Background technology
Air is most basic environmental key-element for the survival of mankind, but as rapid industrial development, energy-consuming increase,
The discharge of dust, industrial waste gas and vehicle exhaust causes the sour gas such as sulfur dioxide and nitrogen oxides to get into the air, to big
Gas causes serious pollution.Air contaminant treatment turns into the focus studied now.China's atmosphere pollution is mainly sulphur
Oxide and nitrogen oxide, effective contamination containment enterprise exhaust emission gas will be helpful to improve air quality.19 end of the centurys
So far, the development of power industry is more and more faster for invention electric power, and electrical equipment manufacturing technology is also improving constantly, and this causes electricity usage
Field is constantly expanding, and the arrival for power information acquisition system (i.e. " electric extraction system ") provides good place mat.
As country is constantly lifted to the attention degree for improving air quality, prevention and control of air pollution work tends to more careful
Change, generalization, generate a series of data of sign air regimes and pollution control.Internet, ecommerce and movement are logical
The prosperous of letter promotes the intercommunication of every profession and trade data message and shared, and has concatenated the information " isolated island " of script.One extensive raw
Produce, share and opened with the epoch of application data.How above-mentioned data are applied by information sharing, preferably serve reality
Work, just turn into hot subject.To be effectively improved air quality, although environmental administration of governments at all levels proposes to close down contaminating enterprises
Management and control measures, but still suffer from the phenomenon of enterprise's " vacation row, production of draging out an ignoble existence ".
A kind of atmosphere pollution monitoring method based on Internet of Things of the disclosure of the invention of Patent No. 201410442826.3 and
System, this method include:The concentration of pollutant in air is detected, obtains first pollution thing concentration value;It is dense based on first pollution thing
Angle value and first pollution thing concentration threshold, judge whether first pollution thing concentration value is more than first pollution thing concentration threshold, obtain
First judged result;When first pollution thing concentration value is more than first pollution thing concentration threshold, atmospheric cleaning operation is performed, one
After preset time, the concentration of pollutant in air is detected, obtains the second pollutant concentration value;When the second pollutant concentration value not
During more than first pollution thing concentration threshold, stop atmospheric cleaning operation.With it, being capable of pollutant concentration value in an atmosphere
When exceeded, corresponding atmospheric cleaning operation is carried out in time, effectively alleviates State of Air pollution, and it is net to reduce air in time
Change a large amount of electric power that operation expends, save the energy.
The patent of Patent No. 201520792900.4 discloses a kind of on-line monitoring atmosphere pollution system, including sampling
Device, A/D converter, controller, APP devices and WiFi module, the output end of the sampler are electrically connected monitoring smoke dust
System, flue gas monitoring subsystem, flow velocity measuring unit, heavy metal analysis device, the monitoring smoke dust subsystem, flue gas monitoring subsystem
System, flow velocity measuring unit, the output end of heavy metal analysis device electrical connection A/D converter.The on-line monitoring atmosphere pollution system,
By monitoring smoke dust subsystem, flue gas monitoring subsystem, flow velocity measuring unit, heavy metal analysis device can to the flue dust of air,
Flue gas, flow velocity, content of beary metal are monitored, and when content exceedes normal value, can carry out remote alarms online by alarm,
Also, testing result can be sent remotely to by WiFi module on the APP softwares of mobile phone or electronic equipment, realize that remote online is supervised
Control, measurement, analysis, contrast.
However, said system or device can not solve the problems, such as to propose in background very well, so the present invention will be with monitoring
Contaminating enterprises are starting point, are support by power information acquisition system, build up an atmosphere pollution shared based on big data
Prevent and treat electricity consumption data analysis platform.
The content of the invention
Present invention aims at provide based on the shared prevention and control of air pollution electricity consumption data analysis platform of big data, with electricity consumption
Based on information acquisition system, with the shared analysis of big data and technological means is predicted as, is realized to She Qi contaminating enterprises and various regions
Commercial power data remote detecting, strong support is provided to pollute the formulation of management and control measures.
To achieve the above object, the technical solution adopted by the present invention is:Used based on the prevention and control of air pollution that big data is shared
Electric Data Analysis Platform, including power information collecting unit, transport module and system main website, the power information collecting unit bag
Measuring terminal equipment and the apparatus for acquiring and storing installed at the scene are included, the system main website is included in server group and data analysis
The heart, the system main website are connected with the power information collecting unit by the transport module, the data analysis center bag
Include the big data based on BP neural network and share analysis model.
Further, the shared analysis model of the big data passes through the analysis to electric power and environment protection digital, structure BP nerves
Neural network forecast unit, cyclic forecast is carried out to air quality, the BP neural network includes input layer, hidden layer and output
Layer, the hidden layer include single hidden layer and more hidden layers.
Further, the big data shares analysis model and realizes that step includes, and 1) achievement data input:With environmental protection with
Electric power shared data, the input pointer for carrying out BP neural network are analyzed and are acquired;2) achievement data is handled:To different type
Input pointer, be classified by the difference of physical significance, dimension and the order of magnitude, and initial data is normalized place
Reason;3) training sample is formulated:Using the data of current acquisition time section as a cycle, the training sample as BP neural network
This, to be predicted inspection to the air quality index in next cycle;4) network training is carried out:Using single hidden layer or more hidden layers
BP neural network carry out air quality index prediction, desired value is set, and using the air quality index of predetermined period as defeated
Go out value to be exported;5) prediction result is analyzed:Next circulated air quality is predicted by BP neural network, and makes prediction curve
Value carries out registration with actual value and compared, and draws prediction accuracy.
Further, the input pointer of the BP neural network include air pollution granule density, air quality index with
Management and control enterprise electricity, the air pollution particle include dust, industrial waste gas and the pollution of motor vehicle exhaust emission particle.
Further, the server group include database server, disk array, application server, front server,
Interface server, work station, global positioning system (GPS) clock, the network equipment of fire wall and correlation.
Further, the measuring terminal equipment includes intelligent electric meter, and the apparatus for acquiring and storing includes data collecting card
And data concentrator, the hierarchical memory being connected respectively with the data collecting card, some intelligent electric meters in star distribution with
One data collecting card connection.
Further, the transport module includes short range measuring terminal network and remote wireless network, the short range metering
Terminal network is transmitted using electric force carrier transmission or WIFI, and the remote wireless network is transmitted using long range point-to-point type.
Further, the remote wireless network includes transceiving radio station, and some apparatus for acquiring and storing are in vertical
Gauze distribution is connected with a transceiving radio station.
Further, the data analysis center connection digital-scroll technique unit, the digital-scroll technique unit connect string respectively
Row communication interface, human-machine interface unit.
Further, the digital-scroll technique unit includes data setting panel and LED display panel, the data setting face
Plate includes several data input buttons and order button.
The beneficial effects are mainly as follows the following aspects:
1st, the present invention utilizes electric power and the shared support air contaminant treatment data of environmentally friendly big data, including various regions air quality
Index (AQI values) PM2.5 and PM10 contents and enterprise's electricity meter information etc., belong to pioneering in the whole nation, weight are provided for science pollution treatment
The technical assistance wanted, and using power information acquisition system to support, realize that environmental protection supervise and examine positioning, enterprise are slacked off regularly, boosting ring
Protect enterprise governance.
2nd, the present invention is with BP neural network model algorithm prediction enterprise future electricity consumption situation, rate of accuracy reached 98.55%,
Data reference is provided for company's good service and risk profile, and the following air quality index of prediction, rate of accuracy reached can be analyzed
99.59%, provide strong support to pollute the formulation of management and control measures.
3rd, the present invention prediction enterprise future electricity consumption situation, shares environment protection digital, with reference to various regions air matter by model analysis
Volume index and industrial electro meter information have accurately reflected the reasonability of urban industrial structure, power-assisted company good service hair
Exhibition, data reference is provided for company's good service and risk profile.
4th, the present invention has good ductility, both Data Share System can be established with the administration for industry and commerce, by enterprise's reference
Bring into shared data bank, formulate client's degrees of comparison, data supporting is carried out for tariff recovery and good service, can also be with
Petrochina shared data.The charged stake charge volume of electric automobile can be monitored using electric extraction system, by contrasting oil, electricity accounts for
Than grasping electric automobile in time and promoting occupation rate, should be widely promoted and use.
Embodiment
The present invention program is further elaborated below by specific embodiment.
As shown in Figures 1 to 7, the prevention and control of air pollution electricity consumption data analysis platform shared based on big data, including electricity consumption
Information acquisition unit 1, transport module 2 and system main website 3, power information collecting unit 1 include the measuring terminal of installation at the scene
Equipment 4 and apparatus for acquiring and storing 5, system main website 3 include server group 6 and data analysis center 7, and system main website 3 is with using telecommunications
Breath collecting unit 1 is connected by transport module 2, and data analysis center includes shared point of the big data based on BP neural network
Analyse model 8.Server group includes database server, disk array, application server, front server, interface server, work
Stand, the network equipment of global positioning system (GPS) clock, fire wall and correlation, for providing net for data analysis center
The series of computation machine services such as network, interface, Database vendors, sequential calculating.Measuring terminal equipment includes intelligent electric meter, and collection is deposited
Storage equipment 5 includes data collecting card 9 and the data concentrator 10, the hierarchical memory 11 that are connected respectively with data collecting card 9, if
Dry intelligent electric meter is connected in star distribution with a data collecting card.Transport module 2 is including short range measuring terminal network 12 and far
Journey wireless network 13, short range measuring terminal network are transmitted using electric force carrier transmission or WIFI, remote wireless network using it is long away from
Transmitted from point-to-point type.Remote wireless network includes transceiving radio station, some apparatus for acquiring and storing in the distribution of ordinate net with
One transceiving radio station connection.Data analysis center 7 connects digital-scroll technique unit 14, and digital-scroll technique unit 14 connects respectively
Serial communication interface 15, human-machine interface unit 16.Digital-scroll technique unit includes data setting panel and LED display panel, data
Setting panel includes several data input buttons and order button.In the specific implementation, the data in data setting panel are defeated
Enter button, including ' 0 ' to ' 9 ' ten Arabic numerals enter key, order button include alarm settings, shut down setting, return,
Confirmation, machine open/close, automatic detection etc..The storage core of apparatus for acquiring and storing is hierarchical memory, and this is carried out according to content
The memory CAM of classification, this is a kind of special storage array RAM, and its groundwork mechanism is exactly by an input data
With all data item for being stored in CAM automatically simultaneously compared with, differentiate the data stored in the input data item and CAM
Whether match, and match information corresponding to exporting the data item, the data class that is particularly suitable for use in is various, calls complicated feelings
Shape, this kind of memory can make data concentrator carry out the accurate of data according to the data category information to be recorded or called
Extraction, it is convenient and swift.
The power information collecting device of prevention and control of air pollution electricity consumption data analysis platform of the present invention is the use to power consumer
The system that power information is acquired, handles and monitored, the automatic of the power informations such as user's electricity, load, voltage, electric current can be achieved
Collection, mainly by the power information acquisition monitoring of the heavily contaminated type industrial and mining enterprises to management and control, the change of analysis electric power data with
The relation of atmosphere pollution change, and the electric data index to collect, are analyzed by data model, are drawn dirty with air
The air quality index prediction of dye preventing and treating correlation, to carry out specific management and control of the air contaminant treatment proposition to management and control enterprise in next step
Scheme,
So the present invention in realization principle, it is necessary to first carry out contrast opinion to industry watt-hour meter and the internal relation of atmosphere pollution
Card, could then carry out data target forecast analysis, production Methods curve and analytical plan on the basis of this internal relation, under
Face by the city of A, B two in 2017 3, the industry watt-hour meter gathered data in April based on, carry out analysis contrast with air quality.
Due to air contaminant treatment in 2017, the situation is tense, starts to increase pollution management and control dynamics from early March, polluting weather is pre-
During alert generation, start corresponding prediction scheme, close down polluted enterprises, it is pointedly right by power information collecting unit between 3, April
She Qi contaminating enterprises carry out information gathering, daily from 8 points of morning to 22 points at night, run-down task per hour, for adopting
The new benefit of collecting system is copied data and pushed again.Fig. 4 shows 3-4 month industry watt-hour meters and air quality relation.From Fig. 4 (I),
- 5 days on the 3rd March, A cities start heavily contaminated weather yellow early warning, and industry watt-hour meter gently declines, electricity when weather was excellent than March 2
Reduce about 740,000 kilowatt hours (about 2.46%).On March -28 on the 17th, during starting blue early warning, city's industry watt-hour meter was than 16 days
Amount declines about 1,850,000 kilowatt hours (about 6.2%).On April -4 on the 3rd started heavily contaminated weather orange warning, city's industry on April 3
Electricity ratio electricity on the 2nd declines 1,530,000 kilowatt hours (about 4.4%).From Fig. 4 (II), B cities industry watt-hour meter and air quality
Relation.- 5 days on the 3rd March, B cities start heavily contaminated weather yellow early warning.4, city's industry watt-hour meter increased by 2,850,000 kilowatt hours (about
15%).On March -18 on the 17th, during starting blue early warning, than 16 days electricity of city's industry watt-hour meter decline about 70,000 kilowatt hours.4
On the moon -4 on the 3rd started heavily contaminated weather orange warning, and city's electricity of industrial electricity ratio 2 days on April 3 declines 1,220,000 kilowatt hours (about
10.2%).Can be to air with the strict management and control of preliminary judgement industry watt-hour meter from industrial electro spirogram and air quality graph of a relation
Performance figure causes actively impact, and the Some Enterprises in A, B city have implemented management and control measures within this period, also can be to atmosphere pollution
Improvement causes actively impact, moreover, the data that measuring terminal equipment collects can also carry out more targeted monitoring analysis,
It can such as determine major polluting factories' electricity and air quality relation by monitoring management and control enterprise of B cities electricity, further sentence
Duan Gai cities management and control measures implementation of conditions.From figure 5 it can be seen that on March -5 on the 3rd, during B cities start heavily contaminated weather yellow early warning, pipe
Control enterprise electricity gradually reduces.Industrial electricity ratio reduced by 500,000 kilowatt hours (about 4%) on 3rd on 5th.Comparison diagram 5 and Fig. 4 (II).March
- 18 days on the 17th and -4 days on the 3rd April, blue early warning and orange warning is respectively started in B cities, city's management and control enterprise electricity and air matter
The variation tendency and industry watt-hour meter and Air Quality Change Trend of amount are basically identical.Thus it can determine whether that city's management and control measures fall substantially
Actual arrival position.In addition, by monitoring management and control business electrical amount, the on-line monitoring to its production status can be achieved.
At the same time, find according to the study during polluting weather, PM2.5 values and air quality index (AQI) are into linear positive
Relation, and PM2.5 is influenceed by the pollutant emission of She Qi contaminating enterprises.To monitor enterprise pollution thing emission behaviour, environmental administration pair
Monitoring device is installed by contaminating enterprises, but still has enterprise to forge emissions data., can be by monitoring enterprise to avoid the above situation
Power consumption strengthens supervision.Analysis with reference to warning grade is comprehensive description, PM2.5 is quantitative analysis.In Fig. 6, y-axis
Same day management and control enterprise electricity is that B cities pollute management and control enterprise's same day electricity and proxima luce (prox. luc) electricity ratio.As can be seen from Fig. 6, March with
Come, the PM2.5 values in B cities are with polluting the same day electricity of management and control enterprise into positive correlation.So adopted by power information collecting device
Ji Sheqi contaminating enterprises electricity meter information, reflects its production status, assists environmental administration's control atmosphere pollution, can turn into and use telecommunications
Cease an important application of acquisition system.Can region-by-region, at times monitoring management and control enterprise and industrial electro meter letter by electric extraction system
Breath, while the multi dimensional analysis of enterprise's electricity meter data such as information and air quality can be realized.
It is determined that positive connection existing for industry watt-hour meter and air quality, and after collecting a large amount of power information data,
The big data for resting against data analysis center of the present invention shares analysis model, shared based on electric power with environmentally friendly big data
Forecast analysis to air quality, by the shared of environmental protection and electric power data information, with air quality index (AQI), PM2.5,
Management and control enterprise electricity builds single hidden layer BP neural network model, deploys forecast analysis to air quality, realize as data source
Air quality high-accuracy prediction, for it is next pollution management and control measures formulation strong data supporting is provided, for produce with
Life provides reference.
Big data shares analysis model based on BP neural network to realize, BP (Back Propagation) is calculated
Method is also known as error backpropagation algorithm, is a kind of learning algorithm of supervised in artificial neural network, has well non-
Linear Mapping ability, adaptive learning ability and stronger generalization ability, precision of prediction is high, algorithmic stability, is at present using more
Extensive algorithm.BP neural network used in this programme is divided into input layer, hidden layer and output layer, and adjacent layer neuron is entirely mutual
Even, it is connectionless with layer neuron, as shown in Figure 3.There is research to represent, the BP networks for having a hidden layer can approach any one and close
Continuous function in section.This programme uses implicit number of layers as 1.Its model training process by input signal forward-propagating and
The backpropagation composition of error signal.During information forward-propagating, input information is successively handled through hidden layer neuron
And it is transmitted to output layer.Compare the error between the actual value of output layer and desired value, error is continuous according to the direction of reduction
Output layer is changed to hidden layer, the connection weight and threshold value of amendment hidden layer to input layer, through " forward direction, which calculates, to be exported-reversely pass
Broadcast error " process iterate, until error is down in tolerance interval.The input pointer of neutral net is that have with predicted value
The element of close relation.Research represents that the discharge of dust, industrial waste gas and vehicle exhaust causes serious pollution to air.
Thus relation factor SO2, NO2, PM2.5 of predicted value, PM10, air quality index and management and control enterprise electricity etc. are judged.But
It is for a neutral net, is not that input pointer is The more the better, it is more model to be made to be more prone to plan on the contrary
Close or make the training time excessively very long.Consider the factor of each side, the input pointer of this programme selection is history
PM2.5, air quality index and management and control enterprise electricity, output are to predict the air quality index of day.
We are analyzed by taking B cities as an example below, comprehensive based on management and control enterprise of the B cities electric quantity data collected
Air quality index data, by B cities in January, 2017 to the historical data analysis during March, the sky of prediction B city's early Aprils
Gas quality condition:
During 1-3 months in 2017, the air quality index (AQI) in B cities is by air pollution degree and Air Quality
Classification represents, is suitable for representing the short-term Air Quality and variation tendency in city.PM2.5It is adapted to surrounding air hollow pneumatic
Mechanics equivalent diameter is less than or equal to 2.5 microns of particulate matter.Management and control enterprise electricity be according to different air regimes need to perform stopping production/
Enterprise's electricity of limited production, the big data in embodiment share analysis model and realize that step is as follows:
1) achievement data inputs:With environmental protection and electric power shared data, air quality index (AQI), PM2.5, management and control enterprise
Electricity is data source, carries out the forecast analysis of air quality index.
2) achievement data is handled:Due to air quality index, PM2.5, management and control enterprise electricity have different physical significances,
Dimension and the order of magnitude, therefore initial data is normalized by the method for formula (1) before network training.Data
Method for normalizing is a lot, and this programme uses minimax method, and formula (1) is as follows:In formula,For the data Jing Guo standardization, x is initial data, and xmax and xmin are maximum number in data sequence, minimum number respectively.
After data normalization processing, [- 1,1] section is in.
3) training sample is formulated:Training sample using the data in 1-3 months as BP neural network model, to 1-16 in April
Day air quality index is predicted inspection.
4) network training is carried out:Air quality index prediction is carried out using single hidden layer BP neural network model.Hidden layer nerve
First number is determined by formula (2):Wherein, the constant before a is 0-10, this experiment use a=3.
It is final to determine:Single hidden layer BP neural network Runoff Analysis model structure is 3-5-1, and hidden layer transmission function uses
Tansig, output layer transmission function use purelin, and training function uses trainlm, sets anticipation error as 0.0001, most
Big training samsara is 500 times, and by training, network has reached preferable precision of prediction.
5) prediction result is analyzed:By single hidden layer BP neural network model prediction early April air quality, as shown in Figure 7.
AQI prediction curves value essentially coincides with actual value, and prediction result error rate is shown in Table 1.
The B cities air quality index prediction result of table 1 and its comparison sheet
Thus B cities air quality index can be predicted exactly using single hidden layer BP neural network model, can from table 1
See, average error rate 0.41% (accuracy rate 99.59%).High-accuracy be predicted as city's environment protection treating and next step management and control is arranged
The formulation applied provides strong data supporting, can provide reference for life and enterprise's production.
The present invention in the specific implementation, realizes that the data between field acquisition unit and main station system are mutual using transport module
Connection, transport module include short range measuring terminal network and remote wireless network, and short range measuring terminal network is passed using power carrier
Defeated or WIFI transmission, remote wireless network are transmitted using long range point-to-point type.The data transfer mode of transport module can be certainly
By selecting, if being only that a small range carries out real-time Data Transmission, direct line transmission can be achieved, as power carrier is disobeyed
Auxiliary element is held in the palm, WIFI can also be used to transmit, form the short range measuring terminal network for being readily applicable to a small range interconnection;Remotely
Wireless network can be both wirelessly transferred using GSM, can be interconnected with various public communication networks, strong interference immunity, communication quality
Height, do not limited by distance, long range point-to-point type can also be used to transmit, transmitted by apparatus for acquiring and storing using a GSM,
For gsm wireless transmission, it is not necessary to using SIM card, greatly reduce long-term use of expense, be especially used in this programme
When realizing teledata interconnection, Remote Data Analysis center and metering collecting device can be directly realized by a manner of preferred GSM message
Live real-time interactive.
For the present invention since coming into operation, enterprise's violation productivity ratio is reduced to 2.51% from 11.04%.Pass through power information
Acquisition system monitors the family of the whole province industrial user about 400,000, and only verifying cost reduces nearly more than 200 ten thousand yuan., can using the present invention program
To establish Data Share System with the administration for industry and commerce, enterprise's reference is brought into shared data bank, client's degrees of comparison is formulated, is
Data supporting is carried out in tariff recovery and good service, can also be with petrochina shared data.It can be monitored using electric extraction system electronic
The charged stake charge volume of automobile, by contrasting oil, electricity accounting, electric automobile is grasped in time and promotes occupation rate.Further come
Say, palm APP can also be made, its information security management method borrows palm machine management mode.Checked whenever and wherever possible by mobile phone
All departments' data.
Finally illustrate, the above embodiments are merely illustrative of the technical solutions of the present invention and it is unrestricted, this area is common
Other modifications or equivalent substitution that technical staff is made to technical scheme, without departing from technical solution of the present invention
Spirit and scope, all should cover among scope of the presently claimed invention.