CN107300550A - A kind of method based on BP neural network model prediction atmosphere heavy metal concentration - Google Patents
A kind of method based on BP neural network model prediction atmosphere heavy metal concentration Download PDFInfo
- Publication number
- CN107300550A CN107300550A CN201710476059.1A CN201710476059A CN107300550A CN 107300550 A CN107300550 A CN 107300550A CN 201710476059 A CN201710476059 A CN 201710476059A CN 107300550 A CN107300550 A CN 107300550A
- Authority
- CN
- China
- Prior art keywords
- heavy metal
- concentration
- atmosphere
- data
- neural network
- 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.)
- Pending
Links
- 229910001385 heavy metal Inorganic materials 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000003062 neural network model Methods 0.000 title claims abstract description 18
- 239000002245 particle Substances 0.000 claims abstract description 30
- 238000010276 construction Methods 0.000 claims abstract description 9
- 238000001514 detection method Methods 0.000 claims abstract description 4
- 230000004044 response Effects 0.000 claims abstract description 4
- 239000008277 atmospheric particulate matter Substances 0.000 claims abstract description 3
- 238000007405 data analysis Methods 0.000 claims abstract description 3
- 238000012549 training Methods 0.000 claims description 20
- 238000004458 analytical method Methods 0.000 claims description 8
- 239000003344 environmental pollutant Substances 0.000 claims description 7
- 229910052751 metal Inorganic materials 0.000 claims description 7
- 231100000719 pollutant Toxicity 0.000 claims description 7
- 238000011160 research Methods 0.000 claims description 7
- 239000007789 gas Substances 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 239000012528 membrane Substances 0.000 claims description 5
- 239000002184 metal Substances 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 229910052785 arsenic Inorganic materials 0.000 claims description 4
- 230000036541 health Effects 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 239000013618 particulate matter Substances 0.000 claims description 4
- 238000003915 air pollution Methods 0.000 claims description 3
- 229910052782 aluminium Inorganic materials 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 229910052793 cadmium Inorganic materials 0.000 claims description 3
- 229910052804 chromium Inorganic materials 0.000 claims description 3
- 229910052802 copper Inorganic materials 0.000 claims description 3
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 239000000428 dust Substances 0.000 claims description 3
- 229910052742 iron Inorganic materials 0.000 claims description 3
- 229910052745 lead Inorganic materials 0.000 claims description 3
- 229910052759 nickel Inorganic materials 0.000 claims description 3
- 238000000918 plasma mass spectrometry Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 230000008878 coupling Effects 0.000 claims description 2
- 238000010168 coupling process Methods 0.000 claims description 2
- 238000005859 coupling reaction Methods 0.000 claims description 2
- 230000029087 digestion Effects 0.000 claims description 2
- 230000001939 inductive effect Effects 0.000 claims description 2
- 238000006396 nitration reaction Methods 0.000 claims description 2
- 229910052725 zinc Inorganic materials 0.000 claims description 2
- 238000004088 simulation Methods 0.000 claims 2
- 239000007788 liquid Substances 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 abstract description 10
- 230000002265 prevention Effects 0.000 abstract description 5
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 238000013459 approach Methods 0.000 abstract description 2
- 238000005070 sampling Methods 0.000 description 6
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 241001465754 Metazoa Species 0.000 description 4
- PXHVJJICTQNCMI-UHFFFAOYSA-N Nickel Chemical compound [Ni] PXHVJJICTQNCMI-UHFFFAOYSA-N 0.000 description 4
- 230000000875 corresponding effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000011133 lead Substances 0.000 description 4
- 239000011651 chromium Substances 0.000 description 3
- 239000010949 copper Substances 0.000 description 3
- 239000005022 packaging material Substances 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000001932 seasonal effect Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 description 1
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- BUGBHKTXTAQXES-UHFFFAOYSA-N Selenium Chemical compound [Se] BUGBHKTXTAQXES-UHFFFAOYSA-N 0.000 description 1
- 239000004411 aluminium Substances 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- RQNWIZPPADIBDY-UHFFFAOYSA-N arsenic atom Chemical compound [As] RQNWIZPPADIBDY-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- BDOSMKKIYDKNTQ-UHFFFAOYSA-N cadmium atom Chemical compound [Cd] BDOSMKKIYDKNTQ-UHFFFAOYSA-N 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 210000002249 digestive system Anatomy 0.000 description 1
- 238000007865 diluting Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000010419 fine particle Substances 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- QSHDDOUJBYECFT-UHFFFAOYSA-N mercury Chemical compound [Hg] QSHDDOUJBYECFT-UHFFFAOYSA-N 0.000 description 1
- 229910052753 mercury Inorganic materials 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000009659 non-destructive testing Methods 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 230000000505 pernicious effect Effects 0.000 description 1
- 238000007747 plating Methods 0.000 description 1
- 238000004094 preconcentration Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 239000010453 quartz Substances 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 229910052711 selenium Inorganic materials 0.000 description 1
- 239000011669 selenium Substances 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N silicon dioxide Inorganic materials O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000004876 x-ray fluorescence Methods 0.000 description 1
- 239000011701 zinc Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/71—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light thermally excited
- G01N21/73—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light thermally excited using plasma burners or torches
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
- G01N15/06—Investigating concentration of particle suspensions
- G01N15/0656—Investigating concentration of particle suspensions using electric, e.g. electrostatic methods or magnetic methods
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
- G01N27/62—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating the ionisation of gases, e.g. aerosols; by investigating electric discharges, e.g. emission of cathode
-
- 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
-
- G01N15/075—
Abstract
The invention discloses a kind of method based on BP neural network model prediction atmosphere heavy metal concentration, comprise the following steps:1) typical small survey region selection;2) survey region meteorological data is obtained;3) survey region atmosphere particle concentration data acquisition;4) survey region atmosphere particle concentration data detection;5) data analysis;6) model construction;7) heavy metal concentration in regional atmospheric particulate matter to be evaluated is carried out according to constructed BP neural network model to predict.The present invention sets up the fast prediction model of urban atmosphere heavy metal by the response relation between BP neural network and each Monitoring factors, so as to which the concentration to atmosphere heavy metal is estimated, a kind of new approaches are provided to the pollution prevention of China's urban atmosphere heavy metal, with extremely important realistic meaning.
Description
Technical field
The present invention relates to atmosphere pollution studying technological domain, it is specifically related to a kind of big based on BP neural network model prediction
The method of gas heavy metal concentration.
Background technology
Heavy metal-polluted in air is infected with natural source and artificial two kinds of source, by universe celestial body act on and the earth on it is various
Geologic process and make some heavy metal elements enter air in belong to natural source, the heavy metal artificially originated is mainly industrial life
A large amount of pernicious gases and dust containing heavy metal that production, motor vehicle exhaust emission and auto tire wear are produced etc..With motor vehicle
Increasingly increase, the atmosphere heavy metal pollution produced therewith more serious, there are some researches show atmosphere heavy metal pollution is to human body
Have significant impact in terms of health, animals and plants ecotoxicological, these heavy metal elements include lead, copper, cadmium, arsenic, chromium, nickel,
The elements such as selenium, mercury, in addition to nutrient iron, aluminium influential on sea hydrobiont.Harm of the heavy metal to environment is depended on first
It is the content in environment again in the physicochemical properties of its own, when these heavy metals run up to certain journey in animal body
When spending, i.e., the growing of animal, physiological loading can be directly affected, until causing the death of animal, while heavy metal
It can be absorbed by the body along food chain by digestive system, to the very harmful of crowd.Therefore to heavy metal concentration in air
Side's monitoring is just particularly important.
The monitoring of current heavy metal largely carries out pre-concentration sampling using film or ram, then (atom is inhaled by AAS
Receive spectrum), the measurement of the detecting instrument such as ICP-MS (inductivity coupled plasma mass spectrometry), AAS and ICP-MS can detect many simultaneously
Heavy Metallic Elements, test limit is low, and sensitivity is high.But be due to this off-line analysis mode temporal resolution it is not high, it is difficult to
The need for meeting research heavy metal element Transport And Transformation, and ICP-MS is expensive, and complex operation is also not susceptible to promote the use of.
It is also a kind of semicontinuous detecting instrument sampled based on film using the detecting instrument of XRF (X-ray fluorescence spectra analysis) technology, though
So temporal resolution is improved to a certain extent, and its detection belongs to Non-Destructive Testing, can carry out multiple element and detect simultaneously, but
Its expensive price causes the instrument to be not easy to popularization.So existing heavy metal monitoring technology is difficult to meet in real time, on a large scale
Monitoring, heavy metal on-line checking is a great problem of present atmosphere heavy metal research field.
At present, although many cities have been set up the monitoring station of Atmospheric particulates, but due to there is weather condition, it is artificial
Factor and other the reason for can not manipulate, the data of many websites can all be lacked or disconnected section, and for a huge sum of money in Atmospheric particulates
The research work of the monitoring analysis of category just more lacks.Correlative study by multiple linear regression model, artificial neural network,
SVMs etc. linearly or nonlinearly mathematical method be successfully established atmosphere pollution (including SO2, CO, NO2, O3,
PM2.5 etc.) forecast model, but on predict Heavy Metals In Atmospheric Particles concentration research be rarely reported.Therefore, to typical case
Heavy metal pollution species, time space distribution, pollution sources are identified and set up assessment China city in urban atmosphere particulate matter
The fast prediction model of atmosphere heavy metal pollution feature, has extremely important to the pollution prevention of China's urban atmosphere heavy metal
Meaning.
The content of the invention
Purpose be to provide a kind of method based on BP neural network model prediction atmosphere heavy metal concentration, for predict with
The concentration of assessment area atmosphere heavy metal, support is provided for work such as the formulations of corresponding prevention and control measure.
The technical scheme is that:
A kind of method based on BP neural network model prediction atmosphere heavy metal concentration, comprises the following steps:
1) typical small survey region selection:
Survey region of the selection with typical representative, typical small research area should possess following items essential characteristic:The
One, the region possesses typical traffic activity, construction and life discharge, and a large amount of gases of generation, particulate matter, road are raised
Dirt;Second, there is high energy-consuming enterprises and disposal of pollutants rich and influential family in the region, the air pollution in the region is by industrial pollution short distance
The influence of transmission, has hidden danger to health;3rd, the meteorological data in the region is complete, and modeling precision is high;4th,
The regional atmospheric sample is easily obtained;
2) survey region meteorological data is obtained:According to step 1) requirement select survey region, sampled point is determined, using normal
The meteorological data of automatic monitor station near atmospheric particulate cascade sampling device collection Atmospheric particulates, synchronous recording sampled point is advised,
Including:Temperature, humidity, air pressure and wind speed;
3) survey region atmosphere particle concentration data acquisition:The concentration of analysis test Heavy Metals In Atmospheric Particles, is carried
Take mainly includes with analysis heavy metal element:Al, As, Cd, Cr, Cu, Fe, Ni, Pb and Zn;Extracting method is:By 1/8 size
Sample filter membrane is cut into fine strip shape in after digestion tube, and total metalses are extracted using nitration mixture HNO3-HCl-HF-HClO4, are then treated
Survey GOLD FROM PLATING SOLUTION and belong to the concentration of element using inductive coupling plasma emission spectrograph and inductivity coupled plasma mass spectrometry survey
It is fixed;
4) survey region atmosphere particle concentration data detection:Using Grubbs methods to surveyed Atmospheric particulates initial data
Middle special datum is tested and rejected;The normal distribution-test of data is carried out using SPSS19.0;
5) data analysis:Analyze Heavy Metals In Atmospheric Particles concentration and meteorological data and corresponding particle diameter Atmospheric particulates are dense
The dependency relation of degree;By SPSS23.0 by 14 heavy metal species concentration in Atmospheric particulates and meteorological data (temperature, humidity, gas
Pressure, wind speed) and particle diameter atmosphere particle concentration progress multiple linear correlation analysis is corresponded to, relative coefficient is higher, shows two
Person's correlation is better;
6) model construction:Using meteorological factor and pollutant data as the input factor, atmosphere heavy metal concentration for output because
Son, builds the model based on BP neural network model prediction atmosphere heavy metal concentration, and related to measured value according to predicted value
The quality of property and error judgment model construction;
7) heavy metal concentration in regional atmospheric particulate matter to be evaluated is carried out according to constructed BP neural network model to predict:
According to the concentration of the meteorological factor in city period Heavy Metals In Atmospheric Particles corresponding with the pollutant data prediction city.
Further, in such scheme, the BP networks (Fig. 2) are multitiered networks, are divided into input layer, hidden layer and defeated
Go out between layer, each layer and carry out full connection;It realize multilayer study imagination, when given one input pattern of network, it by
Input layer unit passes to implicit layer unit, is then sent through output layer unit after successively being handled through implicit layer unit, is produced after processing
One output mode, if output response has error with desired output pattern, is unsatisfactory for requiring, is then transferred to error back propagation,
Error amount is successively reversely transmitted along connecting path and each layer connection weight is corrected, when each training mode, which is all met, to be required,
Then study terminates;In hands-on, first have to provide one group of training sample, each training sample therein by input sample and
Ideal output is to composition;When all reality outputs of network are consistent with its ideal output, training terminates;Otherwise, error is passed through
The inverse method propagated makes the preferable output of network consistent with reality output to correct weights;When repetition learning until sample lump is missed
Poor (such as following formula) reaches some required precision, i.e. E<Stop during ε (previously given precision), and record the power after now adjusting
Value.
Yk is desired output in formula, and ck is reality output, and m is learning sample number.
Further, in such scheme, during modeling, 70% is randomly selected as training data, is left 30% and is made
For checking data;To ensure the representativeness of training data, region, season, the data of day and night are randomly selected, and include it
In very big and minimum;Among 100 times successfully model, mould is simulated in maximum being once used as of selection training pattern coefficient correlation
Type, heavy metal concentration is carried out and prediction.
Further, in such scheme, whether the model for judging to build has the specific measurement index of universality
Have:
1. coefficient correlation (R), for reflecting dependency relation level of intimate, the fitness for measurement model between variable
Energy;
2. mean absolute error (MAE), is that the absolute value of all single analogues value and the deviation of measured value is averaged, is used for
The prediction effect of assessment models;
3. root-mean-square error (MSE), is the square root of the average value of the quadratic sum of error, for judging the reliable of the model
Property.
Further, in such scheme, after model training is finished, the meteorological data (wind of Free Region random time section
Speed, air pressure, temperature, humidity) and correspondence particle diameter atmosphere particle concentration as input the factor, with the model prediction trained this when
The concentration of section, survey region correspondence atmosphere heavy metal.
The beneficial effects of the invention are as follows:The dirt of Heavy Metals In Atmospheric Particles under the Various Seasonal of city is probed into present invention analysis
Feature is contaminated, while the routine monitoring factors such as meteorological data are collected, by the response between BP neural network and each Monitoring factors
Relation sets up the fast prediction model of urban atmosphere heavy metal, so that the concentration to atmosphere heavy metal is estimated, it is corresponding
The work such as the formulation of prevention and control measure provide support, establish the fast prediction model for assessing China's urban atmosphere heavy metal pollution,
A kind of new approaches are provided to the pollution prevention of China's urban atmosphere heavy metal, with extremely important realistic meaning.
Brief description of the drawings
Fig. 1 is the techniqueflow block diagram of the present invention;
Fig. 2 is the flow chart of BP neural network model construction;
Fig. 3 is atmosphere particle concentration and meteorological data;
Fig. 4 is the variation tendency of Various Seasonal atmosphere heavy metal concentration;
Fig. 5 Nanjing meteorological data of 2015, PM2.5 concentration and the heavy metal concentration of prediction.
Embodiment
Modeled by taking the celestial woods school district of Nanjing University as an example, concentration of the application model to heavy metal in the PM2.5 in 2015 of Nanjing
Predicted.
Nanjing is the second largest city of China Yangtze River Delta Area, the important comprehensive Industrial Complex of China, East China
The particularly important transport hub in area.The quick Process of Urbanization Construction in Nanjing is generated than more serious air pollution in recent years,
Essentially from a large amount of gases produced by industrial production, traffic activity, construction and life discharge, particulate matter, dust on the roads
Deng.The main economic development zone in Nanjing, technological development zone etc. are around whole Nanjing, the wherein chemical work in the Nanjing of direct north
Industry garden is the compact district of chemical enterprise, summarizes the main high energy-consuming enterprises in Nanjing and disposal of pollutants rich and influential family.Nanjing University
Celestial woods school district is located in Xian Lin university cities, but also relatively near (5.6 kilometers) according to Nanjing petrochemical industry.It can be seen that, the air in this region is dirty
Dye will be influenceed by industrial pollution short-distance transmission, have huge hidden danger to health, should be by government's dependent part
Door and the great attention of the public.
Sampler has used classification cutting head (2.5 μm, 5 μm, 10 μm and 100 μm), and sampling media is high purity quartz filter membrane
(QM-Whatman, 20.3cm × 25.4cm), sampling flow is 100L/min, and sampler atmospheric connection height is left away from ground 1.5m
It is right.The sampling time of celestial woods be 22 days-May 7 April in 2014 (spring), 9-July of July 26 (summer), October 13-
October 22 (autumn), 27 days 18 days-January of January in 2015 (winter), total acquire 156 samples altogether, and synchronous recording is certainly
The meteorological data of dynamic monitoring station, including temperature, humidity, air pressure and wind speed etc..The equal constant temperature and humidity 24h (temperature of filter membrane before and after sampling
25 DEG C, humidity 50%) and weigh, to determine quality, it is finally stored in drying box to pretreatment.Every filter membrane is bisected into 8 parts,
Wherein 1 part is used for total amount and determines, and 7 parts give over to other experiment use in addition.
Statistical analysis, Nanjing atmospheric particle concentration and meteorological factor are as shown in figure 3, result is shown:Nanjing is big
Aerated particle thing is more gathered in fine particle PM2.5, account for the 49% of total amount.In addition, celestial woods PM2.5 concentration is 78 μ
G/m3, the sample for having 38% has exceeded the average daily limit value of national standard (75 μ g/m3).Generally speaking, celestial woods winter PM2.5 and
Heavy metal concentration is significantly higher than summer in PM2.5-5.The variation tendency of atmosphere heavy metal concentration is as shown in figure 4, result is shown:Greatly
The concentration in part atmosphere heavy metal winter is significantly higher than summer.
For checking mode input parameter and the internal association of metal, correlation analysis is carried out with SPSS23.0, is as a result shown:
Metal concentration, in notable negatively correlated, is in notable positive correlation with air pressure, most of metal is with humidity without significantly correlated with wind speed, temperature
Property.Wind speed has certain diluting effect to atmosphere pollution, and wind speed is more high more is easily reduced atmosphere particle concentration.In general,
Temperature is higher, and atmospheric pressure is lower, and air convection movement is more obvious, and Atmospheric particulates diffusion rate is faster so that Atmospheric particulates
The concentration of middle heavy metal concentration is also lower.Metal concentration and PM are in notable positive correlation, have common denominator such as between this and they
Industrial discharge, transportation emission, daily culinary art, biomass combustion etc. are relevant.
The processing handled well is uniformly called in through the compiled program of MATLAB 2013a instruments, carry out model training and
Checking, obtained final result is as shown in table 1:Training pattern r values are maintained at 0.423-0.765 (PM2.5), 0.540-0.672
(PM2.5-5), between 0.543-0.682 (PM5-10) and 0.560-0.684 (PM10-100), model checking r is maintained at
0.502-0.752 (PM2.5), 0.559-0.786 (PM2.5-5), 0.560-0.772 (PM5-10) and 0.622-0.783
(PM10-100) between.The result of comprehensive R values, MAE values and RMSE value, it can be seen that the predicted value and actual value correlation of model compared with
Good, the model is relatively reliable.
Table 1:Coefficient correlation, mean absolute error and root mean square between BP neural network pattern die analog values and measured value are missed
Difference
After model training is finished, by Nanjing annual meteorological datas (wind speed, air pressure, temperature, humidity) in 2015 and
PM2.5Concentration predicts the concentration of period correspondence atmosphere heavy metal as the input factor, as a result shows:2015, Nanjing
PM2.5Middle heavy metal concentration is ordered as Al>Fe>Zn>Pb>Cu>As>Cr>Ni>Cd, winter>Spring>Autumn>Summer, pollute most tight
The period of weight is in December and January.Wherein, typical element As and Pb annual tendency chart is as shown in Figure 5.
It is described above, only it is the case study on implementation of the present invention, is not the limitation for making other forms to the present invention, it is any
Those skilled in the art are changed or are modified as the equivalent of equivalent variations possibly also with the technology contents of the disclosure above
Embodiment.But it is every without departing from technical solution of the present invention content, the technical spirit according to the present invention implements what is made to more than
Any simple modification, equivalent variations and remodeling, still belong to the protection domain of technical solution of the present invention.
Claims (6)
1. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration, it is characterised in that comprise the following steps:
1) typical small survey region selection:
Survey region of the selection with typical representative, typical small research area should possess following items essential characteristic:First, should
Region possesses typical traffic activity, construction and life discharge, and produces a large amount of gases, particulate matter, dust on the roads;The
Two, there is high energy-consuming enterprises and disposal of pollutants rich and influential family in the region, the air pollution in the region is by industrial pollution short-distance transmission
Influence, there is hidden danger to health;3rd, the meteorological data in the region is complete, and modeling precision is high;4th, the area
Domain atmospheric sample is easily obtained;
2) survey region meteorological data is obtained:According to step 1) requirement select survey region, sampled point is determined, using conventional big
Aerated particle thing andersen sampler gathers the meteorological data of automatic monitor station near Atmospheric particulates, synchronous recording sampled point, including:
Temperature, humidity, air pressure and wind speed;
3) survey region atmosphere particle concentration data acquisition:Analysis test Heavy Metals In Atmospheric Particles concentration, extract and
Analysis heavy metal element mainly includes:Al, As, Cd, Cr, Cu, Fe, Ni, Pb and Zn;Extracting method is:By the sample of 1/8 size
Filter membrane is cut into fine strip shape in after digestion tube, and total metalses are extracted using nitration mixture HNO3-HCl-HF-HClO4, then to be measured molten
The concentration of metallic element is determined using inductive coupling plasma emission spectrograph and inductivity coupled plasma mass spectrometry in liquid;
4) survey region atmosphere particle concentration data detection:Using Grubbs methods to special in surveyed Atmospheric particulates initial data
Different value is tested and rejected;The normal distribution-test of data is carried out using SPSS19.0;
5) data analysis:Analyze Heavy Metals In Atmospheric Particles concentration and meteorological data and corresponding particle diameter atmosphere particle concentration
Dependency relation;By SPSS23.0 by 14 heavy metal species concentration in Atmospheric particulates and meteorological data (temperature, humidity, air pressure, wind
Speed) and particle diameter atmosphere particle concentration progress multiple linear correlation analysis is corresponded to, relative coefficient is higher, shows that both are related
Property is better;
6) model construction:Using meteorological factor and pollutant data as the input factor, atmosphere heavy metal concentration is the output factor, structure
The model based on BP neural network model prediction atmosphere heavy metal concentration is built, and according to predicted value and the correlation and mistake of measured value
The quality that poor judgment models are built;
7) heavy metal concentration in regional atmospheric particulate matter to be evaluated is carried out according to constructed BP neural network model to predict:According to
The concentration of the meteorological factor in city period Heavy Metals In Atmospheric Particles corresponding with the pollutant data prediction city.
2. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration according to claim 1, it is special
Levy and be, the BP networks are multitiered networks, be divided between input layer, hidden layer and output layer, each layer and carry out full connection;It is real
The imagination of multilayer study is showed, when given one input pattern of network, it passes to implicit layer unit by input layer unit, through hidden
Output layer unit is then sent through after successively being handled containing layer unit, an output mode is produced after processing, if output response and phase
Hope output mode have error, be unsatisfactory for requiring, be then transferred to error back propagation, error amount is successively reversely transmitted along connecting path
And each layer connection weight is corrected, when each training mode, which is all met, to be required, then study terminates;In hands-on, first have to
One group of training sample is provided, each training sample therein is by input sample and preferable output to constituting;When all realities of network
When border exports consistent with its ideal output, training terminates;Otherwise, network is made to correct weights by the method for error Back-Propagation
Ideal output is consistent with reality output;When repetition learning until sample set overall error (such as following formula) reaches some required precision, i.e. E
<Stop during ε (previously given precision), and record the weights after now adjusting.
<mrow>
<mi>E</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msup>
<mi>y</mi>
<mi>k</mi>
</msup>
<mo>-</mo>
<msup>
<mi>c</mi>
<mi>k</mi>
</msup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
Yk is desired output in formula, and ck is reality output, and m is learning sample number.
3. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration according to claim 2, it is special
Levy and be, during modeling, randomly select 70% as training data, be left 30% as checking data;Successfully built at 100 times
Among mould, selection training pattern coefficient correlation it is maximum once as simulation model, heavy metal concentration is carried out and prediction.
4. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration according to claim 2, it is special
Levy and be, during modeling, randomly select 70% as training data, be left 30% as checking data;To ensure to train number
According to representativeness, randomly select region, season, the data of day and night, and include it is therein greatly and minimum;100
It is secondary successfully model among, selection training pattern coefficient correlation it is maximum once as simulation model, heavy metal concentration carry out with
Prediction.
5. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration according to claim 1, it is special
Levy and be, whether the specific measurement index with universality has the model for judging to build:Coefficient correlation (R), average absolute
Error (MAE) and root-mean-square error (MSE).
6. a kind of method based on BP neural network model prediction atmosphere heavy metal concentration according to claim 1, it is special
Levy and be, after model training is finished, the meteorological data (wind speed, air pressure, temperature, humidity) of Free Region random time section and correspondingly
Particle diameter atmosphere particle concentration is as the input factor, with the model prediction trained the period, survey region correspondence air weight
The concentration of metal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710476059.1A CN107300550A (en) | 2017-06-21 | 2017-06-21 | A kind of method based on BP neural network model prediction atmosphere heavy metal concentration |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710476059.1A CN107300550A (en) | 2017-06-21 | 2017-06-21 | A kind of method based on BP neural network model prediction atmosphere heavy metal concentration |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107300550A true CN107300550A (en) | 2017-10-27 |
Family
ID=60135927
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710476059.1A Pending CN107300550A (en) | 2017-06-21 | 2017-06-21 | A kind of method based on BP neural network model prediction atmosphere heavy metal concentration |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107300550A (en) |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107976515A (en) * | 2017-11-20 | 2018-05-01 | 安徽优思天成智能科技有限公司 | A kind of city pollutant of vehicle exhaust concentration distribution Forecasting Methodology |
CN108120661A (en) * | 2017-12-19 | 2018-06-05 | 北京理工大学 | Particle content spatial and temporal distributions assay method in a kind of urban air |
CN108593557A (en) * | 2018-03-13 | 2018-09-28 | 杭州电子科技大学 | Based on TE-ANN-AWF mobile pollution source telemetry errors compensation methodes |
CN109900865A (en) * | 2019-04-02 | 2019-06-18 | 西京学院 | A kind of air pollution detection system neural network based |
CN110398446A (en) * | 2019-08-01 | 2019-11-01 | 成都信息工程大学 | A kind of PM of combination AOD and meteorologic parameter2.5Concentration approximating method |
CN110487689A (en) * | 2018-05-14 | 2019-11-22 | 扬州大学 | Measuring and calculating of the fugitive dust particulate matter in proximate region with respect to mean concentration and comparative approach in city |
CN110514626A (en) * | 2019-07-23 | 2019-11-29 | 北京英视睿达科技有限公司 | The data calibration method and air pollution surveillance system of air pollution surveillance system |
CN110619384A (en) * | 2019-08-13 | 2019-12-27 | 浙江工业大学 | PM2.5 concentration value prediction method based on neural network |
CN110779843A (en) * | 2019-09-30 | 2020-02-11 | 成都市环境保护科学研究院 | Typing analytic system of atmosphere secondary particle pollution process |
CN110796284A (en) * | 2019-09-20 | 2020-02-14 | 平安科技(深圳)有限公司 | Method and device for predicting pollution level of fine particulate matters and computer equipment |
CN110852493A (en) * | 2019-10-25 | 2020-02-28 | 天津大学 | Atmospheric PM2.5 concentration prediction method based on multiple model comparisons |
CN111060467A (en) * | 2020-01-03 | 2020-04-24 | 四川中衡检测技术有限公司 | Method for rapidly determining trace metals in atmospheric suspended particles |
CN111380787A (en) * | 2020-05-18 | 2020-07-07 | 南京信息工程大学 | LSTM-NN time sequence prediction-based health risk evaluation method for heavy metals in PM2.5 |
CN111476258A (en) * | 2019-01-24 | 2020-07-31 | 杭州海康威视数字技术股份有限公司 | Feature extraction method and device based on attention mechanism and electronic equipment |
CN112649337A (en) * | 2020-12-21 | 2021-04-13 | 张家口市杰星电子科技有限公司 | Oil smoke online monitoring method and device |
CN112819217A (en) * | 2021-01-27 | 2021-05-18 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Method, system and storage medium for predicting main influence factors of mobile source pollution emission |
CN113092450A (en) * | 2021-04-13 | 2021-07-09 | 南京信息工程大学 | Atmospheric aerosol rapid detection traceability device and method |
CN113176216A (en) * | 2021-01-12 | 2021-07-27 | 生态环境部卫星环境应用中心 | Ozone precursor VOCs high-value area satellite remote sensing identification method |
CN113533644A (en) * | 2021-06-26 | 2021-10-22 | 西北工业大学 | Method for monitoring normal atmospheric pollutants on face source based on unmanned aerial vehicle |
CN113627529A (en) * | 2021-08-11 | 2021-11-09 | 成都佳华物链云科技有限公司 | Air quality prediction method, device, electronic equipment and storage medium |
CN113723534A (en) * | 2021-09-02 | 2021-11-30 | 南京润内克西信息科技有限公司 | Urban raise dust on-line monitoring system based on BP neural network |
CN114487284A (en) * | 2021-12-31 | 2022-05-13 | 武汉怡特环保科技有限公司 | Method and system for measuring concentration of heavy metal in air |
TWI812051B (en) * | 2022-03-08 | 2023-08-11 | 國立臺灣師範大學 | Volatile organic compounds concentration forecast method |
CN116758342A (en) * | 2023-06-01 | 2023-09-15 | 中国地质科学院矿产资源研究所 | Atmospheric pollution grade assessment method and device based on rare earth mineral area |
CN117313428A (en) * | 2023-11-28 | 2023-12-29 | 西安中核核仪器股份有限公司 | Pollutant concentration online monitoring and correcting method for thermal power plant |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105911037A (en) * | 2016-04-19 | 2016-08-31 | 湖南科技大学 | Manganese and associated heavy metal distribution prediction method of soil-water interface contaminated flow in manganese mine area |
-
2017
- 2017-06-21 CN CN201710476059.1A patent/CN107300550A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105911037A (en) * | 2016-04-19 | 2016-08-31 | 湖南科技大学 | Manganese and associated heavy metal distribution prediction method of soil-water interface contaminated flow in manganese mine area |
Non-Patent Citations (2)
Title |
---|
冷湘梓 等: ""基于非线性数学方法的PM2.5中重金属浓度模拟"", 《中国环境科学》 * |
王俭 等: ""基于人工神经网络的区域水环境承载力评价模型及其应用"", 《生态学杂志》 * |
Cited By (34)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107976515A (en) * | 2017-11-20 | 2018-05-01 | 安徽优思天成智能科技有限公司 | A kind of city pollutant of vehicle exhaust concentration distribution Forecasting Methodology |
CN108120661A (en) * | 2017-12-19 | 2018-06-05 | 北京理工大学 | Particle content spatial and temporal distributions assay method in a kind of urban air |
CN108593557A (en) * | 2018-03-13 | 2018-09-28 | 杭州电子科技大学 | Based on TE-ANN-AWF mobile pollution source telemetry errors compensation methodes |
CN108593557B (en) * | 2018-03-13 | 2020-08-11 | 杭州电子科技大学 | Remote measurement error compensation method based on TE-ANN-AWF (transverse electric field analysis) -based mobile pollution source |
CN110487689B (en) * | 2018-05-14 | 2022-02-18 | 扬州大学 | Method for measuring, calculating and comparing relative average concentration of dust particles in similar areas in city |
CN110487689A (en) * | 2018-05-14 | 2019-11-22 | 扬州大学 | Measuring and calculating of the fugitive dust particulate matter in proximate region with respect to mean concentration and comparative approach in city |
CN111476258B (en) * | 2019-01-24 | 2024-01-05 | 杭州海康威视数字技术股份有限公司 | Feature extraction method and device based on attention mechanism and electronic equipment |
CN111476258A (en) * | 2019-01-24 | 2020-07-31 | 杭州海康威视数字技术股份有限公司 | Feature extraction method and device based on attention mechanism and electronic equipment |
CN109900865A (en) * | 2019-04-02 | 2019-06-18 | 西京学院 | A kind of air pollution detection system neural network based |
CN110514626B (en) * | 2019-07-23 | 2024-01-26 | 北京英视睿达科技股份有限公司 | Data calibration method of atmosphere pollution monitoring system and atmosphere pollution monitoring system |
CN110514626A (en) * | 2019-07-23 | 2019-11-29 | 北京英视睿达科技有限公司 | The data calibration method and air pollution surveillance system of air pollution surveillance system |
CN110398446A (en) * | 2019-08-01 | 2019-11-01 | 成都信息工程大学 | A kind of PM of combination AOD and meteorologic parameter2.5Concentration approximating method |
CN110619384A (en) * | 2019-08-13 | 2019-12-27 | 浙江工业大学 | PM2.5 concentration value prediction method based on neural network |
CN110796284A (en) * | 2019-09-20 | 2020-02-14 | 平安科技(深圳)有限公司 | Method and device for predicting pollution level of fine particulate matters and computer equipment |
CN110779843A (en) * | 2019-09-30 | 2020-02-11 | 成都市环境保护科学研究院 | Typing analytic system of atmosphere secondary particle pollution process |
CN110852493A (en) * | 2019-10-25 | 2020-02-28 | 天津大学 | Atmospheric PM2.5 concentration prediction method based on multiple model comparisons |
CN111060467A (en) * | 2020-01-03 | 2020-04-24 | 四川中衡检测技术有限公司 | Method for rapidly determining trace metals in atmospheric suspended particles |
CN111380787B (en) * | 2020-05-18 | 2022-04-01 | 南京信息工程大学 | LSTM-NN time sequence prediction-based health risk evaluation method for heavy metals in PM2.5 |
CN111380787A (en) * | 2020-05-18 | 2020-07-07 | 南京信息工程大学 | LSTM-NN time sequence prediction-based health risk evaluation method for heavy metals in PM2.5 |
CN112649337A (en) * | 2020-12-21 | 2021-04-13 | 张家口市杰星电子科技有限公司 | Oil smoke online monitoring method and device |
CN113176216B (en) * | 2021-01-12 | 2022-03-15 | 生态环境部卫星环境应用中心 | Ozone precursor VOCs high-value area satellite remote sensing identification method |
CN113176216A (en) * | 2021-01-12 | 2021-07-27 | 生态环境部卫星环境应用中心 | Ozone precursor VOCs high-value area satellite remote sensing identification method |
CN112819217A (en) * | 2021-01-27 | 2021-05-18 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Method, system and storage medium for predicting main influence factors of mobile source pollution emission |
CN113092450A (en) * | 2021-04-13 | 2021-07-09 | 南京信息工程大学 | Atmospheric aerosol rapid detection traceability device and method |
CN113533644A (en) * | 2021-06-26 | 2021-10-22 | 西北工业大学 | Method for monitoring normal atmospheric pollutants on face source based on unmanned aerial vehicle |
CN113627529A (en) * | 2021-08-11 | 2021-11-09 | 成都佳华物链云科技有限公司 | Air quality prediction method, device, electronic equipment and storage medium |
CN113723534A (en) * | 2021-09-02 | 2021-11-30 | 南京润内克西信息科技有限公司 | Urban raise dust on-line monitoring system based on BP neural network |
CN114487284A (en) * | 2021-12-31 | 2022-05-13 | 武汉怡特环保科技有限公司 | Method and system for measuring concentration of heavy metal in air |
CN114487284B (en) * | 2021-12-31 | 2023-09-08 | 武汉怡特环保科技有限公司 | Method and system for measuring concentration of heavy metal in air |
TWI812051B (en) * | 2022-03-08 | 2023-08-11 | 國立臺灣師範大學 | Volatile organic compounds concentration forecast method |
CN116758342B (en) * | 2023-06-01 | 2023-12-15 | 中国地质科学院矿产资源研究所 | Atmospheric pollution grade assessment method and device based on rare earth mineral area |
CN116758342A (en) * | 2023-06-01 | 2023-09-15 | 中国地质科学院矿产资源研究所 | Atmospheric pollution grade assessment method and device based on rare earth mineral area |
CN117313428A (en) * | 2023-11-28 | 2023-12-29 | 西安中核核仪器股份有限公司 | Pollutant concentration online monitoring and correcting method for thermal power plant |
CN117313428B (en) * | 2023-11-28 | 2024-02-23 | 西安中核核仪器股份有限公司 | Pollutant concentration online monitoring and correcting method for thermal power plant |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107300550A (en) | A kind of method based on BP neural network model prediction atmosphere heavy metal concentration | |
CN112905560B (en) | Air pollution prediction method based on multi-source time-space big data deep fusion | |
Ding et al. | Transport characteristics and origins of carbon monoxide and ozone in Hong Kong, South China | |
Pekney et al. | Application of PSCF and CPF to PMF-modeled sources of PM2. 5 in Pittsburgh | |
Vignati et al. | Global scale emission and distribution of sea-spray aerosol: Sea-salt and organic enrichment | |
Makra et al. | Airborne pollen in three European cities: Detection of atmospheric circulation pathways by applying three‐dimensional clustering of backward trajectories | |
CN108268935B (en) | PM2.5 concentration value prediction method and system based on time sequence recurrent neural network | |
Engvall et al. | Changes in aerosol properties during spring-summer period in the Arctic troposphere | |
Kukkonen et al. | Modelling of the urban concentrations of PM 2.5 on a high resolution for a period of 35 years, for the assessment of lifetime exposure and health effects | |
CN110989041B (en) | Method and system for forecasting dust-haze and sand-dust weather | |
CN107121650A (en) | Atmospheric particulates heavy metal magnetics appraisal procedure based on SVMs technology | |
CN115438848A (en) | PM based on deep mixed graph neural network 2.5 Long-term concentration prediction method | |
Kajino et al. | Modeling transition metals in East Asia and Japan and its emission sources | |
CN110503348B (en) | Individual air pollution exposure simulation measurement method based on position matching | |
CN108170927A (en) | A kind of PM2.5 remote sensing inversion methods based on MODIS | |
CN108169316A (en) | Atmosphere heavy metal pollution appraisal procedure based on support vector machines and leaf magnetics | |
dos Santos et al. | Analysis of atmospheric aerosol (PM2. 5) in Recife city, Brazil | |
Masri et al. | Use of visual range measurements to predict fine particulate matter exposures in Southwest Asia and Afghanistan | |
CN108072597A (en) | Atmosphere particle concentration monitoring method based on support vector machines and leaf magnetics | |
Wu et al. | STUDY ON PREDICTION MODEL OF SPACE-TIME DISTRIBUTION OF AIR POLLUTANTS BASED ON ARTIFICIAL NEURAL NETWORK. | |
Seares et al. | Optimization of operational ICP OES parameters and application to PM10 monitoring associated to sugarcane burning | |
Rochimawati et al. | Prediction and modelling of total suspended particulate generation on ultisol and andisol soil | |
CN113010850A (en) | Method for predicting concentration of trivalent arsenic in atmospheric fine particles based on GIS | |
CN113032729A (en) | Convenient method for predicting atmospheric particulate heavy metal health risk spatial distribution | |
Shaziayani et al. | Three-days ahead prediction of daily maximum concentrations of PM10 using decision tree approach. |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171027 |