CN109740301A - A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network - Google Patents
A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network Download PDFInfo
- Publication number
- CN109740301A CN109740301A CN201910192902.2A CN201910192902A CN109740301A CN 109740301 A CN109740301 A CN 109740301A CN 201910192902 A CN201910192902 A CN 201910192902A CN 109740301 A CN109740301 A CN 109740301A
- Authority
- CN
- China
- Prior art keywords
- data
- carbon emission
- neural network
- carbon
- generator set
- 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
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Feedback Control In General (AREA)
Abstract
The invention discloses the accounting methods of the Gas Generator Set carbon emission amount based on BP neural network, the accounting method includes the following steps: the carbon emission concentration data for obtaining Gas Generator Set, air inflow data, combustion engine load data, steam turbine load data and status data, carbon emission qualitative data is converted by carbon emission concentration data using carbon emission concentration data and status data, unit carbon displacement data is determined by carbon emission qualitative data and air inflow data, utilize unit carbon displacement data, air inflow data, combustion engine load data and steam turbine load data are trained initial BP neural network model, obtain the BP neural network model for being calculated to Gas Generator Set carbon emission amount.The present invention can carry out screening to abnormal carbon emission data and carry out reasonable supplement to the data of rejecting, to realize the accurate and real-time accounting to Gas Generator Set carbon emission amount, carbon emission transaction is participated in for Thermal Power Generation Industry, solid data support is provided, thus the liveness for pushing carbon market to trade.
Description
Technical field
The present invention relates to carbon emission detection technique fields, specifically for, the present invention is a kind of to be based on BP neural network
Gas Generator Set carbon emission amount accounting method.
Background technique
As people are to the pay attention to day by day of environmental problem, many countries propose stringent carbon emission reduction target, for example,
2017, China started national carbon emission trade market comprehensively, and power industry is received as maximum CO2 emission source
Enter carbon emission trade system, and be in the first stage the breach of national carbon emission trade system with power industry, to wish
Total carbon emission is set to be effectively controlled.
It is calculated currently, the accounting of carbon emission amount tends to rely on emission factor, but the actual conditions thousand of Thermal Power Enterprises
Poor ten thousand are not, and the existing carbon emission amount accounting method based on emission factor is often difficult to accurately calculate the carbon emission of different Thermal Power Enterprises
Amount calculates result for example, the Gas Generator Set to different enterprises carries out carbon emission amount accounting using identical emission factor and parameter
Accuracy is poor, confidence level is poor, inaccuracy, unfair, confidence level is poor accounting result will affect the encouragement of carbon transaction
Mechanism and penalty mechanism, and then will lead to and the enterprise for playing an active part in carbon emission reduction is motivated not enough, to the enterprise for failing to carry out carbon emission reduction
The problems such as restraining force is small, so the accounting method of existing carbon discharge capacity hinders the liveness of carbon market transaction.In addition, due to existing
There is accounting method that can only be calculated after discharge, so the real-time of existing accounting method is also poor, cannot achieve
Carbon discharge capacity is calculated in time, carbon discharge capacity can not be reported and submitted to calculate result in time.
Therefore, how to realize the accurate and timely accounting to Gas Generator Set carbon discharge capacity, become those skilled in the art
Member's technical problem urgently to be resolved and the emphasis studied always.
Summary of the invention
To solve, accuracy existing for the existing Gas Generator Set carbon emission amount accounting method based on emission factor is poor, Wu Faji
When the problems such as reporting and submitting, present invention innovation provides a kind of accounting method of Gas Generator Set carbon emission amount based on BP neural network,
Specially a kind of Gas Generator Set carbon emission abnormal data screening mechanism based on BP neural network, it is issuable to Gas Generator Set
Carbon emission abnormal data monitor value carries out screening, guarantees that carbon emission amount calculates the real effectiveness of result；Moreover, the present invention will have
Help accurately and completely obtain Thermal Power Generation Industry carbon emission real time data, be provided for Thermal Power Generation Industry participation carbon emission transaction solid
Data support.
To realize the above-mentioned technical purpose, the Gas Generator Set carbon emission amount based on BP neural network that the invention discloses a kind of
Accounting method, the accounting method include the following steps；
Step 1, carbon emission concentration data, air inflow data, combustion engine load data, the steam turbine load number of Gas Generator Set are obtained
According to and status data, wherein gas flowing pressure data, flue gas when the status data includes measurement it is average per hour
Gas temperature data when product data on flows and measurement；
Step 2, it is converted the carbon emission concentration data to using the carbon emission concentration data and the status data
Carbon emission qualitative data；
Step 3, unit carbon displacement data, and unit are determined by the carbon emission qualitative data and the air inflow data
Carbon discharge capacity is the ratio of the carbon emission quality and air inflow under the same time；
Step 4, by the unit carbon displacement data, the air inflow data, the combustion engine load data and the steam turbine
Load data is trained initial BP neural network model collectively as sample data, using the sample data, thus
BP neural network model after to training；Wherein, the air inflow data, the combustion engine load data and the steam turbine are born
Input data of the lotus data as BP neural network model, and enabling the output of BP neural network model is unit carbon discharge capacity amplitude；
Step 5, carbon emission amount accounting is carried out to Gas Generator Set using the BP neural network model after the training.
Based on above-mentioned technical solution, the present invention can obtain being used for core according to Gas Generator Set carbon emission relevant historical data
The BP neural network model of carbon emission is calculated, thus realize accurate and Gas Generator Set carbon emission data are calculated in time, with
Preferably solve problems of the existing technology.
Further, step 5 includes the following steps；
Step 5-1 determines the unit carbon for changing with air inflow and changing using the BP neural network model after the training
The forecast interval of flow oscillation amplitude；
Step 5-2, in the identical situation of air inflow, by the unit carbon displacement data calculated with pending carbon emission amount with
The forecast interval of the unit carbon flow oscillation amplitude is compared, and the pending carbon emission amount except forecast interval is calculated
Unit carbon displacement data is rejected.
Based on above-mentioned improved technical solution, the present invention provides a kind of carbon emission amount exception number based on BP neural network
According to screening mechanism, Gas Generator Set carbon emission abnormal data can be effectively removed, guarantee the standard of carbon emission amount data
True property.
Further, in step 5-1, the pre- of the unit carbon flow oscillation amplitude changed with air inflow variation is utilized
It surveys section and determines that carbon emission reference line, the carbon emission reference line are the unit carbon discharge curve for changing with air inflow and changing；
In step 5-2, the corresponding air inlet of unit carbon displacement data that the pending carbon emission amount being removed is calculated is obtained
Data are measured, and using the corresponding carbon displacement data on reference line of the air inflow data as revised carbon displacement data.
Based on above-mentioned improved technical solution, on the basis of guaranteeing the accuracy of carbon emission data, the present invention can also
Gas Generator Set carbon emission abnormal data is supplemented, ensure that the integrality of carbon emission amount data, accurately to gas engine
Group carbon emission amount is calculated.
Further, in step 2, carbon emission mass number is converted by the carbon emission concentration data in the following way
According to；
Wherein,Indicate carbon emission quality, MWCO2Indicate CO2Molecular weight, P indicate measurement when gas flowing pressure, Q
Indicate the average product flow per hour of flue gas, CCO2Indicating carbon emission concentration, T indicates gas temperature when measurement,
0.08314 indicates gas constant, and 273.15 indicate temperature inversion coefficient.
It further, further include that carbon emission qualitative data is converted to the carbon emission matter after moisture correction process in step 2
The step of measuring data；
In step 3, the carbon emission qualitative data unit of account carbon after air inflow data and the moisture correction process is utilized
Discharge capacity.
Further, in step 2, moisture correction process is carried out to carbon emission qualitative data in the following way；
Wherein,Carbon emission quality after indicating moisture correction process, %H2O indicates the water in flue gas per hour
Divide percentage,Carbon emission quality before indicating moisture correction process.
Based on above-mentioned improved technical solution, by directly handling carbon emission butt data, the present invention can also be further
Improve the accuracy that carbon emission amount is calculated in ground.
Further, the BP neural network model in step 4, after being trained in the following way；
Step 4-1 initializes each neuron in the hidden layer of initial BP neural network model；
Air inflow data, combustion engine load data and steam turbine load data in sample data is fed through BP mind by step 4-2
Input layer through network model, then calculate the output result that provides of output layer of BP neural network model in sample data
Error between unit carbon discharge capacity；
Step 4-3 thens follow the steps 4-4 if the error is greater than or equal to preset accuracy value；If the error
Less than preset accuracy value, 4-5 is thened follow the steps；
Step 4-4 corrects the network weight and threshold value of each neuron in the hidden layer of current BP neural network model, then returns
Return step 4-2；
Step 4-5, using current BP neural network model as the BP neural network model after training.
Further, the initial BP neural network model be 3 × k × 1 three layers of BP neural network model, i.e., described three
The input layer number of layer BP neural network model is 3, output layer number is 1, hidden layer quantity is k, and each hidden layer has l
Neuron；Wherein, k >=1,3≤l≤12.
Further, three layers of BP neural network model has a hidden layer, is determined as follows hidden layer tool
Some neuron numbers；
Wherein, l indicates that the neuron number that hidden layer has, n indicate that the neuron number of input layer, m indicate output layer
Neuron number, a indicate the constant between [1,10].
Further, in step 1, from fume continuous monitoring system obtain Gas Generator Set carbon emission concentration data, into
Tolerance data, combustion engine load data, steam turbine load data and status data, and step 1 further includes when will arrange mouth calibration to be not zero
Data and in row mouth fault time section in data reject the step of.
The invention has the benefit that the present invention can the carbon emission data to Gas Generator Set accurately checked, including
Screening is carried out to abnormal carbon emission data and reasonable supplement is carried out to the data of rejecting, to realize to Gas Generator Set carbon emission amount
It is accurate and calculate in real time, guarantee the accurate integrality of Thermal Power Generation Industry carbon emission data, and then realize to playing an active part in carbon emission reduction
Enterprise encourage and carry out to the enterprise for failing to carry out carbon emission reduction the effect of strength constraint, participates in carbon emission for Thermal Power Generation Industry and hands over
Solid data are easily provided to support, thus the liveness for pushing carbon market to trade, and then thoroughly solve of the existing technology all
More problems.
Detailed description of the invention
Fig. 1 is the flow diagram of the accounting method of the Gas Generator Set carbon emission amount based on BP neural network.
Fig. 2 is the forecast interval schematic diagram for the unit carbon flow oscillation amplitude for changing with air inflow and changing.
Fig. 3 is that the unit carbon discharge capacity amplitude fluctuations state and carbon emission reference line for changing with air inflow and changing are illustrated
Figure.
Fig. 4 is the composition schematic diagram of BP neural network application model.
Specific embodiment
With reference to the accompanying drawings of the specification to a kind of Gas Generator Set carbon emission amount based on BP neural network provided by the invention
Accounting method carry out detailed explanation and illustration.
As shown in Figures 1 to 3, the accounting for the Gas Generator Set carbon emission amount based on BP neural network that present embodiment discloses a kind of
Method, the accounting method of specially a kind of Gas Generator Set CO2 emissions based on BP neural network；It, should in the present embodiment
Accounting method includes the following steps.
Step 1, carbon emission concentration data, air inflow data, combustion engine load data, the steam turbine load number of Gas Generator Set are obtained
According to and status data, wherein status data include measurement when gas flowing pressure data, flue gas average product stream per hour
Measure gas temperature data when data and measurement.Specifically, the present invention is for being monitored flue gas continuous emission, from flue gas
Carbon emission concentration data, air inflow data, combustion engine load data, the steam turbine load number of Gas Generator Set are obtained in continuous monitor system
It accordingly and status data, can be by mounted for pollutant emissions such as Thermal Power Generation Industry dust, sulfide, nitride when implementation
On-line continuous monitoring system or maturation flue gas discharge continuous monitoring system (continuous emission
Monitoring system of flue gas, CEMS) above-mentioned data are obtained, moreover, this step further includes that will arrange mouth calibration not
The step of data when being zero and the data in row's mouth fault time section are rejected, to improve the accurate of carbon emission amount accounting
Property.
Step 2, carbon emission mass number is converted by carbon emission concentration data using carbon emission concentration data and status data
According to, for example pass through the continuous monitoring model of carbon emission formerly constructed and convert carbon emission for the carbon emission concentration data continuously monitored
Qualitative data.
Specifically, in step 2, the present embodiment converts carbon emission mass number for carbon emission concentration data in the following way
According to；
Wherein,Indicate that carbon emission quality, unit can be ton/second, MWCO2Indicate CO2Molecular weight, be 44.01 × 10-3, gas flowing pressure, unit when P expression measures can be kPa, and Q indicates that the flow of average product per hour of flue gas, unit can
For a cube meter per second, CCO2Indicate carbon emission concentration (CO2%), T indicate measurement when gas temperature, unit be degree Celsius,
0.08314 indicates gas constant, and 273.15 indicate temperature inversion coefficient.
In addition, further including converting carbon emission qualitative data in step 2 to further increase the accuracy that the present invention calculates
The step of for carbon emission qualitative data after moisture correction process, to realize that the butt to carbon dioxide measures.
Moisture correction process is carried out to carbon emission qualitative data especially by such as under type in step 2:
Wherein, above-mentionedCarbon emission quality (ton/second) after indicating moisture correction process, %H2O is indicated per hour
Moisture percentage in flue gas,Carbon emission quality (ton/second) before indicating moisture correction process.
Step 3, determine unit carbon displacement data by carbon emission qualitative data and air inflow data, the present embodiment utilize into
Carbon emission qualitative data unit of account carbon discharge capacity after tolerance data and moisture correction process, wherein unit carbon discharge capacity is same
The ratio of carbon emission quality and air inflow under one time, the CO2 emissions under unit carbon discharge capacity, that is, unit air inflow.
Step 4, and random fluctuation relatively stable using unit carbon discharge capacity, by air inflow variation influenced the features such as, this reality
Example is applied by unit carbon displacement data, air inflow data, combustion engine load data and steam turbine load data collectively as sample data, so
Initial BP neural network model (also known as " reverse neural network ") is trained using sample data afterwards, thus after being trained
BP neural network model；Wherein, using air inflow data, combustion engine load data and steam turbine load data as BP neural network
The input data of model, and enabling the output of BP neural network model is unit carbon discharge capacity amplitude, i.e., is defeated with unit carbon emission amount
Out；BP neural network model is made of input layer, hidden layer and output layer, in this step, after being trained in the following way
BP neural network model.
Step 4-1 initializes each neuron in the hidden layer of initial BP neural network model, the initial BP of the present embodiment
Neural network model is three layers of BP neural network model of 3 × k × 1, i.e., the input layer number of three layers BP neural network model is
3, output layer number is 1, hidden layer quantity is k, and each hidden layer has l neuron；Wherein, k >=1,3≤l≤12.Hidden layer
K can be one or more layers, have the neural network of a hidden layer, as long as hidden node is enough, so that it may approach with arbitrary accuracy
One nonlinear function, the present embodiment establish prediction mould using the BP network of three layers of multiple input single output containing a hidden layer
Type, in network design process, the determination of hidden nodes is particularly significant, this is because: hidden neuron number is excessive, meeting
It increases network query function amount and is easy to produce overfitting problem；Neuron number is very few, then will affect network performance, is not achieved pre-
Phase effect, the present embodiment are final to determine that three layers of BP neural network model of the present embodiment have one by a large amount of test and demonstration
A hidden layer is determined as follows the neuron number (i.e. number of nodes) that the hidden layer has；
Wherein, l indicates that the neuron number that hidden layer has, n indicate that the neuron number of input layer, m indicate output layer
Neuron number, a indicate the constant between [1,10].As shown in figure 4, " Gas inflow " indicates air inflow, " Gas
Turbine load " indicates that combustion engine load, " Steam turbine load " indicate that steam turbine load, " Input layer " indicate
Input layer, " Implicit layer " indicate hidden layer, and " Output layer " indicates that output layer, " Amplitude " indicate carbon row
Measure amplitude.
In addition, the BP neural network of training is using Sigmoid differentiable function and linear function as network in the present embodiment
Excitation function, and excitation function of the S type tangent function tansig as hidden neuron is selected, due to the output normalizing of network
Into [- 1,1] range, therefore prediction model chooses excitation function of the S type logarithmic function tansig as output layer neuron.
Air inflow data, combustion engine load data and steam turbine load data in sample data is fed through BP mind by step 4-2
Input layer through network model, the present embodiment to the carbon emission related data of Gas Generator Set 9E as sample data, for example, choosing
The monitoring data in Gas Generator Set June, and shutdown status data (" combustion engine load < 0.1MW or combustion engine air inflow are negative ") is rejected,
Then the mistake between the output layer output result provided of BP neural network model and the unit carbon discharge capacity of sample data is calculated again
Difference.When it is implemented, sample data can be divided into training set and test set by the present embodiment, can randomly select in each segment
Carbon emission monitoring actual value is as test set, to obtain screening a series of forecast interval of abnormal carbon emission data.
Step 4-3 thens follow the steps 4-4 if error is greater than or equal to preset accuracy value；If error is less than default
Accuracy value, then follow the steps 4-5.
Step 4-4 corrects the network weight and threshold value of each neuron in the hidden layer of current BP neural network model, to make
The error function of BP neural network declines along negative gradient direction, result is made to approach desired output, then return step 4-2.
Step 4-5, using current BP neural network model as the BP neural network model after training.
Step 5, carbon emission amount accounting, this reality are carried out to Gas Generator Set using the BP neural network model after above-mentioned training
The step 5 for applying example includes the following steps 5-1 and step 5-2.
Step 5-1 determines the unit carbon discharge capacity for changing with air inflow and changing using the BP neural network model after training
The forecast interval of fluctuation amplitude, the present embodiment using combustion engine air inflow as abscissa, with unit carbon discharge capacity (carbon emission amount/combustion engine into
Tolerance) it is that ordinate makees curve graph, and deletes the abnormal point of unit carbon discharge capacity>100 or<10, due to air inflow<12kNm3/h
When it is unstable, therefore abscissa air inflow is since 12, in addition, when it is implemented, due to air inflow < 40kNm3/ h data mistake
It is few, air inflow < 40kNm in supplement 1,2,3,4 month3The point of/h, the present embodiment obtain air inflow < 40kNm3The data point of/h is total
7757, air inflow > 40kNm3The data point of/h totally 9121, by air inflow 12kNm3/ h~48kNm3/ h points are 36 length
For 1 section, point set mean value in 36 segments is calculated separately, obtains the carbon emission reference line in 36 sections；Carbon will be surveyed
The mean value that emissions data subtracts corresponding section obtains carbon emission actual measurement fluctuation amplitude, as shown in Figure 2,3, wherein
" CO2Emissions/Gas inflow " indicates the ratio of the carbon emission quality and air inflow under the same time, " Gas
Inflow " indicates air inflow, and " Base line " indicates that carbon emission reference line, " Upper boundary " indicate coboundary,
" Lower boundary " indicates that lower boundary, " Text point " indicate the unit carbon discharge capacity test that pending carbon emission amount is calculated
Point, " Raw data " indicate to change with air inflow and the unit carbon flow oscillation amplitude of variation；It is utilized in this step with air inlet
Amount variation and the forecast interval of unit carbon flow oscillation amplitude changed determines carbon emission reference line, carbon emission reference line be with into
The unit carbon discharge curve of tolerance variation and variation.As shown in figure 3, Gas Generator Set is in and opens when gas inlet amount is less than 40
Dynamic or stopping is with state, and air inflow fluctuates larger at any time during this period, since combustion of natural gas deficiency leads to unit carbon discharge capacity
It reduces, when air inflow is greater than 40, Gas Generator Set enters steady operational status, and the full combustion of rate of load condensate > 75%, natural gas makes
Unit carbon discharge capacity is obtained to increase.BP neural network input layer number is 3, and respectively combustion engine load, air inflow and steam turbine load are defeated
It is unit carbon emission amplitude that node layer number, which is 1, out, and hidden nodes 10, the present embodiment is using MATLAB neural network tool
Case 8.2 is trained 16878 groups of sample datas it is found that when setting is trained for 10000, the optimal fitting error of curve convergence
It is 1.67%.
Step 5-2, in the identical situation of air inflow, by the unit carbon displacement data calculated with pending carbon emission amount with
The forecast interval (data) of unit carbon flow oscillation amplitude is compared, and the pending carbon emission amount except forecast interval is calculated
Unit carbon displacement data reject；Exceptional data point in Fig. 2 is 3, and as the technical solution advanced optimized, this implementation
Example obtains the corresponding air inflow number of unit carbon displacement data that the pending carbon emission amount being removed is calculated in step 5-2
According to, and using the corresponding carbon displacement data on reference line of the air inflow data (can be multiple data) as revised carbon discharge capacity
Data.
In the description of this specification, reference term " the present embodiment ", " one embodiment ", " some embodiments ", " show
The description of example ", " specific example " or " some examples " etc. mean specific features described in conjunction with this embodiment or example, structure,
Material or feature are included at least one embodiment or example of the invention.In the present specification, above-mentioned term is shown
The statement of meaning property is necessarily directed to identical embodiment or example.Moreover, specific features, structure, material or the spy of description
Point may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, without conflicting with each other,
Those skilled in the art can be by different embodiments or examples described in this specification and different embodiments or examples
Feature is combined.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modification, equivalent replacement and simple modifications etc., should all be included in the protection scope of the present invention in content.
Claims (10)
1. a kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network, it is characterised in that: the accounting method packet
Include following steps；
Step 1, obtain the carbon emission concentration data of Gas Generator Set, air inflow data, combustion engine load data, steam turbine load data and
Status data, wherein the average product stream per hour of gas flowing pressure data, flue gas when the status data includes measurement
Measure gas temperature data when data and measurement；
Step 2, carbon is converted by the carbon emission concentration data using the carbon emission concentration data and the status data to arrange
Put qualitative data；
Step 3, unit carbon displacement data is determined by the carbon emission qualitative data and the air inflow data, and unit carbon is arranged
Amount is the ratio of the carbon emission quality and air inflow under the same time；
Step 4, by the unit carbon displacement data, the air inflow data, the combustion engine load data and the steam turbine load
Data are trained initial BP neural network model collectively as sample data, using the sample data, to be instructed
BP neural network model after white silk；Wherein, by the air inflow data, the combustion engine load data and the steam turbine load number
According to the input data as BP neural network model, and enabling the output of BP neural network model is unit carbon discharge capacity amplitude；
Step 5, carbon emission amount accounting is carried out to Gas Generator Set using the BP neural network model after the training.
2. the accounting method of the Gas Generator Set carbon emission amount according to claim 1 based on BP neural network, feature exist
In: step 5 includes the following steps；
Step 5-1 determines the unit carbon discharge capacity for changing with air inflow and changing using the BP neural network model after the training
The forecast interval of fluctuation amplitude；
Step 5-2, in the identical situation of air inflow, by the unit carbon displacement data calculated with pending carbon emission amount with it is described
The forecast interval of unit carbon flow oscillation amplitude is compared, the unit that the pending carbon emission amount except forecast interval is calculated
Carbon displacement data is rejected.
3. the accounting method of the Gas Generator Set carbon emission amount according to claim 2 based on BP neural network, feature exist
In:
In step 5-1, the forecast interval of the unit carbon flow oscillation amplitude changed using described change with air inflow determines carbon
Reference line is discharged, the carbon emission reference line is the unit carbon discharge curve for changing with air inflow and changing；
In step 5-2, the corresponding air inflow number of unit carbon displacement data that the pending carbon emission amount being removed is calculated is obtained
According to, and using the corresponding carbon displacement data on reference line of the air inflow data as revised carbon displacement data.
4. according to claim 1 to the Gas Generator Set carbon emission amount described in any claim in 3 based on BP neural network
Accounting method, it is characterised in that: in step 2, convert carbon emission quality for the carbon emission concentration data in the following way
Data；
Wherein,Indicate carbon emission quality, MWCO2Indicate CO2Molecular weight, P indicate measurement when gas flowing pressure, Q indicate
The average product flow per hour of flue gas, CCO2Indicate carbon emission concentration, T indicates gas temperature when measurement, 0.08314 table
Show gas constant, 273.15 indicate temperature inversion coefficient.
5. the accounting method of the Gas Generator Set carbon emission amount according to claim 4 based on BP neural network, feature exist
In:
In step 2, further include the steps that being converted to carbon emission qualitative data into the carbon emission qualitative data after moisture correction process；
In step 3, the carbon emission qualitative data unit of account carbon discharge capacity after air inflow data and the moisture correction process is utilized.
6. the accounting method of the Gas Generator Set carbon emission amount according to claim 5 based on BP neural network, feature exist
In: in step 2, moisture correction process is carried out to carbon emission qualitative data in the following way；
Wherein,Carbon emission quality after indicating moisture correction process, %H2O indicates the moisture hundred in flue gas per hour
Divide ratio,Carbon emission quality before indicating moisture correction process.
7. the accounting method of the Gas Generator Set carbon emission amount according to claim 1 or 6 based on BP neural network, feature
It is: the BP neural network model in step 4, after being trained in the following way；
Step 4-1 initializes each neuron in the hidden layer of initial BP neural network model；
Air inflow data, combustion engine load data and steam turbine load data in sample data is fed through BP nerve net by step 4-2
Then the input layer of network model calculates output result and the unit in sample data that the output layer of BP neural network model provides
Error between carbon discharge capacity；
Step 4-3 thens follow the steps 4-4 if the error is greater than or equal to preset accuracy value；If the error is less than
Preset accuracy value, thens follow the steps 4-5；
Step 4-4 corrects the network weight and threshold value of each neuron in the hidden layer of current BP neural network model, then returns to step
Rapid 4-2；
Step 4-5, using current BP neural network model as the BP neural network model after training.
8. the accounting method of the Gas Generator Set carbon emission amount according to claim 7 based on BP neural network, feature exist
In: the initial BP neural network model is three layers of BP neural network model of 3 × k × 1, i.e., described three layers of BP neural network mould
The input layer number of type is 3, output layer number is 1, hidden layer quantity is k, and each hidden layer has l neuron；Wherein, k
>=1,3≤l≤12.
9. the accounting method of the Gas Generator Set carbon emission amount according to claim 8 based on BP neural network, feature exist
In: three layers of BP neural network model has a hidden layer, is determined as follows the neuron number that the hidden layer has；
Wherein, l indicates that the neuron number that hidden layer has, n indicate that the neuron number of input layer, m indicate the nerve of output layer
First number, a indicate the constant between [1,10].
10. the accounting method of the Gas Generator Set carbon emission amount according to claim 1 or described in 9 based on BP neural network, special
Sign is:
In step 1, carbon emission concentration data, the air inflow data, combustion engine of Gas Generator Set are obtained from fume continuous monitoring system
Load data, steam turbine load data and status data, and step 1 further includes data and the row of being in that will be arranged when mouth calibration is not zero
The step of data in mouth fault time section are rejected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910192902.2A CN109740301A (en) | 2019-03-14 | 2019-03-14 | A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910192902.2A CN109740301A (en) | 2019-03-14 | 2019-03-14 | A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109740301A true CN109740301A (en) | 2019-05-10 |
Family
ID=66370343
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910192902.2A Pending CN109740301A (en) | 2019-03-14 | 2019-03-14 | A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109740301A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110544015A (en) * | 2019-08-09 | 2019-12-06 | 内蒙古自治区计量测试研究院 | Enterprise carbon data or carbon asset intelligent management and control platform based on big data analysis |
| CN112785843A (en) * | 2020-12-26 | 2021-05-11 | 清华四川能源互联网研究院 | Carbon emission monitoring method, device, server and computer-readable storage medium |
| CN113640466A (en) * | 2021-08-03 | 2021-11-12 | 广东电网有限责任公司 | Carbon emission intensity measuring method, equipment and medium |
| CN115099522A (en) * | 2022-07-19 | 2022-09-23 | 东南大学溧阳研究院 | Active and reactive carbon emission prediction method for special transformer user based on BP neural network |
| CN115147012A (en) * | 2022-08-31 | 2022-10-04 | 南方电网数字电网研究院有限公司 | Carbon emission accounting method and device based on neural network model |
| CN115496286A (en) * | 2022-09-26 | 2022-12-20 | 重庆德宜高大数据科技有限公司 | Neural network carbon emission prediction method based on big data environment and application |
| CN115759450A (en) * | 2022-11-28 | 2023-03-07 | 广东省科学院江门产业技术研究院有限公司 | Carbon emission accounting early warning analysis system based on big data |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975799A (en) * | 2016-06-01 | 2016-09-28 | 广东电网有限责任公司电力科学研究院 | Method and system for calculating carbon emissions |
| CN106529732A (en) * | 2016-11-18 | 2017-03-22 | 国网山东省电力公司电力科学研究院 | Carbon emission efficiency prediction method based on neural network and random frontier analysis |
| CN106649579A (en) * | 2016-11-17 | 2017-05-10 | 苏州航天***工程有限公司 | Time-series data cleaning method for pipe net modeling |
| CN108280047A (en) * | 2018-01-24 | 2018-07-13 | 江苏方天电力技术有限公司 | A kind of fired power generating unit carbon emission accounting method based on field monitoring data |
| CN108734359A (en) * | 2018-06-08 | 2018-11-02 | 上海电机学院 | A kind of wind power prediction data preprocessing method |
| CN108846526A (en) * | 2018-08-08 | 2018-11-20 | 华北电力大学 | A kind of CO2 emissions prediction technique |
| CN109443423A (en) * | 2018-10-10 | 2019-03-08 | 江苏方天电力技术有限公司 | A kind of Gas Generator Set carbon emission data correlation degree method of inspection based on covariance analysis |
-
2019
- 2019-03-14 CN CN201910192902.2A patent/CN109740301A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975799A (en) * | 2016-06-01 | 2016-09-28 | 广东电网有限责任公司电力科学研究院 | Method and system for calculating carbon emissions |
| CN106649579A (en) * | 2016-11-17 | 2017-05-10 | 苏州航天***工程有限公司 | Time-series data cleaning method for pipe net modeling |
| CN106529732A (en) * | 2016-11-18 | 2017-03-22 | 国网山东省电力公司电力科学研究院 | Carbon emission efficiency prediction method based on neural network and random frontier analysis |
| CN108280047A (en) * | 2018-01-24 | 2018-07-13 | 江苏方天电力技术有限公司 | A kind of fired power generating unit carbon emission accounting method based on field monitoring data |
| CN108734359A (en) * | 2018-06-08 | 2018-11-02 | 上海电机学院 | A kind of wind power prediction data preprocessing method |
| CN108846526A (en) * | 2018-08-08 | 2018-11-20 | 华北电力大学 | A kind of CO2 emissions prediction technique |
| CN109443423A (en) * | 2018-10-10 | 2019-03-08 | 江苏方天电力技术有限公司 | A kind of Gas Generator Set carbon emission data correlation degree method of inspection based on covariance analysis |
Non-Patent Citations (2)
| Title |
|---|
| LI MENG等: "Abnormal Data Screening Method of Thermal Power Based on BP Neural Network Algorithm", 《2018 2ND INTERNATIONAL CONFERENCE ON MODELING, SIMULATION AND OPTIMIZATION TECHNOLOGIES AND APPLICATIONS (MSOTA 2018)》 * |
| 刘荣昌 编著: "《曲轴滚压关键参数设计理论与数值模拟》", 28 February 2010 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110544015A (en) * | 2019-08-09 | 2019-12-06 | 内蒙古自治区计量测试研究院 | Enterprise carbon data or carbon asset intelligent management and control platform based on big data analysis |
| CN112785843A (en) * | 2020-12-26 | 2021-05-11 | 清华四川能源互联网研究院 | Carbon emission monitoring method, device, server and computer-readable storage medium |
| CN113640466A (en) * | 2021-08-03 | 2021-11-12 | 广东电网有限责任公司 | Carbon emission intensity measuring method, equipment and medium |
| CN115099522A (en) * | 2022-07-19 | 2022-09-23 | 东南大学溧阳研究院 | Active and reactive carbon emission prediction method for special transformer user based on BP neural network |
| CN115147012A (en) * | 2022-08-31 | 2022-10-04 | 南方电网数字电网研究院有限公司 | Carbon emission accounting method and device based on neural network model |
| CN115496286A (en) * | 2022-09-26 | 2022-12-20 | 重庆德宜高大数据科技有限公司 | Neural network carbon emission prediction method based on big data environment and application |
| CN115759450A (en) * | 2022-11-28 | 2023-03-07 | 广东省科学院江门产业技术研究院有限公司 | Carbon emission accounting early warning analysis system based on big data |
| CN115759450B (en) * | 2022-11-28 | 2023-10-27 | 广东省科学院江门产业技术研究院有限公司 | Carbon emission accounting early warning analysis system based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109740301A (en) | A kind of accounting method of the Gas Generator Set carbon emission amount based on BP neural network | |
| CN109740977A (en) | The evaluation method of Gas Generator Set carbon emission influence factor based on grey correlation analysis | |
| CN108280047A (en) | A kind of fired power generating unit carbon emission accounting method based on field monitoring data | |
| CN112232571B (en) | Method for predicting concentration of main pollutants in waste gas | |
| CN107301617B (en) | Method and equipment for evaluating quality of waste gas monitoring data | |
| CN114994264B (en) | Park digital monitoring platform based on block chain enterprise-level ledger | |
| CN109190967B (en) | Carbon emission accounting method and system for thermal generator set | |
| CN105590283A (en) | Examination data analysis method on the basis of fuzzy synthetic evaluation model | |
| CN109443423A (en) | A kind of Gas Generator Set carbon emission data correlation degree method of inspection based on covariance analysis | |
| CN110094251A (en) | SCR catalyst performance deterioration analysis method based on multi-model modeling at times | |
| CN109100468A (en) | A kind of the abnormal data early warning system and method for pollution sources environment automatic monitoring platform | |
| CN114169424A (en) | Discharge capacity prediction method based on k nearest neighbor regression algorithm and electricity utilization data | |
| CN109783934A (en) | A kind of mean velocity in section fitting rating method based on H-ADCP | |
| Duflo et al. | The value of discretion in the enforcement of regulation: Experimental evidence and structural estimates from environmental inspections in india | |
| CN111797892A (en) | Electric power market generator market force monitoring method based on random forest regression | |
| CN114372729A (en) | Multi-dimension-based thermoelectric enterprise carbon emission level evaluation method | |
| CN105224730A (en) | The original NO of a kind of high pressure common rail electric-controlled diesel engine 2forecasting of discharged quantity method | |
| Blair | Boundary-layer transition in accelerating flows with intense freestream turbulence: part 2—the zone of intermittent turbulence | |
| CN106446570A (en) | Conference influence assessment method based on academic big data | |
| RU2010118406A (en) | METHOD FOR CREATING MIXED CARDS FOR THE NUCLEAR REACTOR ACTIVE ZONE AND ITS APPLICATION FOR CALIBRATION OF STATIONARY MONITORING AND MEASURING INSTRUMENTS | |
| CN109655489A (en) | A kind of metering method and device of coal unit standard coal consumption | |
| CN106651094B (en) | Enterprise energy efficiency evaluation method based on equivalent energy saving | |
| CN115204453A (en) | Carbon emission prediction method, equipment and storage medium | |
| Lim et al. | Trend change in mean residual life | |
| Saramud et al. | A new approach to multi-version decision-making to improve the reliability of environmental monitoring parameters |
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: 20190510 |