AI intelligent traceability data processing method

Abstract

The invention discloses an AI intelligent traceability data processing method, which comprises the following steps: S1, monitoring point deployment: deploying VOCPprobe monitoring points at sensitive points, transmission channels and pollution sources by utilizing the principle that foul and peculiar smell gas diffuses along with the wind direction, and integrating online wind direction and wind speed data to obtain detection data of fine monitoring granularity. VOCPprobe is deployed at an emission source, a sensitive point and a transmission path for online monitoring, full space-time supervision and all-weather 24-hour * 30-second fine granularity monitoring of the emission source, a transmission channel and the sensitive point are achieved, the correlation between the sensitive point and the emission source high value is analyzed by means of big data, an algorithm is combined, intelligent traceability of mass monitoring data is achieved, and real-time monitoring of the emission source is achieved. The whole system is convenient to install, accurate in monitoring and scientific in traceability.

Description

Technical Field

[0001] The present invention relates to the field of AI smart provenance data technology, and in particular to a method for processing AI smart provenance data. Background Art

[0002] In the field of environmental monitoring, when pollution occurs, tracing the source is required. Traditional methods generally use sampling and analysis to trace the source. However, due to the small number of data samples, the pollution source often does not emit pollutants at the sampling time, making it difficult to achieve the purpose of tracing the source. With the rapid development of the Internet of Things and big data, a large amount of monitoring data has also been obtained in environmental monitoring. How to use these big data for tracing the source is a new challenge in the field of environmental monitoring.

[0003] Taking the tracing of complaints about bad odors, which has been the focus of attention in the environmental field in recent years, as an example, this paper explores an AI intelligent tracing data processing method. Previous tracing methods are often carried out through sampling and analysis. For example, in the tracing of bad odors, the traditional idea is to sample and analyze the VOCs components of sensitive points, and compare them with the VOCs components emitted by surrounding enterprises, and trace the source based on the degree of component consistency. However, due to the complexity of VOCs, the number of components, and the low olfactory threshold, the monitoring range and monitoring accuracy requirements are relatively high. Factors such as short emission periods, rapid diffusion, and untimely sampling make it difficult to respond in a timely manner and collect samples.

[0004] Electronic noses are generally composed of multiple sensors testing different gases, and then using software fitting weights to mimic the output OU values ​​of human nose sensations, i.e. a combination of hardware + software. VOCProbe is a nano-metal oxide sensor that responds to TVOC, ammonia, and hydrogen sulfide. The cloud platform removes interference factors through AI algorithms and outputs OU values. Nano-sensors can be understood as human noses, reflecting whether there is a smell or not, rather than reflecting the concentration of each substance. Compared with the means of distinguishing component monitoring, it is more consistent with the feeling of the human nose. The characteristics of VOCProbe include: 1) Standardization: the broad-spectrum nano-metal oxide sensing technology, coupled with automatic calibration of humidity compensation and baseline calibration patented algorithms, has been successfully applied to more than 20 industries and has taken the lead in formulating group standards; 2) Complete set: It consists of three parts: ultra-small monitor, smart platform and mobile app; 3) Automation: The ultra-small monitor collects data and transmits it to the cloud platform through the Internet of Things NB-IoT technology, realizing automatic data collection, transmission, storage, calibration, alarm, playback, statistics, analysis and other automation functions; 4) Green: Ultra-low energy consumption design, average power consumption of 550uA, built-in battery power supply for more than one year, plug and play;

[0005] The principle of the diffusion of foul-smelling gases with wind direction is as follows: the diffusion of atmospheric pollutants depends largely on the distribution of the flow field. The diffusion plume varies with the change of wind direction. The main body of pollutants is transported in the direction of the dominant wind. The wind speed determines the speed of pollutant transportation. The flow field conditions in urban areas are very complex. Affected by buildings, topography and landforms, when airflow encounters obstacles such as buildings, it produces circumference, backflow, and narrow pipe effects around the buildings, causing significant changes in wind direction and wind speed, which also has a significant impact on the transportation of pollutants. Due to the obstruction of buildings, airflow bypasses buildings and forms a convergence of airflow between adjacent buildings, which increases the wind speed in these areas and produces the so-called narrow pipe effect; in other areas (the leeward side of buildings), the wind speed decreases or even calms. In the distribution of the flow field on the vertical section, the area higher than the buildings is less affected by the buildings, and the airflow exhibits a uniform laminar flow; near the ground, the buildings have a strong impact and the flow field becomes very complicated. Not only does the friction and other factors reduce the wind speed on the leeward side and the lower layers of the building, but also when the airflow bypasses the building from the top, the wind speed at the top of the building increases, forming a high-rise wind; when the wind direction, wind speed and the height and distance of adjacent buildings meet certain conditions, a vortex phenomenon will form between the buildings; therefore, in the small meteorological field within 10KM, atmospheric pollutants diffuse irregularly, and the settings of factors such as the specific gravity of the air, turbulence intensity, and thermal effects are different from the actual atmospheric environment, making it difficult to simulate accurately; traditional sampling and analysis methods rely on atmospheric transmission models to simulate tracing, which has great uncertainty; for this reason, the present application proposes a method for processing AI intelligent tracing data. Summary of the invention

[0006] Based on the technical problems existing in the background technology, the present invention proposes a method for processing AI intelligent traceability data.

[0007] The present invention proposes a method for processing AI intelligent traceability data, comprising the following steps:

[0008] S1: Monitoring point deployment: Based on the principle that odorous gases diffuse with the wind direction, VOCProbe monitoring points are deployed at sensitive points, transmission channels and pollution sources, and online wind direction and speed data are integrated to obtain detection data with fine monitoring granularity;

[0009] S2: Source tracing correlation matrix: Establish a monitoring high-value correlation matrix in the monitoring system window, find out the high value of sensitive points within the set time period according to the high-value threshold set by the monitoring standard, and establish a correlation matrix corresponding to the transmission channel and pollution source point related to the high value of the sensitive point, and use big data to find the target sensitive points and suspected pollution sources, and the corresponding time period;

[0010] S3: Normalization processing: normalizing the point monitoring data of the correlation matrix;

[0011] S4: DTW calculation similarity: The DTW algorithm is used to calculate the curve similarity between the sensitive point and the related transmission channel and pollution source point in the correlation matrix;

[0012] S5: Result output: Analyze the correlation between sensitive points and high values ​​of emission sources, achieve the purpose of tracing the source, and output the results.

[0013] Preferably, in S1, the binding device used for integrating online wind direction and wind speed data is a wind direction and anemometer.

[0014] Preferably, in S2, the time period is one hour, and considering that the high value may span the hour, in order to better calculate the curve similarity, the curves of one hour before and after the high value, that is, a total of three hours, are taken for calculation.

[0015] Preferably, in S4, the DTW algorithm calculates the similarity by calculating the DTW value of the time period corresponding to the high value of the sensitive point and each suspected pollution source point. The smaller the DTW value, the higher the curve similarity.

[0016] Preferably, in S2, the monitoring high value correlation matrix is ​​established according to the four-dimensional association of "monitoring high value time, high value value, wind direction and speed, and point geographical location", and the specific steps are:

[0017] S201: The user selects the time of the analysis project, the location of the sensitive point, target area and transmission channel through the monitoring system window, and sets the high threshold of the OU value of the sensitive point, target area and transmission channel;

[0018] S202: Filter out data with high values ​​of sensitive points in the database, and record the relevant time and points;

[0019] S203: querying the wind direction corresponding to each high value of the sensitive point, and querying the time corresponding to the high value and the target area and transmission channel in the upwind direction that are also high value points in the previous hour;

[0020] S204: Record and establish a traceability correlation matrix for each sensitive point.

[0021] Preferably, in S3, the specific steps of normalization processing are:

[0022] S301: Convert the number into a decimal between (0, 1);

[0023] S302: The dimensional expression is converted into a dimensionless expression, and a linear normalization method is used to normalize the data. The expression formula of the linear normalization method is:

[0024]

[0025] Here we assume that the value of “x” is 20, the maximum number is 55, and the minimum number is 5. To normalize this number, we start with the denominator, which is 50 (55-5). The same idea is used to calculate the numerator: x-min = 15 (20–5). The standardized x or x' is 15 / 50 = 0.3.

[0026] Preferably, in S4, the specific steps of the DTW algorithm are:

[0027] S401: Calculate the Euclidean distance between each time point of two time series;

[0028] S402: Select an appropriate stepping mode, calculate row by row or column by column to construct a cumulative cost matrix between two time series, and use the most commonly used stepping mode of the DTW algorithm to calculate the cumulative cost matrix. The calculation formula is:

[0029] M(i,j)=dis(pi,qi)+min(M(i-1,j),M(i,j-1),M(ij,j-1));

[0030] S403: Find a path with the smallest sum of cumulative cost matrix values ​​between the start point and the end point of the two time series alignment. This path is called the optimal alignment path between the two time series.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention deploys VOCProbe at emission sources, sensitive points, and transmission paths for online monitoring, thereby realizing full-time and space supervision of emission sources, transmission channels, and sensitive points, and fine-grained monitoring 24 hours a day x 30 seconds around the clock. It also uses big data to analyze the correlation between sensitive points and high values ​​of emission sources, and combines algorithms to realize intelligent tracing of massive monitoring data. The entire system is easy to install, accurate in monitoring, and scientific in tracing. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flowchart of a method for processing AI intelligent traceability data proposed in the present invention. DETAILED DESCRIPTION

[0034] The present invention will be further explained below in conjunction with specific embodiments.

[0035] Example

[0036] Reference Figure 1 , this embodiment proposes a method for processing AI intelligent traceability data, including the following steps:

[0037] S1: Monitoring point deployment: Based on the principle that odorous gases diffuse with wind direction, VOCProbe monitoring points are deployed at sensitive points, transmission channels and pollution sources, and online wind direction and wind speed data are integrated to obtain detection data with fine monitoring granularity. The binding device used for integrating online wind direction and wind speed data is a wind direction and anemometer.

[0038] S2: Source tracing correlation matrix: Establish a monitoring high-value correlation matrix in the monitoring system window, find out the high value of sensitive points within the set time period according to the high-value threshold set by the monitoring standard, and establish a correlation matrix corresponding to the transmission channel and pollution source point related to the high value of the sensitive point, and use big data to find out the target sensitive points and suspected pollution sources, and the corresponding time period; the time period is one hour, and considering that the high value may span the hour, in order to better calculate the curve similarity, take the curve one hour before and after the high value, that is, a total of three hours for calculation;

[0039] The monitoring high-value correlation matrix is ​​established based on the four-dimensional association of "monitoring high-value time, high-value value, wind direction and speed, and point geographical location". The specific steps are as follows:

[0040] S201: The user selects the time of the analysis project, the location of the sensitive point, target area and transmission channel through the monitoring system window, and sets the high threshold of the OU value of the sensitive point, target area and transmission channel;

[0041] S202: Filter out data with high values ​​of sensitive points in the database, and record the relevant time and points;

[0042] S203: querying the wind direction corresponding to each high value of the sensitive point, and querying the time corresponding to the high value and the target area and transmission channel in the upwind direction that are also high value points in the previous hour;

[0043] S204: Record and establish a traceability correlation matrix for each sensitive point;

[0044] S3: Normalization processing: normalizing the point monitoring data of the correlation matrix;

[0045] The specific steps of normalization are:

[0046] S301: convert the number into a decimal between (0, 1); this is mainly proposed for the convenience of data processing. Mapping the data to the range of 0 to 1 for processing is more convenient and faster, and should be classified as digital signal processing;

[0047] S302: The dimensional expression is converted into a dimensionless expression, and a linear normalization method is used to normalize the data. The expression formula of the linear normalization method is:

[0048]

[0049] Assume that the value of “x” is 20, the maximum number is 55, and the minimum number is 5. To normalize this number, start with the denominator, which is 50 (55-5). The same idea is used to calculate the numerator: x-min = 15 (20–5). The standardized x or x' is 15 / 50 = 0.3.

[0050] S4: DTW similarity calculation: The DTW algorithm is used to calculate the curve similarity between the sensitive point and the related transmission channel and pollution source point in the correlation matrix; the DTW algorithm calculates the similarity by calculating the DTW value of the time period corresponding to the high value of the sensitive point and each suspected pollution source point. The smaller the DTW value, the higher the curve similarity; the DTW algorithm is a commonly used and reliable similarity measurement method in sequence data mining. Compared with the traditional lock-step measurement method that aligns sequence points one-to-one, the DTW algorithm allows one-to-many alignment between sequence points and can well realize the similarity measurement of local features of time series. This measurement method can not only solve the problem that traditional measurement methods cannot be applied to unequal time series, but also effectively adapt to the characteristics of amplitude changes and phase shifts of time series data;

[0051] The specific steps of the DTW algorithm are:

[0052] S401: Calculate the Euclidean distance between each time point of two time series;

[0053] S402: Select an appropriate stepping mode, calculate row by row or column by column to construct a cumulative cost matrix between two time series, and use the most commonly used stepping mode of the DTW algorithm to calculate the cumulative cost matrix. The calculation formula is:

[0054] M(i,j)=dis(pi,qi)+min(M(i-1,j),M(i,j-1),M(ij,j-1));

[0055] S403: Find a path with the smallest sum of cumulative cost matrix values ​​between the start point and the end point of the two time series alignment. This path is called the best alignment path between the two time series.

[0056] S5: Result output: Analyze the correlation between sensitive points and high values ​​of emission sources, achieve the purpose of tracing the source, and output the results.

[0057] This embodiment deploys VOCProbe at emission sources, sensitive points, and transmission paths for online monitoring, thereby achieving full-time and space supervision of emission sources, transmission channels, and sensitive points, and fine-grained monitoring 24 hours a day x 30 seconds around the clock. It also uses big data to analyze the correlation between sensitive points and high values ​​of emission sources, and combines algorithms to achieve intelligent tracing of massive monitoring data. The entire system is easy to install, accurate in monitoring, and scientific in tracing.

[0058] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for processing AI intelligent traceability data, characterized in that: The following steps are involved: S1: Monitoring point deployment: Based on the principle that odorous gases diffuse with the wind direction, VOCProbe monitoring points are deployed at sensitive points, transmission channels and pollution sources, and online wind direction and speed data are integrated to obtain detection data with fine monitoring granularity; S2: Source tracing correlation matrix: Establish a monitoring high-value correlation matrix in the monitoring system window, find out the high value of sensitive points within the set time period according to the high-value threshold set by the monitoring standard, and establish a correlation matrix corresponding to the transmission channel and pollution source point related to the high value of the sensitive point, and use big data to find the target sensitive points and suspected pollution sources, and the corresponding time period; S3: Normalization processing: normalizing the point monitoring data of the correlation matrix; S4: DTW calculation similarity: The DTW algorithm is used to calculate the curve similarity between the sensitive point and the related transmission channel and pollution source point in the correlation matrix; S5: Result output: Analyze the correlation between sensitive points and high values ​​of emission sources, achieve the purpose of tracing the source, and output the results.

2. The method for processing AI intelligent traceability data according to claim 1, characterized in that: In S1, the binding device used for integrating online wind direction and wind speed data is a wind direction and anemometer.

3. The method for processing AI intelligent traceability data according to claim 1 is characterized in that: In S2, the time period is one hour, and considering that the high value may span the hour, in order to calculate the curve similarity, the curves of one hour before and after the high value, that is, a total of three hours, are taken for calculation.

4. The method for processing AI intelligent traceability data according to claim 1 is characterized in that: In S4, the DTW algorithm calculates the similarity by calculating the DTW value of the time period corresponding to the high value of the sensitive point and each suspected pollution source point. The smaller the DTW value, the higher the curve similarity.

5. The method for processing AI intelligent traceability data according to claim 1, characterized in that: In S2, the monitoring high value correlation matrix is ​​established according to the four-dimensional association of "monitoring high value time, high value value, wind direction and speed, and point geographical location", and the specific steps are: S201: The user selects the time of the analysis project, the location of the sensitive point, target area and transmission channel through the monitoring system window, and sets the high threshold of the OU value of the sensitive point, target area and transmission channel; S202: Filter out data with high values ​​of sensitive points in the database, and record the relevant time and points; S203: querying the wind direction corresponding to each high value of the sensitive point, and querying the time corresponding to the high value and the target area and transmission channel in the upwind direction that are also high value points in the previous hour; S204: Record and establish a traceability correlation matrix for each sensitive point.

6. The method for processing AI intelligent traceability data according to claim 1, characterized in that: In S3, the specific steps of normalization processing are: S301: Convert the number into a decimal between (0, 1); S302: The dimensional expression is converted into a dimensionless expression, and a linear normalization method is used to normalize the data. The expression formula of the linear normalization method is: Here we assume that the value of “x” is 20, the maximum number is 55, and the minimum number is 5. To normalize this number, we start with the denominator, which is 50 (55-5). The same idea is used to calculate the numerator: x-min = 15 (20–5). The standardized x or x' is 15 / 50 = 0.

3.

7. The method for processing AI intelligent traceability data according to claim 1, characterized in that: In S4, the specific steps of the DTW algorithm are: S401: Calculate the Euclidean distance between each time point of two time series; S402: Select an appropriate stepping mode, calculate row by row or column by column to construct a cumulative cost matrix between two time series, and use the stepping mode of the DTW algorithm to calculate the cumulative cost matrix. The calculation formula is: M(i,j)=dis(pi,qi)+min(M(i-1,j),M(i,j-1),M(ij,j-1)); S403: Find a path with the smallest sum of cumulative cost matrix values ​​between the start point and the end point of the two time series alignment. This path is called the optimal alignment path between the two time series.

Images