CN115682010A - Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit - Google Patents
Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit Download PDFInfo
- Publication number
- CN115682010A CN115682010A CN202211377924.4A CN202211377924A CN115682010A CN 115682010 A CN115682010 A CN 115682010A CN 202211377924 A CN202211377924 A CN 202211377924A CN 115682010 A CN115682010 A CN 115682010A
- Authority
- CN
- China
- Prior art keywords
- electric
- data
- smoke concentration
- coal
- outlet smoke
- 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
Images
Landscapes
- Electrostatic Separation (AREA)
Abstract
The invention provides a neural network-based method for predicting outlet smoke concentration of a dry-type electric precipitator of a coal-fired unit, which mainly comprises the following steps: s1, obtaining outlet smoke concentration of a unit under each typical load and secondary current values of each electric chamber in each electric field of an electric precipitator in each level from a database of a distributed control system DCS of a coal-fired thermal power generating unit; s2, carrying out normalization processing on the obtained data of each load section, and dividing the data into a training set and a test set according to a certain proportion; s3, selecting a BP neural network model, taking secondary current values of all levels of electric fields as a network input layer, taking outlet smoke concentration as a network output layer, and putting data into a training set for iterative training; and S4, replacing the next load point of the data set, and repeating the steps S2 and S3. The invention can avoid the dependence on the complex mechanism process of the electric dust collector, and further can provide a calculation basis for realizing the optimization of energy consumption of the dry-type electric dust collector under the condition of meeting the environmental protection requirement.
Description
Technical Field
The invention relates to the field of energy-saving and environment-friendly operation of coal-fired units, in particular to a method for predicting the concentration of smoke dust at an outlet of a dry type electric precipitator of a coal-fired unit.
Background
In order to prevent a large amount of smoke generated after combustion of pulverized coal in a thermal power generating unit from polluting the atmospheric environment, a dry-type electric dust collector is usually arranged at an outlet of a flue at the tail of a boiler in a coal-fired power plant to reduce the emission of pollutants. Among factors influencing the efficiency of the dry-type electric dust collector, design parameters usually do not change greatly along with the change of working conditions during operation; the flue gas parameters can be changed when the unit load and the coal type are changed, so that disturbance is brought to the electric precipitation process; the operation parameters are main operation amount for adjusting the electric dust removal process, and the outlet smoke concentration is usually controlled by adjusting the power parameters of the dry-type electric dust remover during operation. Under the condition that the flue gas parameters are relatively stable, the dust removal efficiency is closely related to the power supply parameters. The improvement of the dust removal effect of the electric dust remover means that the power consumption of the electric dust remover is increased, thereby increasing the plant power consumption. Because the dry-type electrostatic precipitator entry does not have smoke concentration measurement station, and the smoke load process is complicated, and the electrostatic precipitator operating environment is changeable among the dust removal process, and the intercoupling, nonlinearity are strong between the working variable, so it is difficult to predict the influence of its power parameter to export smoke concentration through the inside operation mechanism of analysis electrostatic precipitator.
Compared with mechanism modeling, data-driven modeling has better adaptability to the variable and complex actual dedusting process, and is convenient to apply in the aspects of intelligent control and operation optimization. At the present time of rapid development of machine learning, various advanced algorithms are developed endlessly, and huge potential exists in establishing a data-driven electric precipitator model. The relation between the power supply parameters of the dry type electric dust remover and the outlet smoke concentration is obtained through the operation data, and the model is a regression model modeling process based on data driving. The traditional statistical regression methods, such as multivariate linear regression, logistic regression, principal component analysis regression and the like, all aim at the linear regression problem, and are not suitable for the electric precipitation process with nonlinearity and time lag.
Disclosure of Invention
The invention provides a neural network-based method for predicting outlet smoke concentration of a dry-type electric dust remover of a coal-fired unit, and aims to establish a prediction model of power supply parameters and outlet smoke concentration of the dry-type electric dust remover in a data-driven mode. The model can be used as a precondition guarantee for optimizing power supply parameters, so that the energy-saving operation of the dry type electric dust remover is realized under the condition that the outlet smoke concentration meets the environmental protection requirement.
The technical scheme adopted by the invention is as follows:
a coal-fired unit dry-type electric precipitator outlet smoke concentration prediction method based on a neural network comprises the following steps:
s1, obtaining outlet smoke concentration of the unit under each typical load and secondary current values of each electric chamber in each stage of electric fields of the electric dust collector from a database of a distributed control system DCS of the coal-fired thermal power generating unit. For each level of electric field, the average value of the secondary current values of the electric chambers in the electric field is used as the secondary current value I of the level of electric field #i (i＝1,...n)。
And S2, carrying out normalization processing on the obtained data of each load section, and dividing the data into a training set and a test set according to a certain proportion. In order to ensure the effectiveness of model training, the training set covers most of the data fluctuation range.
And S3, selecting a BP neural network model, taking secondary current values of all levels of electric fields as a network input layer, and taking outlet smoke concentration as a network output layer. After the proper number of nodes of the input layer, the number of nodes of the hidden layer and the number of nodes of the output layer are set, putting the data into a training set for iterative training until the errors of two adjacent times are smaller than the set values.
And S4, replacing the next load point of the data set, and repeating the steps S2 and S3.
The invention has the following beneficial effects:
the method provided by the invention can accurately predict the outlet smoke concentration of the dry type electric dust remover in each load section through a data-driven strategy, thereby avoiding the dependence on the complex mechanism process of the electric dust remover. Furthermore, a calculation basis can be provided for realizing optimal energy consumption of the dry-type electric dust remover under the condition of meeting the environmental protection requirement.
Drawings
FIG. 1 is a calculation flow chart of a method for predicting outlet smoke concentration of a dry type electric dust remover of a coal-fired unit based on a neural network;
FIG. 2 is a structural view of a dry type electric dust collector;
fig. 3 is a schematic diagram of a neural network process.
Detailed Description
The following describes embodiments of the present invention with reference to the drawings.
The embodiment discloses a method for predicting outlet smoke concentration of a dry type electric dust remover of a coal-fired unit. As shown in fig. 1, a prediction model of electric dust collector power supply parameters to outlet concentration is established by a neural network method in a data-driven manner, so that a calculation basis can be provided for realizing optimal energy consumption of a dry-type electric dust collector under the condition of meeting environmental protection requirements.
Generally, the dry electric dust collector has a multi-stage electric field, and the embodiment adopts a conventional five-electric-field structure for description. The structure of the electric dust removal in the embodiment is shown in fig. 2, and takes a 1000MW coal-fired unit as an example, and typical working conditions generally include 980MW,850MW,750MW and 650MW. The environmental emission requirement of the unit is rho op ＜35mg/Nm 3 The flue gas flows through the electric fields # 1, #2, #3, #4 and #5 in sequence. The specific implementation process is as follows:
s1, obtaining outlet smoke concentration under typical load of a unit from a database of a distributed control system DCS of a coal-fired thermal power generating unit
And the secondary current value of each electric chamber in each stage of electric field of the electric dust collector. For each level of the electric field, each of the electric fields is usedThe average value of the secondary current values of the electric chambers is used as the secondary current value of the electric field
Namely, it is
Where N represents the typical load of the unit, N =980mw,850mw,750mw, and 650mw in this embodiment.
S2, obtaining the secondary current of each load section
And outlet soot concentration
The data were normalized and divided into training and test sets according to the scale of 6. In order to ensure the effectiveness of model training, the training set needs to cover most of the data fluctuation range.
And S3, selecting a three-layer BP neural network model, taking the secondary current value of each level of electric field as a network input layer, and taking the outlet smoke concentration as a network output layer. The number of input layer nodes is set to m =5 due to the secondary current including 5 electric fields. While the number of hidden layer nodes and the number of output layer nodes are set to n =5 and p =1, respectively, the network propagation process is as shown in fig. 3. Putting the data into a training set for iterative training, and regulating iterative calculation in the training process until the error of two adjacent times is less than 10 -3 And then stop.
S31 setting the weight of the input layer to the hidden layer to be omega ij Offset is a j (ii) a The weight from hidden layer to output layer is ω jk Offset is b k (ii) a Setting learning rate as eta and excitation function as f 1 (x) And f 2 (x) All take the following forms:
s32, calculating the output of the hidden layer, wherein the output of the jth node in the hidden layer is as follows:
s33, calculating the output of an output layer, wherein the estimated value of the smoke concentration in the output layer is output as follows:
s34, calculating the iteration error:
in the formula
Is the outlet smoke concentration at typical load taken from historical data.
S35, if the accuracy of the error of the calculation does not meet the requirement, updating the weight and the bias to perform the next iterative calculation, wherein the formula is as follows:
in trainingCalculating the error of two adjacent times to be less than 10 in the training process -3 And stopping iteration and outputting the current neural network parameters.
And S4, replacing the data set with the next load point, and repeating S2 and S3.
This example is merely illustrative of the specific concepts of the present invention. Persons skilled in the art to which the invention pertains may supplement or modify the methods of the described embodiments depending on the actual implementation of the particular assembly, without departing from the specific spirit of the invention, which is either directly or indirectly connected, or beyond the scope of the appended claims.
Claims (1)
1. A coal-fired unit dry-type electric precipitator outlet smoke concentration prediction method based on a neural network is characterized by comprising the following steps: comprises the following steps:
s1, obtaining outlet smoke concentration of the unit under each typical load and secondary current values of each electric chamber in each stage of electric fields of the electric dust collector from a database of a distributed control system DCS of the coal-fired thermal power generating unit. For each level of electric field, the average value of the secondary current values of all the electric chambers in the electric field is used as the secondary current value I of the level of electric field #i (i＝1,...n)。
And S2, carrying out normalization processing on the obtained data of each load section, and dividing the data into a training set and a test set according to a certain proportion. In order to ensure the effectiveness of model training, the training set covers most of the data fluctuation range.
And S3, selecting a neural network model, taking secondary current values of all levels of electric fields as a network input layer, and taking outlet smoke concentration as a network output layer. The number of nodes of the input layer and the number of nodes of the hidden layer are set to be the number of electric fields, and the number of nodes of the output layer is set to be 1, namely representing the output concentration. And putting the data into a training set for iterative training until the error of two adjacent times is smaller than a set value.
And S4, replacing the next load point of the data set, and repeating the steps S2 and S3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211377924.4A CN115682010A (en) | 2022-11-04 | 2022-11-04 | Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211377924.4A CN115682010A (en) | 2022-11-04 | 2022-11-04 | Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115682010A true CN115682010A (en) | 2023-02-03 |
Family
ID=85050329
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211377924.4A Pending CN115682010A (en) | 2022-11-04 | 2022-11-04 | Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115682010A (en) |
-
2022
- 2022-11-04 CN CN202211377924.4A patent/CN115682010A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103576655B (en) | A kind of power boiler burning subspace modeling and Multipurpose Optimal Method and system | |
| RU2559416C2 (en) | Optimised integrated control for oxy-fuel combustion power plant | |
| CN104534507B (en) | A kind of boiler combustion optimization control method | |
| CN107038334B (en) | System and method for predicting CO emission of circulating fluidized bed household garbage incineration boiler | |
| CN104715142B (en) | A kind of station boiler NOxDischarge dynamic soft-measuring method | |
| CN103728879A (en) | Power station boiler emission soft measuring method based on least squares support vector machine and on-line updating | |
| CN104613468A (en) | Circulating fluidized bedboiler combustion optimizing control method based on fuzzy adaptive inference | |
| CN112488145A (en) | NO based on intelligent methodxOnline prediction method and system | |
| CN103729569A (en) | Soft measurement system for flue gas of power-station boiler on basis of LSSVM (Least Squares Support Vector Machine) and online updating | |
| CN110935567A (en) | Thermal power generating unit dry-type electric precipitator optimization control method and system | |
| CN115471014A (en) | Power grid enterprise electricity carbon consumption prediction method based on deep neural network | |
| CN112016754A (en) | Power station boiler exhaust gas temperature advanced prediction system and method based on neural network | |
| CN115437245A (en) | Boiler optimization control method based on combustion state prediction model | |
| CN115682010A (en) | Neural network-based method for predicting outlet smoke concentration of dry-type electric precipitator of coal-fired unit | |
| CN115111601B (en) | Multi-target boiler combustion optimization control method based on embedded algorithm fusion under variable load | |
| CN109101683B (en) | Model updating method for pyrolysis kettle of coal quality-based utilization and clean pretreatment system | |
| CN114021819B (en) | Thermal power plant nitrogen oxide content prediction method based on fractional gray delay model | |
| CN110728031A (en) | Multi-objective optimization method for balancing complex petrochemical process production energy based on ANN modeling | |
| Ilamathi et al. | Predictive modelling and optimization of nitrogen oxides emission in coal power plant using Artificial Neural Network and Simulated Annealing | |
| CN115755624A (en) | Coal-fired boiler multi-objective optimization method based on evolutionary algorithm | |
| CN113869359B (en) | Method for predicting nitrogen oxides in urban solid waste incineration process based on modularized neural network | |
| CN112231978B (en) | Boiler flue gas acid dew point testing method based on artificial neural network | |
| CN111428906B (en) | Industrial boiler steam volume prediction method based on image transformation | |
| CN112598168A (en) | Power station boiler NO based on monkey swarm algorithm and fast learning networkxEmission amount prediction method | |
| Balamurugan et al. | ANN-SQP Approach for NOx emission reduction in coal fired Boilers |
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 |