Rapid identification method for public transport building indoor biological factor propagation source

Abstract

The invention provides a rapid identification method for a public transport building indoor biological factor propagation source, and the method comprises the following steps: carrying out the analysis before an abnormal source event occurs, obtaining the positions of all potential abnormal sources, obtaining an indoor flow field of a public transport building through CFD, and building a response factor matrix between each potential abnormal source and a monitoring point. And after an abnormal source event occurs, a traversal method is used to calculate the correlation between sensor monitoring data and concentration data under different combinations in the first-stage data set, and the similarity between the monitoring data and CFD simulation data is evaluated, so that the position and the source intensity of an abnormal source are identified. The first objective of the invention is to systematically solve the problem of factor emission source identification in the public transport building space, accurately position the emission source of harmful substances, guide timely pollution source management, and powerfully guarantee public health. The indoor environment quality is improved, the operation management efficiency is improved, and the comprehensive development of the traffic building field is promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of indoor air pollutant source identification in public transportation buildings, and in particular relates to a method for indoor multi-biological factor source identification in public transportation buildings. Background Art

[0002] With the rapid development of the national economy, large public transportation buildings are increasing, such as high-speed rail stations, subway stations, and airport terminals. These places are usually characterized by closed spaces and dense population. Once biological factors appear indoors in these places, they will be more likely to cause large-scale infection and spread, posing a major threat to the life, health and safety of station staff and passengers. The traceability methods for biological factors in the air can be mainly divided into fixed sensor methods and mobile sensor methods based on their monitoring carriers. Among them, the mobile sensor method mainly uses mobile robots, drones and other carriers to carry factor monitoring sensors. After the biological factors appear, the mobile sensors are moved toward the source of the factor emission with the help of relevant algorithms, and then the location of the emission source is determined. This method has better adaptability to unknown environments, but it is usually impossible to carry out long-term continuous monitoring, that is, it is impossible to determine the occurrence of factor transmission events in the first place. The fixed sensor method arranges multiple fixed biological factor sensors in the venue. When the sensor detects the concentration data of the relevant factor, the data can be used to reverse the source in the first place. The current method of reverse identification of factor sources based on fixed sensors can be divided into backward method and forward method. The backward method refers to the determination of relevant information of the source by solving the propagation equation of the factor in the air through limited monitoring data. This method requires less data, but the solution time is very long, which will affect the speed of factor tracing. In addition, when the number of sources is unknown, the solution efficiency of the method will be further affected. The forward method models the possible propagation of the factor before the factor appears; then, after the factor propagation event occurs, the source is identified by comparing the monitored and simulated concentrations. This method is fast in the source identification stage, but there is no suitable method for the identification of multiple unknown number of factor emission sources. Therefore, it is urgent to propose a method based on fixed sensors for rapid identification of source characteristics when the number of biological factor emission sources is unknown. Summary of the invention

[0003] In order to solve the current technical problems, the present invention proposes a method for quickly identifying the source of indoor biological factors in public transportation buildings, which provides a prerequisite for timely identifying the source of biological factors, predicting the spread of transmission, and taking relevant control measures in a timely manner.

[0004] To achieve the above objectives, the method for rapid biological factor source identification proposed by the present invention comprises the following steps:

[0005] The method in the invention mainly combines the computational fluid dynamics simulation (CFD) method and the Pearson correlation determination method, and the method can be divided into two stages. First, in the first stage, before the factor propagation event occurs, CFD simulation is performed on the corresponding public building to obtain the indoor airflow field of the building. Then, tracer gas is released at the locations where all potential factor release sources may appear, and the concentration monitoring data of the corresponding sensor position is obtained. Then, the concentration data set of the sensor position under different source numbers and position combinations is obtained through rapid calculation.

[0006] Assuming the flow field is stable, the concentration response at the sensor location is linearly related to the source intensity. For multiple sensors, the concentration response of the i-th sensor to the j-th pollution source can be expressed as:

[0007] c ij =f ij ·q j (1)

[0008] where c ij and f ij The i th The sensor pair j th Considering n potential factor propagation sources, the concentration response of the i-th sensor is the superposition of the effects of n factor sources:

[0009]

[0010] where c i is the concentration response of the ith sensor to the n factor diffusion sources. Therefore, the concentration response of m sensors to n diffusion sources can be expressed as:

[0011]

[0012] Or written as:

[0013] c=Fq (4)

[0014] Among them, c is the concentration response vector, F is the response factor matrix, and q is the source intensity vector. If the safety weaknesses are analyzed and all potential release locations of biological factors are obtained, the corresponding response factors can be obtained through CFD simulation.

[0015] When a biological factor release source appears at the kth location in a building, the monitored biological factor response concentration can be expressed as:

[0016]

[0017] in is the release intensity of the biological factor release source appearing at the kth position, c 1 is the concentration detected by the sensor.

[0018] Formula (5) can be rewritten as:

[0019]

[0020] where F(:,k) is the k-th row of F.

[0021] Formula (6) shows that F(:,k) and c 1 Therefore, by finding which column of the response factor matrix F corresponds to the monitored concentration vector c 1 Linear correlation can determine the location of a single pollution source. For the jth column of matrix F, it is related to c 1 The correlation is calculated by the Pearson correlation coefficient P in formula 4-11 xy Calculation, this coefficient is widely used to measure the degree of correlation between two vectors, the correlation coefficient ranges from -1 to 0 for negative correlation, and from 0 to 1 for positive correlation. xy The closer to 1, the closer the two vectors F(:,k) and c 1 -c 0 The stronger the positive correlation is.

[0022]

[0023] In the source identification process, the correlation coefficient P is calculated by looping j from 1 to n. xy , if P xy is closest to 1, then the location of the abnormal source can be determined by k=j.

[0024] When there are multiple factor release sources, two sources k 1 , k 2 For example, the concentration detected by the sensor after the factor propagation source appears can be written as:

[0025]

[0026] Formula (11) can be written as:

[0027]

[0028] It can be found that F(:,k 1 ) and F(:,k 2 ) and c 1 In order to identify k 1 and k 2 , by adding j 1 Loop from 1 to m-1 and set j 2 From j 1+1 loop to m, calculate F, F(:,j 1 )+aF(:,j 2 ) in any two columns (j 1 <j 2 ) and the monitored concentration vector c 1 Ideally, This is unknown. However, they can usually be assumed. Since biological factors also exist in certain concentrations in public places under normal conditions, this method assumes that the maximum source intensity of biological factors under normal conditions is When the source intensity is greater than When the biological factor dissemination event occurs, the source intensity after the occurrence is affected by Restriction, where 1 lb ub . Therefore, the range of a is:

[0029]

[0030] where b lb and b ub are the lower and upper limit coefficients, which are assumed to be 2 and 10 respectively in this example. 1 and j 2 , a is divided into h uniformly distributed steps from the lower limit (b lb –1) / (b ub -1) loop to the upper limit (b) ub -1) / (b lb -1).

[0031] Assume a 1 =1, a 2 = a, we can define a column vector f 2 :

[0032] f 2 =a 1 F(:,j 1 )+a 2 F(:,j 2 ) (14)

[0033] In order to identify the two sources of factor spread, it is necessary to calculate:

[0034]

[0035] By finding the correlation coefficient closest to 1, the source position k can be determined 1 =j 1 and k 2 =j 2 .

[0036] ​​If there are l factor propagation sources, source identification is to calculate And find the correlation coefficient closest to 1, where: j 1 <j 2 <... <j l .

[0037]

[0038] If the number of sources is unknown, give an upper bound l max , this method needs to loop from 1 to l max To calculate the correlation coefficient, this will be a total of Calculation times.

[0039] If there exists a scalar w such that Then we have:

[0040]

[0041] This means that when and When the vectors are not independent, there are multiple solutions. Therefore, if the column vectors of the matrix F are not independent, there may be multiple solutions. For an m×n matrix F, the column rank of the matrix F is rank(F)≤min(m,n). Therefore, to avoid multiple solutions, the number of sensors should be greater than or equal to the number of potential sources, m≥n. In this study, the minimum number of sensors (m=n) is used to ensure that there are no multiple solutions.

[0042] After determining the source location, for a source, the source intensity can be calculated based on the concentration data monitored by any of the m sensors and the corresponding response factor:

[0043]

[0044] For l factor spreading sources, the source intensity can be calculated based on the concentration data monitored by any l sensors among m sensors (l≤m) and the response factor matrix, as shown in the formula:

[0045]

[0046] Beneficial effects of the present invention:

[0047] The present invention realizes the rapid identification of the source of indoor biological factor transmission in public transportation buildings, systematically solves the problem of difficult identification and low identification efficiency of factor emission sources in public transportation building spaces, accurately locates the source of harmful substance emission, guides timely pollution source management, effectively protects public health, improves indoor environmental quality, improves operational management efficiency, and promotes comprehensive development in the field of transportation buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a diagram that summarizes the method;

[0049] Figure 2 Flowchart for factor source identification;

[0050] Figure 3 Pseudo code diagram for factor source identification method;

[0051] Figure 4 It is a statistical graph of the number of cases of random anomaly sources;

[0052] Figure 5 This is the identification result diagram of the random abnormal source case; DETAILED DESCRIPTION

[0053] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0054] Example:

[0055] In engineering applications, the number of anomaly sources is usually unknown, but a maximum number can be assumed, which is assumed to be 3 in this case study. First, calculate the relationship between each column of F and c 1 -c 0 Then, calculate f 2 、f 3 and c 1 -c 0 The correlation coefficient with P max The corresponding source position is recorded as the abnormal source. r (r=1,2,...,l), the source intensity is calculated using the concentrations monitored by l sensors.

[0056] A r The step size between the lower and upper limits of is set to 0.05. h for two and three anomaly sources is 450 and 202500, respectively. By setting l max =3 Generate 5000 cases with a random number (one, two, or three) of anomaly sources. Figure 4 The statistics of these random cases are shown. The number of cases with different numbers of anomaly sources and locations where the anomalies occur is not much different.

[0057] The abnormal identification results are as follows Figure 5 As shown in (a), the accuracy is 82.54%. Among the 873 misjudgment cases, two types of errors were found. In 72 cases, the number of abnormal sources was correctly identified, but the location was incorrectly identified. In the remaining 801 cases, the number of abnormal sources was incorrectly identified, and two abnormal sources were identified as three abnormal sources. In order to minimize the second error, the a of the two abnormal sources was set to rThe step size is refined from 0.05 to 0.01. For the same 5000 cases, the recognition results are as follows Figure 5 As shown in (b), the source identification accuracy is improved from 82.54% to 93.52%, and the number of cases where the number of abnormal sources are misidentified is reduced to 253. Figure 5 (c) Comparison of the time required for the source identification case of the two strategies. It can be observed that the time taken by the two strategies is 49.3 seconds and 47.8 seconds respectively, which is quite close.

Claims

1. A method for quickly identifying the source of indoor biological factor transmission in public transportation buildings, characterized by: The method combines the computational fluid dynamics (CFD) simulation method and the Pearson correlation determination method. The method is divided into two stages, namely, data preparation before the factor propagation event occurs and identification of the propagation source information after the propagation event occurs.

2. The data preparation stage before the factor propagation time occurs according to claim 1, characterized in that: CFD simulation is performed on the corresponding public buildings to obtain the indoor airflow field of the buildings. Then, tracer gas is released at the locations where all potential factor release sources may appear to obtain concentration monitoring data at the corresponding sensor locations. Then, the concentration data set of the sensor position under different source quantity and position combinations is obtained through rapid calculation.

3. According to claim 1, the stage of identifying the source of transmission information after the transmission event occurs is characterized by: When a known biological factor transmission event occurs, the traversal method is used to calculate the correlation between the sensor monitoring data and the concentration data under different combinations in the first phase data set. Then the correlation is determined, and the source characteristics corresponding to the data set with the highest correlation are the number and location of the actual biological factor release source.

4. The data preparation phase before the factor propagation time occurs according to claim 2, characterized in that: The indoor flow field of the building is obtained by the CFD method and the concentration monitoring data of the corresponding sensor position is obtained by the tracer gas method. The concentration response of m sensors to n diffusion sources is obtained by fast calculation: Or written as c=Fq.

5. The concentration response of m sensors to n diffusion sources according to claim 4, characterized in that: Assuming the flow field is stable, the concentration response at the sensor location is linearly related to the source intensity. For multiple sensors, the concentration response of the i-th sensor to the j-th pollution source can be expressed as c ij =f ij ·q j , c ij and f ij The i th The sensor pair j th There are n potential factor diffusion sources, and the concentration response of the i-th sensor is the superposition of the effects of the n factor sources. c i is the concentration response of the ith sensor to the n-factor diffusion source.

6. The method for identifying the source of transmission information after a transmission event occurs according to claim 1, characterized in that: When there is only one transmission source at the kth position, the monitored biological factor response concentration is: is the release intensity of the biological factor release source appearing at the kth position, c 1 is the concentration detected by the sensor. It can be rewritten as Calculate the columns of the response factor matrix F and the monitored concentration vector c 1 Pearson correlation coefficient, P xy The position closest to 1 is the location of the diffusion source.

7. The method for identifying the source of transmission information after a transmission event occurs according to claim 1, characterized in that: When there are two factor propagation sources, k1 and k2, the concentration monitored by the sensor after the factor propagation source appears can be written as: By looping \(j1\) from 1 to \(m - 1\) and looping \(j2\) from \(j1+1\) to \(m\), calculate the correlation coefficient between the linear combination of any two columns (\(j1 < j2\)) in \(F\), \(F(:,j1)+aF(:,j2)\) and the monitoring concentration vector \(c\). Define a column vector \(f2 = a1F(:,j1)+a2F(:,j2)\), calculate the correlation coefficient between \(f2\) and \(c\). By finding the correlation coefficient closest to 1, the source positions \(k1 = j1\) and \(k2 = j2\) can be determined. 1 Between them. Define a column vector \(f2 = a1F(:,j1)+a2F(:,j2)\), calculate the correlation coefficient between \(f2\) and \(c\). 1 Between them. By finding the correlation coefficient closest to 1, the source positions \(k1 = j1\) and \(k2 = j2\) can be determined.

8. The weighting coefficient a of each diffusion source response matrix according to claim 7, characterized in that: Assume that the maximum source intensity of biological factors under normal conditions is When the source intensity is greater than When the biological factor dissemination event occurs, the source intensity after the occurrence is affected by Restriction, where 1 lb ub . Then the range of a is b lb and b ub are the lower and upper limit coefficients, respectively, and can be assumed in practice. For a specific combination j1 ​​and j2, a is uniformly distributed from the lower limit (b lb –1) / (b ub -1) loop to the upper limit (b) ub -1) / (b lb -1).​​ 9. The method for identifying the source of transmission information after a transmission event occurs according to claim 1, characterized in that: If there are l factors spreading sources, source identification is to calculate And find the correlation coefficient closest to 1, where: j1 <j2<...<j l The response factor matrix for multiple source weights is Coefficient a r The range is 10. The method for identifying the source of transmission after a transmission event occurs according to claim 1, characterized in that: If the number of sources is unknown, we can give an upper limit l max , this method needs to loop l from 1 to l max To calculate the correlation coefficient, this will be a total of After determining the source location, for a source, the source intensity can be calculated based on the concentration data monitored by any of the m sensors and the corresponding response factor. For a source with l factors, the source intensity can be calculated based on the concentration data monitored by any l sensors among m sensors (l≤m) and the response factor matrix:

Images