CN115600643A - Method and system for rapidly predicting toxic gas - Google Patents
Method and system for rapidly predicting toxic gas Download PDFInfo
- Publication number
- CN115600643A CN115600643A CN202211267917.9A CN202211267917A CN115600643A CN 115600643 A CN115600643 A CN 115600643A CN 202211267917 A CN202211267917 A CN 202211267917A CN 115600643 A CN115600643 A CN 115600643A
- Authority
- CN
- China
- Prior art keywords
- space
- working condition
- full
- detector
- cfd
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses a method for quickly predicting toxic gas, which comprises the following steps: constructing a CFD data set of a full-scene working condition based on the actual working condition, intensively setting sampling points in the whole simulation space, and setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition; based on the characteristic data in the constructed CFD data set, a time-space attention mechanism is combined with other deep learning models to output a prediction model of time and space distribution concentration of the toxic gas; and correcting and calibrating the model based on real data of the field detector in actual application. The invention also discloses a system for rapidly predicting toxic and harmful gases. The invention can quickly and accurately predict the spatial-temporal distribution concentration of the toxic gas because of constructing a CFD data set of comprehensive working conditions, introducing a deep learning model of a spatial-temporal attention mechanism and calibrating by using real data of a field detector.
Description
Technical Field
The invention belongs to the field of gas monitoring, and particularly relates to a method and a system for quickly predicting toxic and harmful gas.
Background
Toxic gases generated in industrial production, such as hydrogen fluoride, chlorine, methane, sulfur dioxide and the like, once leaked, threaten the personal safety and environmental safety of factories and surrounding areas. The current research method for toxic gas leakage diffusion rules mainly comprises the following steps:
1. based on various gas diffusion models, such as a gaussian model, a BM model, a box model, and the like, the accuracy of these models is poor, and the accuracy under various complex meteorological environmental conditions and actual earth surface environments (such as various buildings distributed in a leakage area) is not as expected. For example, the Gaussian diffusion model is the most widely used model in atmospheric diffusion prediction, and the model is simple in calculation and easy to understand; the method approximately reflects the change process of the gas concentration, but is mainly suitable for open space, is influenced by surrounding buildings and wind speed and direction, and has larger prediction error in a high-concentration interval.
2. Based on a Computational Fluid Dynamics (CFD) model: basic equation based on fluid mechanics-N-S equation (Navier-Stokes equation); because of huge computation, in order to give consideration to time efficiency and accuracy, large vortex simulation is often used to filter an N-S equation, only large-scale turbulence is calculated, and a model is used to calculate turbulence smaller than the filtering scale. The method has better accuracy; however, although the calculation is simplified by using the large vortex simulation, the calculation amount is still huge, the consumed time is too long, the total space prediction often needs dozens of hours or even longer time, and the high real-time requirement of real-time early warning and emergency evacuation command rescue cannot be met at all.
3. Based on deep learning: at present, various training data based on a deep learning method are based on a plurality of public data sets or detector data of a small number of limited space coordinate points, and a model result is only real-time gas concentration prediction aiming at the small number of coordinate points. Lack of prediction of the various spatial locations of the entire gas leak diffusion region; or the gas concentration at other coordinate points in a wide space is predicted by interpolation based on the gas concentration prediction results of a small number of coordinate points, the accuracy is not in accordance with the expectation, and the prediction results cannot be convinced.
Reference 1 (patent application No. 201410005218.6) discloses a leak gas diffusion prediction method for three-dimensional space. The method comprises the steps of constructing a three-dimensional space coordinate system, representing a leakage area and a peripheral area in the same coordinate system, loading terrain and building distribution data of the leakage area and the peripheral area in a fixed area calculation area in the coordinate system to carry out leakage gas diffusion simulation, and carrying out toxic gas diffusion prediction outside the fixed area calculation area on the basis of the simulation, so that the leakage gas diffusion concentration distribution prediction of the peripheral area of a leakage source is realized. The invention fully considers the geographical information of the current time and the local place when the diffusion prediction is carried out on the leakage gas, and can quickly calculate the range of the polluted area and the concentration of the toxic gas in the polluted area caused by one or more leakage sources. Reference 1 uses the CFD method, which directly uses the basic equation N-S equation (navier-stokes equation) of the fluid mechanics, i.e., the fluid mechanics equation is calculated and solved by Direct Numerical Simulation (DNS); the method has huge calculation amount and very time-consuming full-space prediction.
Reference 2 (patent application No. 201610320272.9) a method for evaluating the security of people evacuation in case of a toxic gas leakage accident, comprising: simulating by adopting a leakage diffusion model to obtain the spatial concentration distribution condition of the toxic gas time sequence, namely a concentration field; simulating the evacuation process of people along a set evacuation path, and acquiring the spatial positions of evacuation individuals at different times; calculating the toxic load Pc of the evacuated individuals on the set evacuation path; calculating the death probability Pr of the evacuated individual according to the toxic load Pc and the type of the toxic gas; and obtaining the death rate Pd of the evacuated individuals according to the death probability Pr, and then obtaining the death rate P of the evacuated individuals according to the death rate P. Reference 2 performs two simulations, i.e., a toxic gas diffusion simulation and a person movement trajectory simulation. And calculating the amount of the toxic gas inhaled in the escape process of the personnel and the death probability by combining the concentration distribution of the toxic gas and the action track of the personnel, and taking the calculated amount of the toxic gas and the death probability as a safety evaluation basis. The toxic gas concentration distribution adopts a gas diffusion model method, has low precision, approximately reflects the change process of the gas concentration, is mainly suitable for open space, is influenced by surrounding buildings and wind speed and direction, and has larger prediction error in a high-concentration interval.
Disclosure of Invention
The technical problem to be solved by the invention is how to meet the requirements of real-time performance and accuracy of toxic gas prediction.
The invention solves the technical problems through the following technical means: a method for rapidly predicting toxic gas, comprising the following steps:
the method comprises the following steps: constructing a CFD data set of a full-scene working condition based on the actual working condition, intensively setting sampling points in the whole simulation space, and setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition;
step two: based on the characteristic data in the constructed CFD data set, a prediction model of time and space distribution concentration of the toxic gas is output by combining a space-time attention mechanism with other deep learning models;
step three: and correcting and calibrating the model based on real data of the field detector in actual application.
As the optimized technical scheme of the invention, the actual working condition in the step one specifically comprises a gas type, a gas leakage condition, a weather element and a surrounding building position coordinate, the gas leakage condition comprises a gas leakage source position and a gas leakage rate, and the weather element comprises a temperature, a humidity, an air pressure, a wind speed, a wind direction and an atmospheric stability.
As a technical solution for optimization in the present invention, the constructing a CFD dataset of a full-scene operating condition in the first step specifically includes:
let the number of features of a working condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
Seed working conditions;
densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting a sampling point every delta distance in 3 directions of the length, width and height, and setting d detectors at each sampling point, so that each working condition has
The detector and the CFD data set of the whole comprehensive working condition comprise
A detector; is provided withThe CFD simulation average time consumption of each working condition is t, and the total time consumption for constructing the CFD data set is t
And (4) obtaining the data of the full-scale detector at all the sampling points in the space under the full-scene working condition after CFD simulation.
As the optimization technical scheme of the invention, the space-time attention mechanism adopted in the second step is combined with an LSTM model, a GRU model or a TCN model.
As a technical solution for the optimization of the present invention, in the second step, the space-time attention mechanism learns weight distribution from the features, and then applies the weight distribution to the original features, so as to change the distribution of the original features, specifically including a space attention mechanism and a time attention mechanism, where the space attention mechanism is used to obtain spatial correlations of different spatial coordinate points, and the time attention mechanism is used to obtain time correlations of different times.
The invention also provides a rapid toxic gas prediction system, which comprises a characteristic acquisition and construction module, a data analysis module and a data correction module;
the system comprises a characteristic acquisition and construction module, a data acquisition and analysis module and a data analysis and analysis module, wherein the characteristic acquisition and construction module is used for acquiring all characteristic parameters in actual working conditions, constructing a CFD data set according to all the acquired characteristic parameters, densely setting sampling points in the whole simulation space, and setting a full detector at the sampling points, wherein the full detector is used for detecting all the characteristic parameters in the actual working conditions;
the data analysis module is used for outputting a prediction model of time and space distribution concentration of the toxic gas by combining a space-time attention mechanism with other deep learning models according to the CFD data set;
and the data correction module is used for correcting and calibrating the prediction model according to all the characteristic parameters actually acquired on site.
As an optimized technical scheme of the invention, in the characteristic acquisition and construction module, the actual working condition specifically comprises gas types, gas leakage conditions, weather elements and surrounding building position coordinates, the gas leakage conditions comprise gas leakage source positions and gas leakage rates, and the weather elements comprise air temperature, humidity, air pressure, wind speed, wind direction and atmospheric stability.
As an optimized technical solution of the present invention, in the feature acquisition and construction module, constructing a CFD data set of a full-scene operating condition specifically includes:
let the number of features of a working condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
Seed working conditions;
densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting a sampling point every delta distance in 3 directions of the length, width and height, and setting d detectors at each sampling point, so that each working condition has
A detector, a CFD data set of the whole comprehensive working condition has
A detector; assuming the CFD simulation average time consumption of each working condition is t, the total time consumption for constructing the CFD data set is t
And (4) obtaining data of the full-quantity detector at all the sampling points in the space under the full scene working condition after CFD simulation.
As the technical scheme of the optimization of the invention, the data analysis module is combined with an LSTM model, a GRU model or a TCN model when a space-time attention mechanism is adopted.
In the data analysis module, a space-time attention mechanism learns weight distribution from the features, and then the weight distribution is applied to the original features, so as to change the distribution of the original features.
The invention has the advantages that:
compared with the reference 1 in the background art, the method only uses the CFD method to construct the data set, and the prediction is based on the deep learning model obtained by training the CFD data set, so that the space-time concentration distribution of the toxic gas can be accurately and quickly predicted.
Compared with the reference 2 of the background art, the method does not consider the movement of people, but directly predicts the space-time concentration value of the whole space range; in addition, the method only calculates the concentration of the toxic gas without further safety evaluation, and the reference document 2 uses a gas diffusion model and does not use a deep learning model in the implementation of concentration prediction.
The invention can quickly and accurately predict the time-space distribution concentration of the toxic gas because of constructing a CFD data set of comprehensive working conditions, introducing a deep learning model of a time-space attention mechanism and calibrating by using real data of a field detector.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention and do not constitute a limitation of the invention. In the drawings:
FIG. 1 is a schematic flow chart of a method for rapidly predicting toxic gases according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a spatiotemporal attention mechanism of an embodiment of the present invention;
FIG. 3 is a schematic diagram of a gas concentration time profile of an embodiment of the present invention.
FIG. 4 is a schematic illustration of the spatial distribution of gas concentration in accordance with an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
Example 1
As shown in fig. 1, the present embodiment discloses a method for rapidly predicting toxic gases, which includes the following steps:
the method comprises the following steps: constructing a CFD data set of the full-scene working condition based on the actual working condition;
the actual working condition specifically comprises: the system comprises a gas type, a gas leakage condition, weather elements and surrounding building position coordinates, wherein the gas leakage condition comprises a gas leakage source position and a gas leakage rate, and the weather elements comprise temperature, humidity, air pressure, air speed, air direction and atmospheric stability;
the method for constructing the CFD data set under the full-scene working condition specifically comprises the following steps:
let the number of features of the operating condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
And (4) carrying out various working conditions.
Densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting a sampling point every delta distance in 3 directions of the length, width and height, and setting d detectors at each sampling point, so that each working condition has
A detector, a CFD data set of the whole comprehensive working condition has
A detector; assuming that the CFD simulation average time consumption of each working condition is t, the total time consumption for constructing the CFD data set is t
The CFD simulation construction data is a mature technology, the accuracy of simulated gas concentration is guaranteed, but the difficulty of constructing the data set is that the number of the full-scene working conditions is large, and the simulation of each working condition is time-consuming, so that the construction of the CFD data set under the full-scene working conditions is extremely time-consuming and labor-consuming.
Step two: based on the characteristic data in the constructed CFD data set, a space-time attention mechanism is combined with an LSTM (Long Short-Term Memory) model, a GRU (Gated Recirculation Unit) model or a TCN (Temporal convolutional network) model to output a prediction model of the time and space distribution concentration of the toxic gas;
the space-time attention mechanism learns weight distribution from the features, and then the weight distribution is acted on the original features, so that the distribution of the original features is changed, and the functions of increasing effective features and inhibiting ineffective features are achieved;
the space-time attention mechanism and the TCN model are combined as an example, and the structure diagram is shown in FIG. 2: the output result of the convolutional layer passes through a time attention module, a space attention module after convolution and then convolution to obtain a weighted output result;
step three: based on real data of the field detector in practical application, the model is corrected and calibrated, and accuracy and universality are further improved.
Inputting coordinate point (x) i ,y i ,z i ) A time density profile at that point may be output as shown in fig. 3; input time t j The gas at that time can be obtainedA spatial distribution map of concentration; the invention can quickly and accurately predict the time-space distribution concentration of the toxic gas because of constructing a CFD data set of comprehensive working conditions, introducing a deep learning model of a time-space attention mechanism and calibrating by using real data of a field detector.
At present, various training data based on a deep learning method are all based on public data sets or detector data of a plurality of small limited space coordinate point positions, and a model result is only real-time gas concentration prediction aiming at the small coordinate points. Lack of prediction of the various spatial locations of the entire gas leak diffusion region; or the gas concentration of other coordinate points in a wide space is predicted by interpolation based on the gas concentration prediction results of a small number of coordinate points, the accuracy is not in line with expectation, and the prediction results cannot be convincing.
The difficulty of the invention lies in how to fully utilize the constructed CFD data set to predict the toxic gas space-time concentration in the whole space range. The prediction results of the existing literature on a plurality of small coordinate point positions are continuous in time and discrete in space; the prediction results of the full space are continuous in time and space. The prediction of the total space is carried out by training a neural network by using the concentration value output by the CFD detector of the sufficiently dense coordinate points in the space, and then predicting the total space range. Because a large amount of coordinate point data which is dense enough needs to be input, the feature dimension is increased, and a reasonable deep learning method needs to be selected. Considering that gas diffusion has a front-back dependency in the time dimension, there is a concentration correlation of adjacent locations in the space dimension. Therefore, a deep learning model adopting a space-time attention mechanism is combined with a TCN (tool control network) or LSTM or GRU (generalized regression unit) method. In addition, the accuracy and the reliability of the model can be further improved by calibrating the real data of a plurality of field sensors.
Example 2
The embodiment provides a toxic gas rapid prediction system which comprises a feature acquisition and construction module, a data analysis module and a data correction module.
And the characteristic acquisition and construction module is used for acquiring all characteristic parameters in actual working conditions and constructing a CFD data set according to all the acquired characteristic parameters.
The actual conditions specifically include: the system comprises a gas type, a gas leakage condition, weather factors and surrounding building position coordinates, wherein the gas leakage condition comprises a gas leakage source position and a gas leakage rate, and the weather factors comprise temperature, humidity, air pressure, wind speed, wind direction and atmospheric stability;
the method for constructing the CFD data set under the full-scene working condition specifically comprises the following steps:
let the number of features of a working condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
And (4) carrying out various working conditions.
Densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting a sampling point every delta distance in 3 directions of the length, width and height, and setting d detectors at each sampling point, so that each working condition has
The detector and the CFD data set of the whole comprehensive working condition comprise
A detector; assuming that the CFD simulation average time consumption of each working condition is t, the total time consumption for constructing the CFD data set is t
And the data analysis module is used for performing data analysis by combining a space-time attention mechanism with an LSTM (Long Short-Term Memory) model, a GRU (Gated recovery Unit) model and a TCN (Temporal convolution network) model according to the characteristic data in the CFD data set, and outputting a prediction model of the time and space distribution concentration of the toxic gas.
The space-time attention mechanism learns weight distribution from the features, and then the weight distribution is acted on the original features, so that the distribution of the original features is changed, and the functions of increasing effective features and inhibiting ineffective features are achieved;
the space-time attention mechanism and the TCN model are combined as an example, and the structure diagram is shown in FIG. 2: the output result of the convolutional layer passes through a time attention module, a space attention module after convolution and then convolution to obtain a weighted output result;
and the data correction module is used for correcting and calibrating the prediction model according to all the characteristic parameters actually acquired on site.
The above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for rapidly predicting toxic gas is characterized by comprising the following steps: the method comprises the following steps:
the method comprises the following steps: constructing a CFD data set of a full-scene working condition based on the actual working condition, intensively setting sampling points in the whole simulation space, and setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition;
step two: based on the characteristic data in the constructed CFD data set, a time-space attention mechanism is combined with other deep learning models to output a prediction model of time and space distribution concentration of the toxic gas;
step three: and correcting and calibrating the model based on real data of the field detector in actual application.
2. The method of claim 1, wherein the method comprises: the actual working condition in the step one specifically comprises gas type, gas leakage condition, weather factors and surrounding building position coordinates, the gas leakage condition comprises gas leakage source position and gas leakage rate, and the weather factors comprise air temperature, humidity, air pressure, air speed, wind direction and atmospheric stability.
3. The method of claim 1, wherein the method comprises: the step one of constructing the CFD data set of the full-scene working condition specifically includes:
let the number of features of a working condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
Seed working conditions;
densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting sampling points at intervals of delta distance in 3 directions of the length, width and height, wherein d detectors are arranged at each sampling point, and each working condition has
A detector, a CFD data set of the whole comprehensive working condition has
A detector; assuming the CFD simulation average time consumption of each working condition is t, the total time consumption for constructing the CFD data set is t
And (4) obtaining data of the full-quantity detector at all the sampling points in the space under the full scene working condition after CFD simulation.
4. The method of claim 1, wherein the method comprises: and combining the space-time attention mechanism with an LSTM model, or a GRU model, or a TCN model in the second step.
5. The method of claim 1, wherein the method comprises: and in the second step, the space-time attention mechanism learns weight distribution from the features, and then the weight distribution is acted on the original features to change the distribution of the original features, wherein the space attention mechanism and the time attention mechanism are specifically included, the space attention mechanism is used for acquiring space correlation of different spatial coordinate points, and the time attention mechanism is used for acquiring time correlation of different times.
6. A toxic gas rapid prediction system is characterized in that: the system comprises a characteristic acquisition and construction module, a data analysis module and a data correction module;
the system comprises a characteristic acquisition and construction module, a data acquisition and construction module and a data acquisition and construction module, wherein the characteristic acquisition and construction module is used for acquiring all characteristic parameters in an actual working condition, constructing a CFD data set according to all the acquired characteristic parameters, intensively setting sampling points in the whole simulation space, and setting a full-quantity detector at the sampling points, wherein the full-quantity detector is used for detecting all the characteristic parameters in the actual working condition;
the data analysis module is used for outputting a prediction model of time and space distribution concentration of the toxic gas by combining a space-time attention mechanism with other deep learning models according to the CFD data set;
and the data correction module is used for correcting and calibrating the prediction model according to all the characteristic parameters actually acquired on site.
7. The system of claim 6, wherein the system is further configured to: in the characteristic acquisition and construction module, the actual working condition specifically comprises gas type, gas leakage condition, weather elements and surrounding building position coordinates, the gas leakage condition comprises gas leakage source position and gas leakage rate, and the weather elements comprise temperature, humidity, air pressure, wind speed, wind direction and atmospheric stability.
8. The system of claim 6, wherein the system comprises: in the feature acquisition and construction module, constructing the CFD dataset of the full-scene operating condition specifically includes:
let the number of features of a working condition be n, each feature having m i The typical value reflects the variation of the characteristic value, and then there are
Seed working conditions;
densely setting sampling points in the whole simulation space, setting a full detector at the sampling points, wherein the full detector is used for detecting all characteristics in the actual working condition: setting the length, width and height of the simulation space as a, b and c, setting sampling points at intervals of delta distance in 3 directions of the length, width and height, wherein d detectors are arranged at each sampling point, and each working condition has
A detector, a CFD data set of the whole comprehensive working condition has
A detector; assuming the CFD simulation average time consumption of each working condition is t, the total time consumption for constructing the CFD data set is t
And (4) obtaining the data of the full-scale detector at all the sampling points in the space under the full-scene working condition after CFD simulation.
9. The system of claim 6, wherein the system comprises: in the data analysis module, a space-time attention mechanism is combined with an LSTM model, a GRU model or a TCN model.
10. The system of claim 6, wherein the system is further configured to: in the data analysis module, a space-time attention mechanism learns weight distribution from the features, and then the weight distribution is acted on the original features, so that the distribution of the original features is changed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211267917.9A CN115600643B (en) | 2022-10-17 | 2022-10-17 | Method and system for rapidly predicting toxic gas |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211267917.9A CN115600643B (en) | 2022-10-17 | 2022-10-17 | Method and system for rapidly predicting toxic gas |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115600643A true CN115600643A (en) | 2023-01-13 |
| CN115600643B CN115600643B (en) | 2023-06-09 |
Family
ID=84846709
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211267917.9A Active CN115600643B (en) | 2022-10-17 | 2022-10-17 | Method and system for rapidly predicting toxic gas |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115600643B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116306377A (en) * | 2023-04-04 | 2023-06-23 | 中国石油大学(华东) | Method and system for rapidly predicting consequences of leakage accident of hydrogen station |
| CN116448965A (en) * | 2023-06-14 | 2023-07-18 | 四川省分析测试服务中心 | Multi-parameter poisonous and harmful gas detection system and method in limited space |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06117600A (en) * | 1992-09-30 | 1994-04-26 | Mitsui Toatsu Chem Inc | System for estimating leakage point and leakage volume of gas, steam or the like |
| CN101260804A (en) * | 2007-03-05 | 2008-09-10 | 普拉德研究及开发有限公司 | Systems and methods for well data compression |
| CN103914622A (en) * | 2014-04-04 | 2014-07-09 | 清华大学 | Quick chemical leakage predicating and warning emergency response decision-making method |
| CN110763809A (en) * | 2019-11-15 | 2020-02-07 | 中国石油大学(华东) | Experimental verification method for optimal arrangement scheme of gas detector |
| CN112270122A (en) * | 2020-10-10 | 2021-01-26 | 清华大学 | Inversion evaluation method for fire source parameters of building fire |
| CN113096343A (en) * | 2021-04-14 | 2021-07-09 | 合肥工业大学 | Multi-sensor cooperative automobile battery fire prevention system |
| CN113139444A (en) * | 2021-04-06 | 2021-07-20 | 上海工程技术大学 | Space-time attention mask wearing real-time detection method based on MobileNet V2 |
| CN114021501A (en) * | 2021-11-09 | 2022-02-08 | 华东理工大学 | Fire temperature field reconstruction method, system, computer equipment, medium and terminal |
| CN114240000A (en) * | 2021-12-31 | 2022-03-25 | 北京工业大学 | Air quality prediction method based on space-time graph convolution network |
| CN114693733A (en) * | 2022-03-15 | 2022-07-01 | 深圳高性能医疗器械国家研究院有限公司 | Motion prediction method and motion prediction device based on deep learning |
| CN114841031A (en) * | 2022-03-07 | 2022-08-02 | 中国人民解放军陆军防化学院 | Method for calculating diffusion concentration of simulated toxic and harmful gas in three-dimensional virtual training environment |
| CN115151808A (en) * | 2020-01-02 | 2022-10-04 | 罗姆有限责任公司 | Method and device for determining particle characteristics by multiparametric detection of scattered light and extinction signals |
-
2022
- 2022-10-17 CN CN202211267917.9A patent/CN115600643B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06117600A (en) * | 1992-09-30 | 1994-04-26 | Mitsui Toatsu Chem Inc | System for estimating leakage point and leakage volume of gas, steam or the like |
| CN101260804A (en) * | 2007-03-05 | 2008-09-10 | 普拉德研究及开发有限公司 | Systems and methods for well data compression |
| CN103914622A (en) * | 2014-04-04 | 2014-07-09 | 清华大学 | Quick chemical leakage predicating and warning emergency response decision-making method |
| CN110763809A (en) * | 2019-11-15 | 2020-02-07 | 中国石油大学(华东) | Experimental verification method for optimal arrangement scheme of gas detector |
| CN115151808A (en) * | 2020-01-02 | 2022-10-04 | 罗姆有限责任公司 | Method and device for determining particle characteristics by multiparametric detection of scattered light and extinction signals |
| CN112270122A (en) * | 2020-10-10 | 2021-01-26 | 清华大学 | Inversion evaluation method for fire source parameters of building fire |
| CN113139444A (en) * | 2021-04-06 | 2021-07-20 | 上海工程技术大学 | Space-time attention mask wearing real-time detection method based on MobileNet V2 |
| CN113096343A (en) * | 2021-04-14 | 2021-07-09 | 合肥工业大学 | Multi-sensor cooperative automobile battery fire prevention system |
| CN114021501A (en) * | 2021-11-09 | 2022-02-08 | 华东理工大学 | Fire temperature field reconstruction method, system, computer equipment, medium and terminal |
| CN114240000A (en) * | 2021-12-31 | 2022-03-25 | 北京工业大学 | Air quality prediction method based on space-time graph convolution network |
| CN114841031A (en) * | 2022-03-07 | 2022-08-02 | 中国人民解放军陆军防化学院 | Method for calculating diffusion concentration of simulated toxic and harmful gas in three-dimensional virtual training environment |
| CN114693733A (en) * | 2022-03-15 | 2022-07-01 | 深圳高性能医疗器械国家研究院有限公司 | Motion prediction method and motion prediction device based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 杜圣东;李天瑞;杨燕;王浩;谢鹏;洪西进;: "一种基于序列到序列时空注意力学习的交通流预测模型", 计算机研究与发展, no. 08, pages 149 - 162 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116306377A (en) * | 2023-04-04 | 2023-06-23 | 中国石油大学(华东) | Method and system for rapidly predicting consequences of leakage accident of hydrogen station |
| CN116306377B (en) * | 2023-04-04 | 2024-04-05 | 中国石油大学(华东) | Method and system for rapidly predicting consequences of leakage accident of hydrogen station |
| CN116448965A (en) * | 2023-06-14 | 2023-07-18 | 四川省分析测试服务中心 | Multi-parameter poisonous and harmful gas detection system and method in limited space |
| CN116448965B (en) * | 2023-06-14 | 2023-09-01 | 四川省分析测试服务中心 | Multi-parameter poisonous and harmful gas detection system and method in limited space |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115600643B (en) | 2023-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115600643B (en) | Method and system for rapidly predicting toxic gas | |
| Borrego et al. | Assessment of air quality microsensors versus reference methods: The EuNetAir Joint Exercise–Part II | |
| CN109492830B (en) | Mobile pollution source emission concentration prediction method based on time-space deep learning | |
| CN108108836B (en) | A kind of ozone concentration distribution forecasting method and system based on space-time deep learning | |
| Dey et al. | Hybrid CNN-LSTM and IoT-based coal mine hazards monitoring and prediction system | |
| CN108053071A (en) | Regional air pollutant concentration Forecasting Methodology, terminal and readable storage medium storing program for executing | |
| CN110298560A (en) | A kind of appraisal procedure, device and the storage medium of air pollution emission control effect | |
| CN114266200B (en) | Nitrogen dioxide concentration prediction method and system | |
| Korunoski et al. | Internet of things solution for intelligent air pollution prediction and visualization | |
| CN113632101B (en) | Method for predicting atmospheric pollution through vectorization analysis | |
| CN108399470B (en) | Indoor PM2.5 prediction method based on multi-example genetic neural network | |
| CN110348074A (en) | The method and device of climate change risk partition | |
| CN112884243A (en) | Air quality analysis and prediction method based on deep learning and Bayesian model | |
| CN113918673A (en) | Emergency evacuation path planning method in toxic gas leakage accident | |
| Mao et al. | A hybrid integrated deep learning model for predicting various air pollutants | |
| Mao et al. | Long time series ozone prediction in China: A novel dynamic spatiotemporal deep learning approach | |
| Campolongo et al. | Sensitivity analysis of the IMAGE Greenhouse model | |
| CN115629160A (en) | Air pollutant concentration prediction method and system based on space-time diagram | |
| Liu et al. | Research on data correction method of micro air quality detector based on combination of partial least squares and random forest regression | |
| Bhat et al. | Machine learning based prediction system for detecting air pollution | |
| Sá et al. | Two step calibration method for ozone low-cost sensor: Field experiences with the UrbanSense DCUs | |
| CN116502539B (en) | VOCs gas concentration prediction method and system | |
| CN111178631B (en) | Water lettuce intrusion distribution area prediction method and system | |
| Padilla et al. | Air quality prediction using recurrent air quality predictor with ensemble learning | |
| Wang et al. | The architecture and application of an automatic operational model system for basin scale water environment management and design making supporting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |