Peculiar smell intelligent database construction method based on big data analysis

By establishing an intelligent database of odors, using big data analysis technology to achieve the olfactory threshold and odor properties prediction of odorous pollutants, identifying odor-causing factors and pollution sources, solving the problem of the existing technology being unable to identify odor-causing substances and locking the source of pollution, and realizing the problem of the existing technology being unable to identify odor-causing substances and locking the source of pollution. Rapid analysis and management support for foul smell pollution.

CN119069040AActive Publication Date: 2024-12-03TIANJIN ACAD OF ECOLOGICAL & ENVIRONMENTAL SCI

Patent Information

Application Number
CN202411256420.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-09
Publication Date
2024-12-03
Estimated Expiration
2044-09-09

AI Technical Summary

Technical Problem

The existing technology cannot effectively identify the odor-causing substances in foul odor pollution and lock in the source of pollution, and cannot quickly analyze the contaminated components and evaluate the degree of pollution, resulting in difficulties in managing foul odor pollution.

Method used

Establish an intelligent odor database based on big data analysis, and realize the prediction of substance olfactory threshold and odor properties, identification of odor factors through process-based methods such as automated data collection, substance information data set, pollution characteristic analysis and visualization, source analysis and report generation, and realize the prediction of substance olfactory threshold and odor properties, identification of odor factors and Contamination traceability diagnosis.

Benefits of technology

It has achieved diversified odor odor evaluation of substance analysis, sensory evaluation and source traceability, which can quickly identify odor-causing substances and lock in pollution sources, and supports pollution management decisions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119069040A_ABST
    Figure CN119069040A_ABST
Patent Text Reader

Abstract

A method for constructing an intelligent odor database based on big data analysis belongs to the field of environmental protection and comprises the steps of establishing an automatic data acquisition workstation, constructing a material information data set, performing pollution characteristic analysis and visualization, analyzing sources, generating reports and the like. The material information in the material information data set comprises smell quality, smell threshold, smell durability, smell health factors, long-distance transmission grade, industry distribution and the like. By means of the method, odor-causing substances can be accurately recognized, pollution sources can be analyzed, and parks and enterprises are helped to effectively solve the odor problem.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of environmental protection, and more specifically relates to a method for constructing an intelligent odor database based on big data analysis. Background Art

[0002] Odor pollution is a typical nuisance pollution, which directly affects the quality of life and comfort of the people and is closely related to the fundamental interests of the people. According to the national ecological environment complaint statistics in the past five years, the pollution complaint rate remains high, second only to noise pollution, and has become "a prominent ecological and environmental problem that must be focused on to overcome around the people." Odor pollution is instantaneous and paroxysmal, and traditional offline testing technology can no longer meet the monitoring needs. In the past three years, on-site rapid monitoring technology has gradually developed, mainly including sensors, rapid mass spectrometry, spectroscopy, ion mobility spectrometry, etc. However, on-site monitoring results can only reflect the type of substance and its content, and cannot identify key odor-causing substances, lock the source of pollution, or evaluate the degree and scope of pollution caused to the surrounding environment. Therefore, how to quickly analyze the pollutant components, assess the degree of pollution, and diagnose odor on-site has become one of the current problems that need to be solved urgently.

[0003] Databases are an important means of monitoring, tracing and evaluating ecological environmental pollution. Domestic and foreign research institutions have developed software tools for viewing a variety of environmental air quality information data based on environmental pollution conditions, making it easier for people to understand environmental pollution conditions. For example, the U.S. Environmental Protection Agency has released the SPECIATE database on volatile organic compounds and particulate matter from air pollution sources, covering more than 8,000 types of material information, including material information, harmfulness and photochemical assessment information of each component; the multi-scale emission inventory model developed by Tsinghua University has constructed a high-resolution inventory of Chinese anthropogenic air pollutants and carbon dioxide emissions; the atmospheric pollution source spectrum database developed by Nankai University covers more than 3,000 source spectrum data of fixed combustion sources, process sources, mobile sources, biomass combustion sources, dust sources, etc. with different particle sizes in more than 40 cities across the country, providing basic emission data support for related scientific research, policy evaluation and air quality management.

[0004] The above databases use health risks, photochemical pollution, carbon emissions, and particulate matter as statistical indicators to assess the pollution status of specific indicators in countries and regions, and then guide pollution management. However, the application of databases in the field of odor pollution is still blank, which brings considerable obstacles to the management of odor pollution in my country. Therefore, it is very necessary to develop a database suitable for my country's odor pollution supervision needs. Summary of the invention

[0005] In order to solve the problem of unclear identification and unknown sources of odor-causing substances in odor pollution, the present invention establishes an intelligent database with a process of "substance analysis-sensory evaluation-source tracing", which has the functions of data statistics, simulation prediction and comparative analysis, and can realize the prediction of substance olfactory threshold and odor properties, accurate identification of odor-causing factors and rapid pollution tracing diagnosis.

[0006] In order to achieve the above object, the present invention is implemented by adopting the following technical scheme: the method comprises:

[0007] Establish an automated data collection workstation with data interface and transmission functions to collect odor concentration monitored by online sensors in real time and input monitoring data at the same time;

[0008] Construct a material information data set, including the odor characteristics, diffusion characteristics, health risks and industry distribution of the material, which is used for querying the odor properties of pollutants, analyzing pollution emission characteristics and calling basic data for screening key odor-causing substances;

[0009] Pollution feature analysis and visualization, including descriptive analysis and statistical analysis: Based on monitoring data, analyze the odor pollution emission characteristics of emission sources, enterprises, and regions in different scenarios, find the pollutants with the main concentration contribution and high detection rate, draw the source component spectrum, further screen the odor-causing substances through odor profile, odor activity, odor intensity and other characteristics and statistical analysis, and establish the source fingerprint spectrum;

[0010] Source analysis and report generation: Based on the spectrum comparison method, the similarity between the source fingerprint spectrum and the environment / sensitive area fingerprint spectrum is calculated, and the similarity between each spectrum is ranked to lock in the "suspected" pollution source; the monitoring data, pollution emission characteristic analysis results and source analysis results are summarized to generate an odor pollution analysis report, and conclusions are drawn on key pollutants, odor-causing factors, source locking, etc.

[0011] In one embodiment, the olfactory threshold is predicted by using a quantitative structure-activity relationship based on linear regression, and the odor quality is predicted by using a convolutional neural network.

[0012] Olfactory threshold prediction, the predictor variables X include molar refractive index, dipole / polarizability, hydrogen bond pH, number of atoms / chemical bonds, and generalized central atom hybridization type, and the predictor variable Y is the olfactory threshold;

[0013] Odor quality prediction, based on the three major theories of olfactory perception and the principles of aroma chemistry, using molecular shape, vibration frequency, and physical adsorption / chemical adsorption, and using convolutional neural networks to establish an odor quality prediction model;

[0014] Among them, the input layer information includes functional groups, the number of functional groups, molar refractive index, molar volume, dipole / polarizability, the number of covalent bonds, the number of hydrogen bond donors, and the number of hydrogen bond acceptors, and the output layer is the odor quality.

[0015] In one scheme, the odor persistence uses saturated vapor pressure, molar mass, olfactory threshold, and relative content as screening factors, determines the weight coefficient and assignment standard of each factor, calculates the comprehensive score of the substance and ranks it;

[0016] The top 20% of the odor persistence scores of the substances in the sample are defined as the “odor persistence factor” of the sample;

[0017] The weight coefficients of each factor are as follows:

[0018] y=0.3a+0.4b+0.2c+0.1d

[0019] Wherein, a represents saturated vapor pressure, b represents molecular weight, c represents olfactory threshold, and d represents relative content.

[0020] In one embodiment, the odor health factor refers to the ratio of the olfactory threshold of a substance to its health risk value, as follows:

[0021]

[0022] In the formula, I i ——i The smell health factor of the substance; OT i ——Odor threshold of substance i, mg / m 3 ;TWA i ——The cumulative toxic gas exposure limit value of substance i measured averagely within 4 to 16 hours, mg / m 3 ;

[0023] If the ratio is greater than 1, it means that the substance has a dominant impact on human health; if the ratio is less than 1, it means that the substance has a dominant impact on human olfactory stimulation.

[0024] In one scheme, the long-distance transmission level is evaluated by constructing the Boston Matrix based on the chemical reaction rate constant and the half-life, and is divided into 1 to 4 levels, with the higher the level, the stronger the transmission capability; level 1 means "low possibility of long-distance transmission", level 2 means "low possibility of long-distance transmission", level 3 means "possibility of certain long-distance transmission", and level 4 means "high possibility of long-distance transmission".

[0025] In one solution, the inferential statistical analysis analyzes the odor characteristics of the emission source / enterprise to form an odor profile map; using odor profile, odor activity analysis, least partial squares fitting, cluster analysis and correlation analysis, the correlation between the substance and its odor properties is found, the key odor-causing substances are locked, and an odor list is formed;

[0026] In one embodiment, the pollution source analysis includes:

[0027] (1) Establish a pollution fingerprint map. Take the production process or enterprise as the observation dimension. Based on the odor list and its concentration value, establish a pollution source fingerprint map with the substance type as the horizontal axis and the relative content of toluene as the vertical axis. Similarly, establish a receptor fingerprint map within the factory / factory boundary / sensitive area;

[0028] (2) Fingerprint comparison: The similarity between the fingerprints of the pollution source and the receptor was calculated using the angle cosine method of Matlab. Fingerprints with a similarity greater than or equal to 0.85 were considered "highly similar", fingerprints with a similarity between 0.70 and 0.85 were considered "relatively similar", and fingerprints with a similarity below 0.70 were considered "relatively similar".

[0029] In one plan, the report generation is divided into enterprise, park and industry categories. The enterprise category focuses on analyzing the characteristics of each emission source to find the characteristic odor pollutants of the emission source; the park category focuses on analyzing the typical odor substances of the enterprise and the pollution source analysis; the industry category focuses on analyzing and clarifying the pollution characteristics and key odor substances of the main processes.

[0030] Beneficial effects of the present invention:

[0031] The intelligent odor database developed by the present invention has the functions of data statistics, simulation prediction and comparative analysis. It can realize the simulation of olfactory threshold, odor property simulation, contribution and source analysis of odor substances, and realize the diversified odor evaluation of "material analysis-sensory evaluation-source tracing", providing technical support for solving unclear material analysis and unknown sources. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Attached Figure 1 The logical structure of the database of the present invention;

[0033] Attached Figure 2 This is a model diagram for predicting odor quality and olfactory threshold of the present invention;

[0034] Attached Figure 3 It is the long-distance transmission level model of the present invention;

[0035] Attached Figure 4 It is a schematic diagram of the change of odor concentration at a certain point of the present invention;

[0036] Attached Figure 5 It is a schematic diagram of the average total concentration of substances at all points in the enterprise of the present invention;

[0037] Attached Figure 6 This is a schematic diagram of average concentration contribution of different types of substances at various points of the present invention;

[0038] Attached Figure 7 It is the total concentration and material distribution of pollutants at a certain time / state of the present invention;

[0039] Attached Figure 8 The least partial squares quasi-analysis of the points and their odor attributes and the substance components and their concentrations of the present invention;

[0040] Attached Fig. 9 It is the cluster analysis of the point difference pollutants of the present invention;

[0041] Attached Fig.10 This is the substance-odor attribute correlation analysis of the present invention. DETAILED DESCRIPTION

[0042] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Typical embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in the present invention. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0043] Unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0044] The present invention provides an odor intelligent database based on big data analysis, which is used for rapid analysis of on-site pollutants, screening of key odor substances and source analysis. The database is mainly divided into four parts, namely, automated data acquisition workstation, material information data set, pollution feature analysis and visualization, source analysis and report generation.

[0045] like Figure 1 As shown in the figure, the functions and effects of each part are as follows:

[0046] 1. Automated data acquisition workstation

[0047] With data interface and transmission function, it can collect the odor concentration monitored by the sensor online in real time, and can also input the monitoring data. The input information includes:

[0048] (1) Company information: company name, company type, location, main products, registered capital, production process, raw materials and auxiliary materials;

[0049] (2) Sample information: company name, location name, location type, chimney information (height, temperature, diameter, exhaust volume, emission rate);

[0050] (3) Pollutant data: site name, sample odor, substance type, substance name, English name, substance CAS number, and substance concentration.

[0051] 2. Material Information Dataset

[0052] It is mainly used for the query of odor properties of pollutants, analysis of pollution emission characteristics and basic data call for the screening of key odor-causing substances. Substance information mainly includes five types, namely basic information of substances, odor properties, atmospheric diffusion, health risks and industry distribution. The components and their indicators are shown in Table 1.

[0053] Table 1 Summary of material basis data

[0054]

[0055] (1) Odor quality and olfactory threshold

[0056] Odor quality and olfactory threshold are important odor information and are often used to evaluate the odor of pollution. However, due to individual differences in smell, the odor quality and olfactory threshold measurement results of substances vary greatly. In addition, there are a large number of substances that lack quality and threshold information.

[0057] In order to improve the accuracy of odor-causing substance screening, the present invention uses a quantitative structure-activity relationship based on linear regression (see formula 1) to establish an olfactory threshold prediction model based on the known molecular structure properties and olfactory threshold of the substance. The prediction variable X includes molar refractive index, dipole / polarizability, hydrogen bond pH, number of atoms / chemical bonds, and generalized central atom hybridization type, and the prediction variable Y is the olfactory threshold.

[0058] ODT = c + β1ε + β2R + β3H A + β4H B + β5V (1)

[0059] Where: ε is the molar refractive index, unit (cm 3 / mol -1 ), R is dipole / polarizability, H A is the acidity of hydrogen bonds and; H B is the basicity and of hydrogen bonds; V is the atomic and chemical bond contribution data; β1~β5 are the regression coefficients of individual variables, which vary with the type of substance.

[0060] Similarly, for the prediction of odor quality, the present invention is mainly based on the three major theories of olfactory perception and the principles of aroma chemistry, starting from the molecular shape (structure), vibration frequency (chemical bond properties) and physical adsorption / chemical adsorption, and using a convolutional neural network to establish an odor quality prediction model. Among them, the input layer information includes functional groups and their number, molar refractive index, molar volume, dipole / polarizability, number of covalent bonds, number of hydrogen bond donors, number of hydrogen bond acceptors, and the output layer is the odor quality. The known odor qualities and related variables in the material information data set are divided into a test set and a training set. The test set is used for model construction, and the training set is used for model generalization evaluation.

[0061] The correlation coefficient r, root mean square error (RMSE), mean absolute percent error (MAPE), loss function and model complexity were selected to evaluate the model and to assess issues such as the difference between the true value and the predicted value and overfitting of the training data.

[0062] The model is optimized by selecting hyperparameter tuning, modifying the network structure, increasing or decreasing the number of hidden layers, adjusting the activation function, etc., and the uncertainty of the prediction results of the olfactory threshold is estimated by combining the independent variables. The final model structure is as follows Figure 2 shown.

[0063] (2) Odor persistence

[0064] Generally, the persistence of odor is related to the molar mass, volatility and content of the substance. The present invention uses fuzzy mathematical analysis method, takes saturated vapor pressure, molar mass, olfactory threshold and relative content as screening factors, determines the weight coefficient (see formula 1) and the assignment standard (see table 2) of each factor, calculates the comprehensive score of the substance and ranks it. The top 20% of the odor persistence scores of the substances in the sample are defined as the "odor persistence factor" of the sample.

[0065] y=0.3a+0.4b+0.2c+0.1d (2)

[0066] Wherein, a represents saturated vapor pressure, b represents molecular weight, c represents olfactory threshold, and d represents relative content.

[0067] Table 2 Screening factor assignment criteria

[0068]

[0069] (3) Odor health factor

[0070] The odor health factor refers to the ratio of the olfactory threshold of a substance to its health risk value (see formula 3). If the ratio is greater than 1, it means that the substance has a dominant impact on human health; if the ratio is less than 1, it means that the substance has a dominant olfactory stimulation to humans. The odor health factor combines sensory impact and health status to further evaluate the relationship between the odor of a substance and its contribution to health.

[0071]

[0072] In the formula, I i ——i The odor health factor of the substance;

[0073] OT i ——Odor threshold of substance i, mg / m 3 ;

[0074] TWA i ——The cumulative toxic gas exposure limit value of substance i measured averagely within 4 to 16 hours, mg / m 3 .

[0075] (4) Long-distance transmission level

[0076] like Figure 3 As shown, the diffusion of pollutants in the atmosphere is mainly related to meteorological conditions and the physical and chemical properties of the substance. The present invention considers the substance from the perspective of the substance, selects the chemical reaction rate constant and half-life of the substance as evaluation indicators, selects the average value of the indicators of all pollutants in the tested sample as the evaluation threshold, and uses the Boston matrix model to evaluate the long-distance transmission level of the substance. There are 4 levels in total, and the higher the level, the stronger the transmission capacity. Level 1 means "low possibility of long-distance transmission", level 2 means "low possibility of long-distance transmission", level 3 means "possibility of certain long-distance transmission", and level 4 means "high possibility of long-distance transmission".

[0077] (5) Industry distribution

[0078] Industry distribution indicators include main uses, industry sources and typical industry odor basic words. Industry sources are based on the pollution emission data (including substances and their concentrations) of 45 industry emission links accumulated in the early stage, and the main uses are obtained by consulting relevant domestic and foreign standards and production and emission data. The industry basic word library is shown in Table 3.

[0079] Table 3 Typical industry odor basic words

[0080]

[0081]

[0082]

[0083] 3. Pollution feature analysis and visualization

[0084] (1) Descriptive analysis

[0085] like Figure 4-Figure 7 As shown in the figure, descriptive analysis is used to analyze the changes in odor concentration and substance concentration distribution of emission sources, enterprises, and regions in different scenarios, sort out the main concentration contributing substances and their odor information, and display the pollution emission characteristics in multiple dimensions.

[0086] ① Overall characteristics analysis: Based on the field test results, record and draw pictures.

[0087] a) Changes in odor concentration at a certain point in the time dimension (line graph);

[0088] b) Comparison of single value / average value of total concentration of substances at monitoring points over a period of time (bar graph);

[0089] c) Average concentration contribution of different types of substances at monitoring points over a period of time (percentage stacked bar chart).

[0090] ②Analysis of characteristics of pollution production and discharge links:

[0091] a) Total pollutant concentration and material distribution corresponding to abnormal odor concentration values ​​(stacked bar graph);

[0092] b) Total pollutant concentration and material distribution at different times and conditions (stacked bar graphs), including daytime and nighttime, before and after process improvements, etc.;

[0093] c) The top 10 pollutants with the highest average concentration in a pollution production and discharge link over a period of time, as well as their basic information, odor information and concentration information.

[0094] (2) Inferential analysis

[0095] Inferential statistical analysis is used to analyze the odor characteristics of emission sources / enterprises and form an odor profile map; odor activity analysis, least partial squares fitting, cluster analysis and correlation analysis are used to find the correlation between substances and their odor properties, identify key odor-causing substances, and form an odor list.

[0096] ①Odor profile map

[0097] Odor profile analysis originates from the field of food flavor, that is, odorists use industry-defined odor vocabulary to conduct sensory evaluation of a certain type of food, and the evaluation results are represented by odor profiles to distinguish the flavor characteristics of foods from different origins, brands, etc. However, there is no relevant research in the field of atmospheric environment, so more stringent quality control is needed to ensure the consistency of odor descriptions in industries / enterprises / waste discharge links. The specific steps for determining odor profiles in the environmental field are as follows:

[0098] a) The olfactory identification team shall consist of 4 persons who meet the requirements of ISO 8586:2012 and the "Three-point comparison odor bag method for the determination of odor in ambient air and exhaust gas" (HJ 1262-2022);

[0099] b) The odor identification team preliminarily determines the odor attributes and intensity of the industry / enterprise / waste discharge link through on-site olfactory analysis and discussion of odor attributes. Generally, odors mostly come from unorganized pollution;

[0100] c) In the laboratory, the olfactory identification team smells the tissue samples, supplements the odor attributes, clarifies the odor attributes of the industry / enterprise / waste discharge link, and evaluates the intensity level of each odor attribute 3 times, taking the average, and finally forming the odor attribute and intensity of the industry / enterprise / waste discharge link. The odor intensity is divided into 0 to 5 levels, "0" means no odor, "1" means a barely perceptible odor, "2" means a faint odor but the odor can be determined, "3" means a clearly perceptible odor, "4" means a strong odor, and "5" means a very strong and unbearable odor;

[0101] d) Based on the above odor attributes and their intensity analysis results, draw an odor profile map of the industry / enterprise / waste discharge link.

[0102] ②Odor activity analysis

[0103] The odor activity value of a substance refers to the ratio of the substance concentration to the olfactory threshold. The substance concentration is derived from the average value of the on-site test at a certain point, and the olfactory threshold is derived from the substance information set. The Chinese and English names, CAS numbers, odor activity values, and odor health factors of the top 20 substances with the odor activity values ​​at a certain point are presented in a list.

[0104] ③ Least partial squares fitting, cluster analysis and correlation analysis

[0105] like Figure 8-Figure 10 As shown in the figure, the least partial squares fitting is used to evaluate the relationship between the odor characteristics of the industry / enterprise / waste discharge link and the odor compounds and their odor attributes. The data comes from the odor attributes of the odor profile map and the material components and their concentration averages measured on site. By observing the distribution of point odor attributes and odor substances, the odor attributes and material types of the emission source / enterprise are locked in; cluster analysis is used to evaluate the differences in odor compounds of emission sources / enterprises and find key odor substances. The data comes from the material components and their concentrations measured on site; correlation analysis is used to evaluate the correlation between material odor characteristics and odor attributes of emission sources / enterprises. The data comes from the point material components and their concentrations measured on site and the odor attributes of the odor profile of the emission source / enterprise.

[0106] Correlation analysis uses the Pearson correlation coefficient method to measure the linear relationship between two continuous variables. The mathematical formula is as follows:

[0107]

[0108] Among them, xi and yi are the specific values ​​of the two variables, mean(X) and mean(Y) are their means, sd represents the standard deviation, and n is the number of samples.

[0109] The value of the correlation coefficient is between -1 and 1, where -1 indicates a completely negative correlation, 1 indicates a completely positive correlation, and 0 indicates no correlation. After calculating the correlation coefficient, the significance of the correlation can be determined by a significance test (such as a t-test). Based on this, the correlation between the odor characteristics of the substance and the odor properties of the emission source / enterprise can be analyzed.

[0110] Finally, take the union of all the above analysis results and combine them with the substance information set to list the odor substances and their properties in the industry / enterprise / waste discharge link.

[0111] 4. Source analysis and report generation

[0112] (1) Analysis of pollution sources

[0113] ① Establishment of pollution fingerprint

[0114] Taking the production process or enterprise as the observation dimension, based on the odor list and its concentration values, a pollution source fingerprint map is established with the substance type as the horizontal axis and the relative content of toluene as the vertical axis; similarly, a receptor fingerprint map is established within the factory / factory boundary / sensitive area.

[0115] ② Fingerprint comparison

[0116] The similarity between the fingerprints of the pollution source and the receptor was calculated using Matlab's angle cosine method. It was stipulated that the fingerprints with a similarity greater than or equal to 0.85 had a "high similarity", the fingerprints with a similarity between 0.70 and 0.85 had a "higher similarity", and the fingerprints with a similarity below 0.70 had a "lower similarity".

[0117] (2) Report Generation

[0118] The report is divided into enterprise, park and industry categories. Among them, the enterprise category focuses on the pollution characteristics and odor analysis of the sewage discharge points, mainly including the pollution emission characteristics of the sewage discharge points, key odor-causing substances and their odor characteristics; the park category mainly evaluates the park situation based on the enterprise, mainly including the enterprise odor profile map, the enterprise-dimensional odor fingerprint map and key odor-causing substances; the industry category mainly generates reports containing information such as the industry odor profile map, the production process odor fingerprint map and key odor-causing substances.

[0119] Embodiment 2:

[0120] Plastic enterprises that produce resins were selected as the analysis objects. Through "2 days and 3 frequencies" sampling analysis, 6 sets of material data and odor concentration data at 4 points were obtained. Using the automated data acquisition workstation, odor concentration monitoring data at 4 points were obtained, and 6 sets of material data at 4 points were imported into the workstation. The odor concentration ranged from 61 to 1308, and the total material concentration was 3.756 to 12.356 mg / m 3 , the types of substances include alcohols, nitrides, aromatic compounds and furans. The odor characteristics of enterprises include rancid smell, mellow fragrance, chemical reagent smell, paint smell, alcohol smell and fishy smell. Through database screening, the odor-causing substances at the four points include ethanol, trimethylamine, ammonia, styrene, ethyl acetate, ethyl butyrate, 2-butanone, furan, butanol and acetic acid. Based on odor activity analysis, least partial squares fitting, cluster analysis and correlation analysis, it is determined that the main odor-causing substances at point 1 are ammonia and ethanol, ethyl acetate, ethyl butyrate, 2-butanone and styrene at point 2, trimethylamine and ammonia at point 3, and furan, butanol and acetic acid at point 4.

[0121] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0122] It should be understood that the detailed description of the technical solutions of the present invention by means of the preferred embodiments is illustrative rather than restrictive. A person skilled in the art may modify the technical solutions described in the embodiments, or replace some of the technical features by equivalents, based on reading the specification of the present invention; and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing an intelligent odor database based on big data analysis, characterized in that: The method includes: Establish an automated data collection workstation with data interface and transmission functions to collect odor concentration monitored by online sensors in real time and input monitoring data at the same time; Construct a material information data set, including the odor characteristics, diffusion characteristics, health risks and industry distribution of the material, which is used for querying the odor properties of pollutants, analyzing pollution emission characteristics and calling basic data for screening key odor-causing substances; Pollution characteristic analysis and visualization, including descriptive analysis and statistical analysis: Based on monitoring data, analyze the emission characteristics of odor pollution in different scenarios by emission sources, enterprises, and regions, screen odor-causing substances through odor profile, odor activity, odor intensity characteristics and statistical analysis, and establish source fingerprint spectrum; Source analysis and report generation: Based on the spectrum comparison method, the similarity between the source fingerprint spectrum and the environment / sensitive area fingerprint spectrum is calculated, and the similarity between each spectrum is ranked to identify the source of pollution; the monitoring data, pollution emission characteristic analysis results and source analysis results are summarized to generate an odor pollution analysis report.

2. The method for constructing an intelligent odor database based on big data analysis according to claim 1, characterized in that: The material information includes: odor quality and olfactory threshold, odor persistence, odor health factor, long-distance transmission level, and industry distribution.

3. The method for constructing an intelligent odor database based on big data analysis according to claim 2, characterized in that: The olfactory threshold is based on a quantitative structure-activity relationship of linear regression to establish an olfactory threshold prediction model, and the odor quality is predicted by a convolutional neural network. Olfactory threshold prediction, the predictor variables X include molar refractive index, dipole / polarizability, hydrogen bond pH, number of atoms / chemical bonds, and generalized central atom hybridization type, and the predictor variable Y is the olfactory threshold; Odor quality prediction, using molecular shape, vibration frequency, and physical adsorption / chemical adsorption, and using convolutional neural networks to establish an odor quality prediction model; Among them, the input layer information includes functional groups, the number of functional groups, molar refractive index, molar volume, dipole / polarizability, the number of covalent bonds, the number of hydrogen bond donors, and the number of hydrogen bond acceptors, and the output layer is the odor quality.

4. The method for constructing an intelligent odor database based on big data analysis according to claim 2, characterized in that: The odor persistence uses saturated vapor pressure, molar mass, olfactory threshold, and relative content as screening factors, determines the weight coefficient and value assignment standard of each factor, calculates the comprehensive score of the substance and ranks it; The top 20% of the smell persistence scores of the substances in the sample are defined as the "smell persistence factor" of the sample; The weight coefficients of each factor are as follows: y=0.3a+0.4b+0.2c+0.1d Wherein, a represents saturated vapor pressure, b represents molecular weight, c represents olfactory threshold, and d represents relative content.

5. The method for constructing an intelligent odor database based on big data analysis according to claim 2, characterized in that: The odor health factor refers to the ratio of the olfactory threshold of a substance to its health risk value, as follows: In the formula, I i ——i The smell health factor of the substance; OT i ——Odor threshold of substance i, mg / m 3 ;TWA i ——The cumulative toxic gas exposure limit value of substance i measured averagely within 4 to 16 hours, mg / m 3 ; If the ratio is greater than 1, it means that the substance has a dominant impact on human health; if the ratio is less than 1, it means that the substance has a dominant impact on human olfactory stimulation.

6. The method for constructing an intelligent odor database based on big data analysis according to claim 2, characterized in that: The long-distance transmission level described above is a Boston matrix constructed based on chemical reaction rate constants and half-life to evaluate the long-distance transmission level of a substance, which is divided into 1 to 4 levels. The higher the level, the stronger the transmission capability. Level 1 means "low possibility of long-distance transmission", level 2 means "low possibility of long-distance transmission", level 3 means "possibility of certain long-distance transmission", and level 4 means "high possibility of long-distance transmission".

7. The method for constructing an intelligent odor database based on big data analysis according to claim 1, characterized in that: The source analysis includes: (1) Establish a pollution fingerprint map. Based on the odor list and its concentration values, establish a pollution source fingerprint map with the substance type as the horizontal axis and the relative content of toluene as the vertical axis; similarly, establish a receptor fingerprint map within the factory / factory boundary / sensitive area; (2) Fingerprint comparison: The similarity between the fingerprints of the pollution source and the receptor was calculated using the angle cosine method of Matlab. Fingerprints with a similarity greater than or equal to 0.85 were considered "highly similar", fingerprints with a similarity between 0.70 and 0.85 were considered "relatively similar", and fingerprints with a similarity below 0.70 were considered "relatively similar".

8. The method for constructing an intelligent odor database based on big data analysis according to claim 1, characterized in that: The report generation is divided into enterprise, park and industry categories. The enterprise category focuses on analyzing the characteristics of each emission source and finding the characteristic odor pollutants of the emission source; the park category focuses on analyzing the typical odor substances of the enterprise and the pollution source analysis; The industry category focuses on analyzing and clarifying the pollution characteristics and key odor substances of the main processes.

Citation Information

Patent Citations

  • Method for analyzing and identifying key aromatic substances in tobacco leaf

    CN102680627A

  • Atmosphere fetor pollution fine source apportionment method

    CN103065198A

  • Odor pollution source key odor causing substance recognition method

    CN105469224A

  • Stink fingerprint spectrum based stink pollution source tracing method and application thereof

    CN108760912A

  • Method for tracing source of odor in rail transit vehicle

    CN110836941A

Cited By

  • Tracing method and device

    CN119963222A