for detecting hazardous gas leaks as described in claim 1, characterized in that, The identification of suspected leak points specifically involves: The gas concentration data at each time point and historical time point of each detection point are used to construct the gas concentration sequence of each detection point. The mean value of the correlation measure between each detection point and all other detection points is calculated, and the detection point with the largest mean value is marked as a suspected leak point.

3. The method for detecting hazardous gas leaks as described in claim 1, characterized in that, The method for determining the diffusion characteristic coefficient of each detection point at each time step is as follows: The collected parameter data also includes wind direction data; For each moment, the wind path is determined by the location of the suspected leak point and wind direction data; a linear fit is performed on the gas concentration data of a predetermined number of detection points closest to the wind path at each moment, and the absolute value of the slope of the fitted line is used as the change measure at each moment; the average value of the correlation measure between all pairs of detection points at each moment is calculated; based on the average value and the change measure, the global diffusion score at each moment is determined. Analyze the gas concentration differences and distance distribution between each detection point and its neighboring detection points along the wind path at each time step, and determine the local variation coefficient of each detection point at each time step. Based on the correlation measurement and local variation coefficient difference between each detection point and the suspected leak point at each time, the local diffusion score of each monitoring point is determined. The ratio between the global diffusion score at each time step and the local diffusion score at each detection point at each time step is used as the diffusion characteristic coefficient of each detection point at each time step.

4. The method for detecting hazardous gas leaks as described in claim 3, characterized in that, The determination of the local variation coefficient of each detection point at each time point is specifically as follows: For each detection point, a preset number of detection points that are closest to it are obtained along the wind path direction and recorded as neighboring detection points; the distance between each detection point and each of its neighboring detection points is calculated. Calculate the gas concentration difference between each detection point and each of its neighboring detection points at each time step, divide it by the distance to obtain the distribution change factor of each neighboring detection point, and positively fuse the distribution change factors of all neighboring detection points of each detection point to obtain the local change coefficient of each detection point at each time step.

5. The method for detecting hazardous gas leaks as described in claim 3, characterized in that, The determination of the local diffusion score for each monitoring point is specifically as follows: For each time point, the difference between the local change coefficients between each detection point and the suspected leak point is calculated and denoted as the change difference; the negative correlation mapping result based on the absolute value of the correlation measure between each detection point and the suspected leak point is positively fused with the change difference to determine the local diffusion score of each detection point.

6. The method for detecting hazardous gas leaks as described in claim 1, characterized in that, The determination of the drift perturbation coefficient for each detection point at each time point is specifically as follows: Obtain the residual between the gas concentration and the corresponding multivariate function value at each detection point at each time step, and use the sequence of residuals obtained at all times as the residual sequence; The specific formula for the drift disturbance coefficient is as follows: : yes Time of the first The drift perturbation coefficient at each detection point yes Time of the first The slope of the fitted line of the residual sequence at each detection point. yes The median of the slope of the fitted straight line of the residual sequence of all detection points at time t. yes The median absolute deviation of the slope of the fitted straight line of the residual sequence at all detection points at time t is given by the time interval t. It is a symbolic function.

7. The method for detecting hazardous gas leaks as described in claim 1, characterized in that, The threshold adjusted at each detection point at each time step is obtained as follows: Calculate the percentage of the diffusion characteristic coefficient of each detection point at each time step, calculate the difference between the natural number 1 and the percentage of the value, and perform positive fusion with the normalized value of the drift perturbation coefficient of each detection point at each time step. Calculate the sum of the natural number 1 and the normalized value, and use the minimum value of the obtained positive fusion result and the sum as the threshold adjustment parameter for each detection point at each time. Calculate the product of the threshold adjustment parameter and the fixed threshold at each detection point at each time moment, and use it as the adjusted threshold at each detection point at each time moment.

8. The method for detecting hazardous gas leaks as described in claim 1, characterized in that, The determination of the gas leakage factor at each moment to determine whether a gas leak has occurred specifically involves: For each time point, when the gas concentration at each detection point is greater than or equal to the adjusted threshold, the threshold coefficient is recorded as 1; otherwise, the threshold coefficient is recorded as 0. The absolute value of the difference between the adjusted threshold and the fixed threshold at each detection point is calculated as the proportion of the values ​​at all detection points, and this proportion is used as the weight of the threshold coefficient. The threshold coefficients at all detection points at each time point are summed by weight to obtain the gas leakage factor at each time point. When the gas leakage factor is greater than a preset leakage threshold, it is determined that a gas leak has occurred; otherwise, it is determined that there is no gas leak.

9. A hazardous gas leak detection device, 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 1-8.

10. A hazardous gas leak detection system, wherein the system stores a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

Images