Method, device, storage medium and equipment for determining concentration of combustible gas leakage cloud

Abstract

The invention discloses a method for determining the concentration of a gas cloud of a combustible gas leak, which is based on a linear beam combustible gas detector and includes obtaining the leakage rate and wind speed of a target area; simulating and calculating the concentration distribution in a calculation domain after the gas leaks; extracting the gas concentration on the corresponding beam path according to the linear coordinates of different beam paths of the detector, and obtaining the beam depression angle; calculating the integral concentration obtained by the detector light path passing through the cloud according to the light path coordinates and the gas concentration at the corresponding position; using the beam depression angle, wind speed and integral concentration as inputs and the probability distribution of the gas concentration peak value on the beam path in multiple concentration intervals as outputs, training a pre-established neural network pattern recognition model to obtain a trained model; inputting data into the trained model to obtain the probability distribution of the maximum gas concentration of the corresponding light path passing through the cloud in different gas concentration intervals, and determining the leaked gas cloud and its concentration. Also disclosed are a device, a storage medium and equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of combustible gas leakage monitoring, and in particular to a method, device, storage medium and equipment for determining the concentration of a combustible gas leakage cloud. Background Art

[0002] Combustible gas leakage monitoring can detect signs of accidents early and prevent fire and explosion accidents. It is the most effective means to ensure safe production. At present, combustible gas detectors are mainly divided into two categories: fixed point type and linear beam type. Fixed point detectors include catalytic combustion and infrared point detectors. The main working principle is that the leaked gas enters the gas chamber of the detector and the gas concentration is fed back through the sensing component. The commonly used unit is %LEL, and the alarm threshold is usually set to 25 and 50%LEL. Linear beam detectors use the principle of spectral absorption to detect combustible gases and vapors such as hydrocarbons, ethers, and alcohols. The measurement result is the mathematical integral value (LEL·m) of the concentration of combustible gas (LEL) along the optical path length (m). The unit is different from that of the fixed point type, resulting in a lack of basis for how to set the alarm threshold of the linear beam combustible gas detector.

[0003] Due to different working modes, the "integrated concentration" measured by the linear beam combustible gas detector is related to the gas cloud concentration distribution and the length of the light path. For example, the test result of 1LEL·m can be expressed as a gas cloud with a width of 4m and a concentration of 25%LEL, or a gas cloud with a width of 2m and a concentration of 50%LEL. Therefore, it is difficult to judge the concentration and risk of the leaking gas cloud based on the "integrated concentration", and it is impossible to set a reasonable alarm threshold. Summary of the invention

[0004] The technical problem to be solved by the present invention is that it is currently impossible to judge the concentration and risk of a leaking gas cloud based on the "integrated concentration" measured by a linear beam combustible gas detector. The purpose is to provide a method, device, storage medium and equipment for determining the concentration of a leaking gas cloud, so that the concentration of the gas cloud on the optical path can be directly judged by the integrated concentration measured by the linear beam combustible gas detector, thereby setting the alarm threshold and rapid response measures more scientifically and reasonably, which is of great significance to ensuring the safe production of hazardous chemicals.

[0005] The first object of the present invention is to provide a method for determining the concentration of a combustible gas leak cloud, the method being based on a linear beam combustible gas detector, the method comprising:

[0006] S1: Obtain historical leakage rate and wind speed data of the target area;

[0007] S2: Simulate and calculate the concentration distribution of gas leakage in the calculation domain under different leakage rates and wind speeds;

[0008] S3: According to the straight line coordinates of the beam paths emitted by the detector at different positions and at different angles in the calculation domain, the gas concentration on the corresponding beam path is extracted, and the beam depression angles of different optical paths are obtained;

[0009] S4: Based on the optical path coordinates and the gas concentration at the corresponding position, the integrated concentration obtained when the detector optical path passes through the cloud is calculated;

[0010] S5: Using the beam pitch angle, wind speed and integrated concentration as inputs and the probability distribution of the gas concentration peak on the beam path in multiple concentration intervals as outputs, a pre-established neural network pattern recognition model is trained;

[0011] S6: Input the data into the trained model to obtain the probability distribution of the maximum gas concentration of the corresponding light path passing through the cloud in different gas concentration ranges, so as to determine the leakage gas cloud and the concentration of the leakage gas cloud.

[0012] In an optional embodiment, in step S2, a computational fluid dynamics model or a Gaussian diffusion equation is used to perform simulation calculations to obtain the concentration distribution of the gas in the calculation domain after leakage.

[0013] In an optional embodiment, in step S3, the method for acquiring the depression angles of the light beams of the different light paths is:

[0014]

[0015] β is the beam depression angle under the beam path, L is the horizontal distance between the detector and the leakage point, H is the installation height of the detector, a is the Y coordinate of the monitoring point, and the light beam is emitted from the location of the detector and incident on the monitoring point.

[0016] In an optional embodiment, the integrated concentration is calculated using an approximate rectangular method.

[0017] In an optional embodiment, the integral concentration is calculated as follows:

[0018]

[0019] IC is the integrated concentration on any beam path; H, L and Z are the vertical height and horizontal spacing of the detector from the leakage point and the z-axis coordinate of the detector location, respectively; a and b are the coordinates of the monitoring point on the y and z axes, respectively; n is the number of grids on the x-axis of any section containing the leakage point and within the calculation domain; x i is the x-direction length of the grid i in the computational domain; c i is the gas concentration of grid i in the computational domain.

[0020] In an optional embodiment, in step S5, the gas concentration peak is the maximum value on the path of the light beam emitted by the detector passing through the leaking gas cloud;

[0021] The method for obtaining the probability distribution of the gas concentration peak in multiple concentration intervals is:

[0022] Under different beam paths and leakage diffusion conditions, t groups of data are formed. The beam depression angle, wind speed and integrated concentration form a t×3 input matrix, and the corresponding gas concentration peak on the beam path is a t×1 output matrix.

[0023] The gas concentration peak t×1 output matrix is ​​converted into a t×N 01 matrix, and the conversion method is:

[0024] Determine N concentration intervals [0~d1, d1~d2...d n-1 ~d n ], and determine the concentration interval d according to the gas concentration peak value on the beam path i-1 ~d i , and mark it as 1, and mark other concentration intervals as 0, forming [0,0..1...0]. In the same way, the gas concentration peak t×1 output matrix is ​​converted into a t×N 01 matrix;

[0025] The training input of the neural network pattern recognition model is a t×3 matrix consisting of beam pitch angle, wind speed and integrated concentration, and the training output is a t×N 01 matrix consisting of gas concentration peaks.

[0026] In an optional embodiment, in step S5, a two-layer feedforward neural network pattern recognition model is used. When establishing the model, a sigmoid transfer function is set in the hidden layer, the number of hidden layer neuron layers is Q, and a softmax transfer function is added to the output layer, and a probability value is assigned to each output classification result.

[0027] In an optional embodiment, in step S6, the integrated concentration of the cloud cluster passed by the detector light path, the instantaneous wind speed and the beam depression angle of the light beam path are input into the trained model.

[0028] A second object of the present invention is to provide a device for determining the concentration of a combustible gas leak cloud, comprising:

[0029] A data acquisition module, used to acquire leakage rate and wind speed data;

[0030] The concentration distribution calculation module is used to calculate the concentration distribution of the leaked gas in the calculation domain according to the leakage rate and wind speed;

[0031] A coordinate acquisition module, used to acquire the straight line coordinates of different beam paths emitted by the linear beam combustible gas detector within the calculation domain;

[0032] A gas concentration extraction module is used to extract the gas concentration on the corresponding beam path from the concentration distribution in the calculation domain according to the linear coordinates;

[0033] A beam depression angle acquisition module is used to calculate the depression angles of different beam paths according to the straight line coordinates;

[0034] The integral concentration calculation module is used to calculate the integral concentration obtained by the detector light path passing through the cloud according to the light path coordinates and the gas concentration at the corresponding position;

[0035] A probability distribution acquisition module is used to obtain the probability distribution of the maximum gas concentration of the cloud group passing through the corresponding light path in different gas concentration intervals according to the gas concentration peak values ​​on different light beam paths;

[0036] The training module is used to train a pre-established neural network pattern recognition model with the beam depression angle, wind speed and integrated concentration as input and the probability distribution of the gas concentration peak in multiple concentration intervals as output to obtain a trained model.

[0037] In an optional embodiment, it also includes: a gas concentration peak acquisition module, which is used to acquire the gas concentration peak according to the gas concentration on different light beam paths.

[0038] The third object of the present invention is to provide a readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the above-mentioned method for determining the concentration of a combustible gas leakage cloud.

[0039] A fourth object of the present invention is to provide an electronic device, the electronic device comprising at least one processor and a memory communicatively connected to the processor;

[0040] The memory stores a program that can be executed by the at least one processor, and the program is executed by the at least one processor so that the at least one processor can execute the above-mentioned method for determining the concentration of the combustible gas leakage cloud.

[0041] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0042] A method for determining the concentration of a combustible gas leak cloud provided by an embodiment of the present invention, on the one hand, obtains the cloud concentration distribution after gas leakage through numerical simulation, and calculates the integral concentration by extracting the coordinates of the light beam path at different angles and positions. Compared with the prior art solution in which the leaking gas cloud is assumed to be a standard gas pool with a fixed width (usually 1m), the technical solution of the present invention is more in line with the actual leakage situation, and the integral concentration database of the light path passing through the cloud under different conditions is more accurate;

[0043] On the other hand, the present invention uses a neural network pattern recognition model to explore the intrinsic relationship between the integral concentration, real-time wind speed, optical path depression angle and the maximum concentration of the gas cloud. Taking into account that the same integral concentration may correspond to different concentration distributions under different optical paths, pattern recognition can deduce the probability distribution of the peak value of the gas cloud concentration on the optical path under the integral concentration in each concentration interval, thereby comprehensively judging the leakage gas cloud concentration and the risk size based on the concentration probability distribution, avoiding the situation where the deviation or error of a single concentration prediction value may lead to misjudgment of the leakage, and the results obtained by the technical solution of the present invention are more reasonable. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0045] Figure 1 A flow chart of a method for determining the concentration of a combustible gas leak cloud provided by an embodiment;

[0046] Figure 2 A schematic diagram of a detector scanning a gas cloud to obtain integrated concentration according to a specific embodiment;

[0047] Figure 3 It is a schematic diagram of the natural gas diffusion concentration results on the detector beam path of a specific embodiment;

[0048] Figure 4 It is a schematic diagram of the integrated concentration, maximum concentration and beam depression angle of natural gas on different scanning paths of a detector in a specific embodiment;

[0049] Figure 5 It is a schematic diagram of the maximum concentration results on the beam path under different wind speed conditions in a specific embodiment;

[0050] Figure 6 It is a schematic diagram of the maximum concentration results on the beam path under different leakage rate conditions of a specific embodiment;

[0051] Figure 7 Schematic diagram of the probability distribution of the maximum value of gas cloud concentration on the beam path under different conditions of a specific embodiment;

[0052] Figure 8 A schematic structural diagram of a device for determining the concentration of a combustible gas leak cloud provided in a specific embodiment;

[0053] The components and corresponding marks in the accompanying drawings are:

[0054] 101-data acquisition module, 201-concentration calculation module, 301-coordinate acquisition module, 401-gas concentration extraction module, 501-beam depression angle acquisition module, 601-integral concentration calculation module, 701-probability distribution acquisition module, 801-training module, 901-gas concentration peak acquisition module. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0056] Example 1: A method for determining the concentration of a combustible gas leak cloud

[0057] like Figure 1 The process shown includes the following steps:

[0058] S1: Investigate the local meteorological historical data and leakage failure database of the target area to determine the wind speed and leakage rate range of the external wind field of the target area.

[0059] S2: Based on the historical leakage rate and wind speed, and by changing the leakage rate and wind speed, the concentration distribution of combustible gas in the calculation domain after leakage under different leakage rates and wind speeds is simulated and calculated.

[0060] Among them, when simulating the concentration distribution of combustible gas in the calculation domain after leakage, a computational fluid dynamics model can be used for simulation calculation (i.e., CFD simulation calculation), and the calculation domain of the computational fluid dynamics model is set to a rectangular space with a length, width, and height of A, B, and C respectively. The coordinates of the leakage point are set to (0, a, b), and the leaked gas flows in from the x=0 surface in the calculation domain together with the air. The bottom y=0 of the calculation domain is a rigid ground, and the other surfaces are free outlets.

[0061] It should be noted that CFD simulation calculation is a prior art, and the specific calculation method is not described in detail here. In addition, the calculation domain of this embodiment can also be a cube, and the shape of the calculation domain can be changed or shielding objects can be added according to actual needs.

[0062] It should be noted that the simulation calculation method of this embodiment is only a preferred simulation calculation method of the present invention. In addition to the CFD simulation calculation method of this embodiment, other existing technologies for obtaining concentration distribution, such as Gaussian diffusion equation, can also be applied to the present invention.

[0063] S3: According to the straight line coordinates of the beam paths emitted by the detector at different positions and at different angles in the calculation domain, the gas concentration on the corresponding beam path is extracted, and the beam depression angles of different optical paths are obtained;

[0064] The computational domain has pre-constructed spatial grid coordinates. The installation height of the linear beam combustible gas detector is set to H, and the horizontal distance from the leakage point is set to L, which represents the location of the detector. For the convenience of explanation, the section with coordinate z = 0 in the computational domain space is taken as an example. This section contains the leakage point, and the coordinates of the leakage point are set to (0, h, 0). The beam emitted by the detector is as follows: Figure 2 The cross section shown is scanned from the bottom to the top, and the beam path obtained is the multiple straight lines incident on the cross section. The multiple points where the light reaches the cross section are monitoring points. The beam path is the straight line light path from the position (L, H, 0) of the linear beam combustible gas detector to the monitoring point (0, a, 0). The straight line coordinates of each straight line light path in the calculation domain can be obtained, so as to extract the diffusion gas concentration distribution of the z=0 section in the calculation domain from the concentration distribution of the combustible gas in the calculation domain after the leakage obtained in S2. Among them, the method for obtaining the straight line coordinates belongs to the prior art and will not be repeated here.

[0065] Using the same method, change the installation height H of the combustible gas detector and the horizontal distance L from the leakage point; select different sections according to the location of the leakage point, and scan the light beam emitted by the detector from the bottom to the top of the selected section, so that the straight light path generated by the detector from different positions and angles can be obtained, and the straight line coordinates of these straight light paths can be obtained, so as to extract the diffusion gas concentration distribution on different sections in the calculation domain, that is, the gas concentration on different light beam paths.

[0066] The method for obtaining the beam depression angle of different optical paths is:

[0067]

[0068] β is the beam depression angle under the beam path, L is the horizontal distance between the detector and the leakage point, H is the installation height of the above-mentioned detector, a is the Y coordinate of the monitoring point mentioned above, and the light beam is emitted from the position where the detector is located and incident on the monitoring point.

[0069] In addition, according to the gas concentration on the beam path under different conditions (different installation heights, horizontal spacings, wind speeds, leakage rates, and different angles) obtained in the above steps, the corresponding gas concentration peak on the beam path can be obtained. The specific method is prior art and will not be described in detail here.

[0070] S4: Calculate the integrated concentration of the detector light path through the cloud based on the gas concentration at the corresponding position of the light path coordinates;

[0071] Specifically, according to the straight-line coordinates of different beam paths and the diffuse gas concentration on the beam path, the integral concentration of the leakage gas cloud concentration on the beam path from the position (L, H, 0) of the linear beam combustible gas detector to the monitoring point (0, a, 0) is calculated. Thus, the integral concentration under different installation heights H, horizontal spacing L, wind speeds and leakage rates, and beam paths can be obtained.

[0072] Preferably, the integral concentration is calculated using the approximate rectangular method, and the specific calculation process is:

[0073] The process of the beam scanning from the bottom to the top of the computational domain is as follows: Figure 2 When the light beam is emitted from the detector (L, H, 0) and incident on the monitoring point (0, a, 0) on the straight line, the path of the light beam on the concentration section can be determined by geometric relationships as follows:

[0074]

[0075] Extract the x on the beam path = [x1, x2, ..., x n ] coordinates, corresponding to gas concentration c = [c1, c2, ... c k ], since the computational domain grid size is small, the approximate rectangular method can be used to calculate the line integral of the leakage gas cloud concentration on the beam path:

[0076] The integrated concentration is calculated according to the formula (1).

[0077]

[0078] The integral concentration calculation formula for the z=0 section is obtained

[0079]

[0080] Where: IC is the integrated concentration on the beam path; n is the number of grids on the x-axis of the z=0 section; x i is the x-direction length of grid i; c i is the gas concentration of grid i; H is the vertical height of the linear beam combustible gas detector from the leak point; a and b are the coordinates of the monitoring point on the y and z axes respectively; L is the horizontal distance between the linear beam combustible gas detector and the leak point, and Z is the z-axis coordinate of the detector. Since the selected section is z = 0, Z and b in formula (1) are both equal to 0. The computational domain grid is a prior art, and the number of computer grids, the coordinates of each grid, etc. are obtained by conventional methods in the field, which will not be elaborated here.

[0081] The monitoring point (0, a, 0) is raised from (0, 0, 0) to (0, B, 0), and the interval of each rise is r, so that B / r groups of data are obtained; B is the Y-axis boundary of the calculation domain. The detector position (L, H, 0), the monitoring point coordinates (0, a, 0) and the rising interval r of the monitoring point are only examples and can be adjusted as needed.

[0082] Then, by changing the installation height H, horizontal spacing L, wind speed and leakage rate, the integrated concentration under different installation heights H, horizontal spacing L, wind speeds, leakage rates and light beam paths can be obtained, totaling t=B / r×p×q×n×m groups of data, where p, q, n, and m are the number of groups of installation height H, horizontal spacing L, wind speed, and leakage rate, respectively.

[0083] S5: taking the beam depression angle, wind speed and integrated concentration as input and the probability distribution of the gas concentration peak in multiple concentration intervals as output, training a pre-established neural network pattern recognition model to obtain a trained model;

[0084] The method for obtaining the probability distribution of the gas concentration peak in multiple concentration intervals is as follows:

[0085] T groups of data are constructed under different beam paths and leakage diffusion conditions, where the beam depression angle, wind speed and integrated concentration constitute a t×3 input matrix, and the corresponding gas concentration peak on the beam path is a t×1 output matrix.

[0086] Determine N concentration intervals [0~d1, d1~d2...d n-1 ~d n ], and determine the concentration interval d according to the gas concentration peak value on the beam path i-1 ~d i , and mark it as 1, and mark other concentration intervals as 0, forming [0,0..1...0]. According to the above method, the gas concentration peak t×1 output matrix is ​​converted into a t×N 01 matrix.

[0087] A neural network pattern recognition model is established, a t×3 matrix consisting of a beam pitch angle, a wind speed and an integrated concentration is used as input data of the neural network pattern recognition model, and a t×N 01 matrix consisting of a probability distribution of a gas concentration peak in N concentration intervals is used as output data of the neural network pattern recognition model; the t groups of data are divided into a training set, a validation set and a test set, and the neural network pattern recognition model is trained using the t groups of data to obtain a trained neural network pattern recognition model.

[0088] Preferably, a double-layer feedforward neural network pattern recognition model is used. When establishing the double-layer feedforward neural network pattern recognition model, a sigmoid transfer function is set in the hidden layer, the number of hidden layer neuron layers is Q, and a softmax transfer function is added in the output layer, and a probability value is assigned to each output classification result.

[0089] Preferably, when training the double-layer feedforward neural network pattern recognition model, the training is terminated when the total number of training rounds reaches threshold one, or the model gradient is less than threshold two, or the error does not rise after multiple iterations. After the training is terminated, the model performance and error are analyzed by confusion matrix, ROC curve and performance curve, and whether the model meets the requirements is determined based on the analysis results. When the model performance does not meet the requirements, the double-layer feedforward neural network pattern recognition model can be retrained, or the number of hidden layer neurons of the double-layer feedforward neural network pattern recognition model can be increased, or the proportion of the training set can be increased to increase the amount of training data.

[0090] It should be noted that the double-layer feedforward neural network pattern recognition model, confusion matrix, ROC curve and performance curve are all prior art, and the specific structure or method will not be repeated here. In addition, in addition to the neural network pattern recognition model adopted in this embodiment, the neural network structure can be optimized according to actual needs, or other big data modeling methods such as deep learning can be used to establish the internal connection between the variables.

[0091] S6: Input the data into the trained model to obtain the probability distribution of the maximum gas concentration of the corresponding light path passing through the cloud in different gas concentration ranges to determine the leakage gas cloud and the concentration of the leakage gas cloud. The data includes the integrated concentration monitored by the linear beam combustible gas detector, the instantaneous wind speed and the beam depression angle.

[0092] Example 2: A method for determining the concentration of a combustible gas leak cloud

[0093] In this embodiment, taking a natural gas leak at a certain place as an example, the method for determining the concentration of a combustible gas leak cloud specifically includes the following steps:

[0094] (1) Investigate local meteorological historical data and leakage failure database to determine that the local wind speed is less than 2m / s and the maximum leakage rate is less than 10m 3 / h.

[0095] (2) A computational fluid dynamics model for gas leakage diffusion is established. The computational domain is a cuboid with length, width and height of A = 0.8 m, B = 0.5 m and C = 0.5 m respectively. The leakage point is located at (0, 0.2, 0). The natural gas leakage rate is 0.012 m 3 / h, the wind speed is 0m / s, and the concentration distribution of natural gas in the calculation domain after leakage is simulated.

[0096] (3) The installation height of the linear beam detector is H = 4m, and the horizontal distance from the leakage point is L = 50m.

[0097] The distribution of diffuse gas concentration on the z = 0 section in the calculation domain. The light beam is emitted from the detector (50,4,0) and incident on the monitoring point (0,a,0) on the straight line. The path of the light beam on the concentration section is determined by geometric relationship as follows:

[0098]

[0099] Taking a=0 as an example, the beam path is:

[0100] y=0.08x 0≤x≤0.8 (6)

[0101] Extract the x-coordinate on the beam path and the corresponding gas concentration c. The results are as follows: Figure 3 As shown. Using formula (1)

[0102] -(2) (Same as Example 1) The integrated concentration of the leakage gas cloud on the light beam path is calculated to be IC=154.83ppm·m, the light beam depression angle is 1.49 (85.4°), and the corresponding gas concentration peak value on the light path is 0.0989%.

[0103] (4) Raise the monitoring point from (0,0,0) to (0,0.5,0) with an interval of 0.02, and simulate the beam from the bottom of the computational domain.

[0104] Scan from the bottom to the top and calculate the integrated concentration under different beam paths. The wind speed is 0m / s and the leakage rate is 0.012m 3 / h, detector installation height 4m, horizontal spacing 50m, integrated concentration IC, gas concentration peak Cmax and beam depression angle β under different paths during beam scanning, a total of B / r=0.5 / 0.02=25 sets of data, the results are as follows Figure 4 shown.

[0105] (5) Changing the detector height, horizontal spacing, wind speed and leakage rate, respectively: H = 4, 6, 8, 10 m, L = 5, 10, 50 m, v = 0, 1, 2 m / s, q = 0.012, 0.036, 5, 10 m 3 / h. Repeat steps (2)-(4) to obtain the integrated concentration IC, gas concentration peak Cmax, and beam depression angle β of different scanning paths under different wind speeds, leakage rates, detector installation heights, and horizontal spacings. Since the detector height, horizontal spacing, wind speed, and leakage rate are set to 4, 3, 3, and 4 groups respectively, there are a total of t = 25 × 4 × 3 × 3 × 4 = 3600 groups of data, among which the results under different wind speed conditions are as follows: Figure 5 The test results under different leakage rate conditions are shown in Figure 6 shown.

[0106] (6) Considering that the lower explosion limit of natural gas is 5-15%, and the alarm threshold is usually set at 25% and 50% of the lower explosion limit, the final design determines 7 concentration intervals, namely [0-0.1%], [0.1%-0.5%], [0.5%-1%], [1%-2.5%], [2.5%-5%], [5%-15%], and [15%-100%]. The 3600 sets of data are extracted and summarized. The beam depression angle, wind speed and integrated concentration form a 3600×3 input matrix. The concentration interval d is determined according to the gas concentration peak Cmax on the beam path in the 3600 sets of data. i-1 ~d i , and mark it as 1, and mark other concentration intervals as 0, forming [0,0..1...0]. For example, Cmax=7% can be represented as [0 000 0 1 0], and Cmax=0.05% can be represented as [1 0 0 0 0 00]. The 3600×1 output matrix of gas concentration peak value is converted into a 3600×7 01 matrix using the above method.

[0107] (7) A two-layer feedforward neural network pattern recognition model is established. The sigmoid transfer function is set in the hidden layer, and the softmax transfer function is added to the output layer. A probability value is assigned to each output classification result. The number of hidden layer neurons is Q = 10, where the input data is the optical path pitch angle β, wind speed v, and integrated concentration IC. The output data is the 7 concentration interval numbers where the maximum cloud concentration Cmax on the optical path is located. The t = 3600 group data is split into a training set, a validation set, and a test set, which occupy 70%, 15%, and 15% of the group data, respectively.

[0108] (8) Train the neural network pattern recognition model, and evaluate the model performance and error through confusion matrix, ROC curve and performance curve to obtain the trained neural network pattern recognition model.

[0109] (9) Using the trained neural network pattern recognition model, the real-time integrated concentration IC, instantaneous wind speed v and optical path depression angle β obtained by the detector scan are input, as shown in Table 1:

[0110] Table 1 Input data of neural network pattern recognition model

[0111]

[0112]

[0113] The model output is the probability distribution P of the peak gas concentration Cmax in the seven concentration intervals when the light beam passes through the cloud. The results are as follows: Figure 7 As shown. Figure 7It can be seen that when the integrated concentration is 13500ppm·m, the wind speed is 0m / s, and the depression angle is 81°, the probability of the peak value of the natural gas concentration on the beam path is 97% in the concentration range of [15%~100%], which is a large leak, and the gas concentration has exceeded the upper explosion limit. When the integrated concentration is 200ppm·m, the wind speed is 1m / s, and the depression angle is 74°, the probability of the peak value of the natural gas concentration on the beam path is 37% in the concentration range of [0~0.1%], 48% in the concentration range of [0.1%~0.5%], and 18% in the concentration range of [0.5%~1%], which is probably a small leak. According to the probability distribution, the size of the leaking gas cloud and concentration can be judged, so as to take corresponding response measures.

[0114] In summary, the embodiment of the present invention obtains the cloud concentration distribution after gas leakage through numerical simulation, and establishes the relationship between the integral concentration and the maximum concentration value under the light beam path, so that the concentration and risk of the leaking gas cloud can be directly judged according to the "integral concentration", and the result is more in line with the actual leakage situation. The large database of integral concentrations of light paths passing through the cloud under different conditions established thereby is more accurate; the concentration and risk of the leaking gas cloud are comprehensively judged through the concentration probability distribution, avoiding the situation where the deviation or error of a single concentration prediction value may lead to misjudgment of the leakage. The results obtained by the technical solution of the present invention are more reasonable, and compared with the prior art, it is a significant improvement. If the concentration and risk of the leaking gas cloud are directly judged according to the "integral concentration", it is impossible to set a reasonable alarm threshold.

[0115] Embodiment 3:

[0116] Corresponding to the above method embodiment, the embodiment of the present invention further provides a device for determining the concentration of a combustible gas leakage cloud. The device for determining the concentration of a combustible gas leakage cloud described below and the method for determining the concentration of a combustible gas leakage cloud described above can be referred to each other.

[0117] See also Figure 8 , a device for determining the concentration of a combustible gas leak cloud, comprising:

[0118] The data acquisition module 101 is used to acquire the leakage rate and wind speed data of the wind field outside the target area;

[0119] The concentration distribution calculation module 201 is used to calculate the concentration distribution of the leaked gas in the three-dimensional space calculation domain according to the leakage rate and wind speed;

[0120] A coordinate acquisition module 301 is used to acquire the straight line coordinates of different light beam paths emitted by the linear light beam combustible gas detector within the calculation domain;

[0121] A gas concentration extraction module 401 is used to extract the gas concentration on the corresponding light beam path from the concentration distribution in the calculation domain according to the linear coordinates;

[0122] A beam depression angle acquisition module 501 is used to calculate the depression angles of different beam paths according to the straight line coordinates;

[0123] The integral concentration calculation module 601 is used to calculate the integral concentration obtained when the detector light path passes through the cloud according to the light path coordinates and the gas concentration at the corresponding position;

[0124] The probability distribution acquisition module 701 is used to obtain the probability distribution of the maximum gas concentration of the cloud group passing through the corresponding light path in different gas concentration intervals according to the gas concentration peak values ​​on different light beam paths;

[0125] The training module 801 is used to train a pre-established neural network pattern recognition model with the beam pitch angle, wind speed and integrated concentration as input and the probability distribution of the gas concentration peak in multiple concentration intervals as output to obtain a trained model.

[0126] Also includes:

[0127] The gas concentration peak value acquisition module 901 is used to acquire the gas concentration peak value according to the gas concentration on different light beam paths.

[0128] When monitoring combustible gas leaks, investigate local meteorological historical data and leakage failure databases to obtain local leakage rates and wind speeds. Based on the obtained leakage rates and wind speeds, and by changing the leakage rates and wind speeds, simulate and calculate the concentration distribution of combustible gas leaks in the calculation domain at different leakage rates and wind speeds; extract the gas concentration on the corresponding beam path according to the linear coordinates of the beam path emitted by the detector at different positions and angles in the calculation domain, and obtain the beam depression angles of different light paths; calculate the integral concentration obtained by the detector light path passing through the cloud based on the light path coordinates and gas concentration; train the pre-established neural network pattern recognition model with the beam depression angle, wind speed and integral concentration as inputs and the probability distribution of the gas concentration peak in multiple concentration intervals as outputs to obtain a trained model; finally, input the integral concentration, instantaneous wind speed and beam depression angle monitored by the combustible gas detector into the trained model to obtain the probability distribution of the maximum gas concentration of the corresponding light path passing through the cloud in different gas concentration intervals to determine the leaking gas cloud and the concentration of the leaking gas cloud. Therefore, the concentration and risk of the leaking gas cloud can be judged based on the "integral concentration", so that the alarm threshold and rapid response measures can be set more scientifically and reasonably, which is of great significance to ensuring the safe production of hazardous chemicals.

[0129] Embodiment 4:

[0130] Corresponding to the above method embodiment, the embodiment of the present invention further provides a computer-readable storage medium, and the computer-readable storage medium described below and the method for determining the concentration of a combustible gas leakage cloud described above can be referenced to each other.

[0131] A computer-readable storage medium stores a computer program, wherein the computer program is executed by a processor to implement the steps of the above-mentioned method for determining the concentration of a combustible gas leak cloud.

[0132] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0133] Embodiment 5:

[0134] Corresponding to the above method embodiment, the embodiment of the present invention further provides an electronic device, and the electronic device described below and the method for determining the concentration of a combustible gas leakage cloud described above can refer to each other.

[0135] An electronic device comprising at least one processor and a memory communicatively connected to the processor;

[0136] The memory stores a program executable by the at least one processor, and the program is executed by the at least one processor so that the at least one processor can execute the steps of the method for determining the concentration of a combustible gas leak cloud as described above.

[0137] The methods, processes, equipment, etc. not mentioned in this patent application are all obtained or performed by using existing methods or by using existing methods.

[0138] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0139] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for determining the concentration of a combustible gas leak cloud, characterized in that: The method is performed based on a linear beam combustible gas detector, and the method comprises: S1: Obtain historical leakage rate and wind speed data of the target area; S2: Simulate and calculate the concentration distribution of gas leakage in the calculation domain under different leakage rates and wind speeds; S3: According to the straight line coordinates of the beam paths emitted by the detector at different positions and at different angles in the calculation domain, the gas concentration on the corresponding beam path is extracted, and the beam depression angles of different optical paths are obtained; S4: Based on the optical path coordinates and the gas concentration at the corresponding position, the integrated concentration obtained when the detector optical path passes through the cloud is calculated; S5: Using the beam pitch angle, wind speed and integrated concentration as inputs and the probability distribution of the gas concentration peak on the beam path in multiple concentration intervals as outputs, a pre-established neural network pattern recognition model is trained; S6: Input the data into the trained model to obtain the probability distribution of the maximum gas concentration of the corresponding light path passing through the cloud in different gas concentration ranges, so as to determine the concentration of the leaking gas cloud; The integral concentration is calculated using the approximate rectangular method; The calculation method of the integrated concentration is: IC is the integrated concentration on any beam path; H, L and Z are the vertical height and horizontal spacing of the detector from the leakage point and the z-axis coordinate of the detector location, respectively; a and b are the coordinates of the monitoring point on the y and z axes, respectively; n is the number of grids on the x-axis of any section containing the leakage point and within the calculation domain; x i is the x-direction length of the grid i in the computational domain; c i is the gas concentration of grid i in the computational domain.

2. A method for determining the concentration of a combustible gas leak cloud according to claim 1, characterized in that: In the step S2, a computational fluid dynamics model or a Gaussian diffusion equation is used to perform simulation calculations to obtain the concentration distribution of the gas in the calculation domain after the gas leaks.

3. A method for determining the concentration of a combustible gas leak cloud according to claim 2, characterized in that: In step S3, the method for acquiring the depression angles of the light beams of the different light paths is: β is the beam depression angle under the beam path, L is the horizontal distance between the detector and the leakage point, H is the vertical height between the detector and the leakage point, a is the Y coordinate of the monitoring point, and the light beam is emitted from the location of the detector and incident on the monitoring point.

4. A method for determining the concentration of a combustible gas leak cloud according to claim 1, characterized in that: In step S5, the gas concentration peak is the maximum value on the path of the light beam emitted by the detector passing through the leaking gas cloud; The method for obtaining the probability distribution of the gas concentration peak in multiple concentration intervals is: Under different beam paths and leakage diffusion conditions, t groups of data are formed. The beam depression angle, wind speed and integrated concentration form a t×3 input matrix, and the corresponding gas concentration peak on the beam path is a t×1 output matrix. The gas concentration peak t×1 output matrix is ​​converted into a t×N 01 matrix. The conversion method is: determine N concentration intervals [0~d1, d1~d2...d n-1 ~d n ], and determine the concentration interval d according to the gas concentration peak value on the beam path i-1 ~d i , and mark it as 1, and mark other concentration intervals as 0, forming [0,0..1...0]. In the same way, the gas concentration peak t×1 output matrix is ​​converted into a t×N 01 matrix; The training input of the neural network pattern recognition model is a t×3 matrix composed of the beam depression angle, wind speed and integrated concentration, and the training output is a t×N 01 matrix composed of gas concentration peak values.

5. A method for determining the concentration of a combustible gas leak cloud according to claim 4, characterized in that: In step S5, a double-layer feedforward neural network pattern recognition model is adopted. When establishing the model, a sigmoid transfer function is set in the hidden layer, the number of hidden layer neuron layers is Q, and a softmax transfer function is added in the output layer, and a probability value is assigned to each output classification result.

6. A method for determining the concentration of a combustible gas leak cloud according to claim 5, characterized in that: In step S6, the integrated concentration of the cloud cluster passed by the detector light path, the instantaneous wind speed and the beam depression angle of the light beam path are input into the trained model.

7. A device for determining the concentration of a combustible gas leak cloud, characterized in that: include: A data acquisition module, used to acquire leakage rate and wind speed data; The concentration distribution calculation module is used to calculate the concentration distribution of the leaked gas in the calculation domain according to the leakage rate and wind speed; A coordinate acquisition module, used to acquire the straight line coordinates of different beam paths emitted by the linear beam combustible gas detector within the calculation domain; A gas concentration extraction module is used to extract the gas concentration on the corresponding beam path from the concentration distribution in the calculation domain according to the linear coordinates; A beam depression angle acquisition module is used to calculate the depression angles of different beam paths according to the straight line coordinates; The integral concentration calculation module is used to calculate the integral concentration obtained by the detector light path passing through the cloud according to the light path coordinates and the gas concentration at the corresponding position; A probability distribution acquisition module is used to obtain the probability distribution of the maximum gas concentration of the cloud group passing through the corresponding light path in different gas concentration intervals according to the gas concentration peak values ​​on different light beam paths; The training module is used to train a pre-established neural network pattern recognition model with the beam depression angle, wind speed and integrated concentration as input and the probability distribution of the gas concentration peak in multiple concentration intervals as output to obtain a trained model.

8. A device for determining the concentration of a combustible gas leak cloud as claimed in claim 7, characterized in that: Also includes: The gas concentration peak value acquisition module is used to obtain the gas concentration peak value according to the gas concentration on different light beam paths.

9. A computer-readable storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method for determining the concentration of a combustible gas leakage cloud according to any one of claims 1 to 6.

10. An electronic device, characterized in that: The electronic device comprises at least one processor and a memory communicatively connected to the processor; The memory stores a program executable by the at least one processor, and the program is executed by the at least one processor so that the at least one processor can execute the method for determining the concentration of a combustible gas leakage cloud as described in any one of claims 1 to 6.

Images