CN114046533B - Pulverized coal furnace combustion optimization method based on flame analysis optimization - Google Patents
Pulverized coal furnace combustion optimization method based on flame analysis optimization Download PDFInfo
- Publication number
- CN114046533B CN114046533B CN202111008401.8A CN202111008401A CN114046533B CN 114046533 B CN114046533 B CN 114046533B CN 202111008401 A CN202111008401 A CN 202111008401A CN 114046533 B CN114046533 B CN 114046533B
- Authority
- CN
- China
- Prior art keywords
- flame
- pulverized coal
- coal furnace
- characteristic value
- furnace
- 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.)
- Active
Links
- 239000003245 coal Substances 0.000 title claims abstract description 219
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000005457 optimization Methods 0.000 title claims abstract description 58
- 238000002485 combustion reaction Methods 0.000 title claims abstract description 36
- 238000004458 analytical method Methods 0.000 title claims abstract description 24
- 238000004364 calculation method Methods 0.000 claims abstract description 43
- 238000013178 mathematical model Methods 0.000 claims abstract description 42
- 238000004519 manufacturing process Methods 0.000 claims abstract description 9
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000003860 storage Methods 0.000 claims abstract description 4
- 239000000843 powder Substances 0.000 claims description 26
- 210000002569 neuron Anatomy 0.000 claims description 18
- 238000013507 mapping Methods 0.000 claims description 12
- 238000010219 correlation analysis Methods 0.000 claims description 11
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 9
- 238000012935 Averaging Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 230000006978 adaptation Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 5
- 238000010438 heat treatment Methods 0.000 claims description 4
- 238000010521 absorption reaction Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000009466 transformation Effects 0.000 abstract description 2
- 239000002817 coal dust Substances 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001311 chemical methods and process Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N5/00—Systems for controlling combustion
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N2223/00—Signal processing; Details thereof
- F23N2223/44—Optimum control
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N2229/00—Flame sensors
- F23N2229/20—Camera viewing
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Regulation And Control Of Combustion (AREA)
Abstract
The invention discloses a pulverized coal furnace combustion optimizing method based on flame analysis optimizing, which comprises the following steps: acquiring operation data of the pulverized coal furnace in real time by using an OPC method; carrying out data preprocessing; collecting flame images burnt by the pulverized coal furnace through an industrial camera, and storing the flame images into a storage medium; establishing a mathematical model between flame characteristics and the operation parameters of the pulverized coal furnace; establishing a pulverized coal furnace optimization model; collecting real-time operation data of a unit, and performing steady-state analysis on the operation data of the pulverized coal furnace by adopting a sliding window method to obtain the current operation state of the pulverized coal furnace; when meeting the steam production requirement, the characteristic value of the flame is maximized as an optimization target, and a queuing competition algorithm is used for optimization calculation; writing in the DCS system in real time to realize closed-loop optimization of the operation of the pulverized coal furnace; deep learning calibration; the influence of the lack of real-time coal quality data on the model precision is overcome, and the efficiency of the pulverized coal furnace is highest by controlling the optimal flame characteristic value; the furnace is not required to be stopped for transformation, the optimizing result is reliable, and the optimizing benefit is high.
Description
Technical Field
The invention belongs to the field of intelligent industry, and particularly relates to a pulverized coal furnace combustion optimization method based on flame analysis optimization.
Background
The pulverized coal furnace system has numerous equipment and complex structure, the adjustment of the combustion process is a multi-input multi-output multi-variable related object with the mutual influence of multiple parameters, and the complex nonlinear relation is presented. In operation, as the change of climate and the load change of a heating pipe network are adapted, the pulverized coal furnace is often operated under a variable working condition under a non-rated load, and the coal quality data of the combustion of the pulverized coal furnace cannot be obtained in real time, the operation parameters of each operation of the pulverized coal furnace are difficult to obtain in real time, the operation efficiency of the pulverized coal furnace is reduced, and a large optimization space exists. At present, two methods are generally adopted for the operation of the pulverized coal furnace, one method is that a plurality of operators manually adjust according to experiences accumulated by the operators for a long time, but the method is limited by the reasons of large change of the coal quality of the entering furnace, variable load operation, complicated operation parameters and the like, and proper decisions are difficult to be made in real time by manpower; secondly, a big data technology and a machine learning algorithm are used for establishing a model among various operation variables, coal quality data and efficiency of the pulverized coal furnace, and optimizing calculation is carried out, and compared with a manual operation adjustment method, the method is greatly improved, but has some problems: for example, during optimization calculation, the coal quality data needs to be known, but the coal quality data of the current combustion of the pulverized coal furnace cannot be obtained due to the longer period of the coal quality test, and if the coal quality data is not considered, the accuracy of the model is lower.
A pulverized coal furnace combustion optimizing method and system based on flame analysis and queue competition algorithm optimizing aims at the pulverized coal furnace operation adjusting problem, a combustion flame model is combined with a machine learning algorithm, an optimal pulverized coal furnace operation operating scheme is obtained in real time through optimizing calculation of an intelligent optimizing algorithm, closed-loop optimization of the pulverized coal furnace is achieved, the pulverized coal furnace is enabled to be kept to safely and efficiently operate, and therefore traditional manual adjustment is replaced. The system converts the problem of 'each operation parameter and coal quality data-coal dust furnace efficiency' of the coal dust furnace into the problem of 'each operation parameter-flame model-coal dust furnace efficiency', overcomes the influence of the lack of real-time coal quality data on model accuracy in the problem of 'each operation parameter and coal quality data-coal dust furnace efficiency', does not need to know the coal quality of the coal dust furnace currently combusted, and ensures the highest efficiency of the coal dust furnace by controlling the optimal flame characteristic value.
Disclosure of Invention
Aiming at the defects or improvement demands of the prior art, the invention aims to solve the problem that in the combustion optimization of the pulverized coal furnace in the prior art, the accuracy of an optimization model is lower because the period of coal quality test is longer and the acquisition of the coal quality data of the current combustion of the pulverized coal furnace cannot be implemented.
In order to achieve the above purpose, the invention relates to a pulverized coal furnace combustion optimizing method based on flame analysis optimizing, which comprises the following steps:
step 1, storing operation data of a pulverized coal furnace acquired by an OPC method into a database;
step 2, taking the database in the step 1 as a sample library, and preprocessing data by removing abnormal values, steady-state analysis, correlation analysis and variable reduction to obtain data for establishing a mathematical model;
step 3, collecting flame images of combustion of the pulverized coal furnace through an industrial camera, and storing the flame images into a storage medium;
step 4, establishing a mathematical model between flame characteristics and coal powder furnace operation parameters by extracting the characteristic values of the flame images obtained in the step 3, wherein the characteristic values of the flame images comprise a flame color characteristic value fc, a flame center characteristic value fctr, a flame shape characteristic value fs and a flame fullness characteristic value ff;
step 5, according to the relation between flame characteristics and the efficiency of the pulverized coal furnace and between the flame characteristics and various operation variables of the pulverized coal furnace, the problem of improving the efficiency of the pulverized coal furnace is converted into the problem of improving flame characteristic parameters, and a pulverized coal furnace optimization model is established;
step 6, collecting real-time operation data of the unit, performing steady-state analysis on the operation data of the pulverized coal furnace by adopting a sliding window method, judging whether the operation data is in a steady state, and if the operation data is in the steady state, judging that the current operation state data can be used for subsequent optimization calculation;
step 7, when the steam production requirement is met, the characteristic value of the flame is maximized as an optimization target, state data of the pulverized coal furnace in the steady state in the step 6 is used as input, and an array competition algorithm is used for optimization calculation;
step 8, an OPC method is used for optimizing and calculating an optimization scheme obtained by a queuing competition algorithm, and the optimization scheme is written into a DCS system in real time to realize closed-loop optimization of the operation of the pulverized coal furnace;
and 9, deep learning automatic calibration, wherein the system continuously and deeply learns according to the running condition of the pulverized coal furnace according to an optimized scheme, and the accuracy of the flame model is continuously improved.
Further, the data collected in the step 1 includes: the main steam flow b_s_m of the pulverized coal furnace, the water supply flow b_w_m of the pulverized coal furnace, the rotating speed c_w of the pulverized coal furnace, the inlet pressure airhead_p_in of the air preheater, the outlet pressure airhead_p_out of the air preheater, the inlet pressure outfan_p_in of the induced draft fan and the outlet pressure outfan_p_out of the induced draft fan.
Further, the correlation analysis and variable reduction method in the step 2 is as follows:
and carrying out correlation analysis on each operation variable and the efficiency of the pulverized coal furnace by using a correlation coefficient method, removing variables with weak correlation, realizing approximate subtraction of the variables, sequencing according to the strength of the correlation, and reserving main operation variables with strong correlation.
Further, the method for establishing the mathematical model between the flame characteristic and the operation parameter of the pulverized coal furnace in the step 4 is to establish the mathematical model between each characteristic value of the flame and the operation parameter of the pulverized coal furnace by using a BP neural network algorithm, and comprises the following steps:
first step, network initialization
The input layer is the operation parameters of the pulverized coal furnace, if the pulverized coal furnace has n operation parameters, the input layer has n neurons, and the input vector is X= (X) 1 ,x 2 ,…,x n ) T The method comprises the steps of carrying out a first treatment on the surface of the The output layer is each characteristic value of the flame, if the flame has m characteristic values, the output layer has m neurons, and the output vector is Y= (Y) 1 ,y 2 ,…,y m ) T The method comprises the steps of carrying out a first treatment on the surface of the Determining the number of network input layer nodes and the number of hidden nodes l according to the system input/output series (X, Y), outputting the number of layer nodes, and initializing the connection weight of the input layer and the hidden layer as w ij The connection weight of the hidden layer and the output layer is w jk Initializing the threshold value of each neuron of the hidden layer to alpha j (j=1, 2, …, 1) with output layer each neuron threshold value b k (k=1, 2, …, m), given a learning rate and a neuron excitation function;
second step, hidden layer output calculation
According to the input vector X, the connection weight w between the input layer and the hidden layer ij Hidden layer threshold a j (j=1, 2, …, 1), the hidden layer output H is calculated j ：
Wherein l is the number of hidden layer nodes, and f is a hidden layer excitation function;
third step, output layer output calculation
Outputting H according to hidden layer j Connection weight w of hidden layer and output layer jk And output layer individual neuron threshold b k (k=1, 2, …, m), calculating the predicted output O of the BP neural network k ：
Fourth, error calculation
Output O from network prediction k And desired output Y k Calculating a network prediction error e k ：
e k ＝Y k -O k ，k＝1，2，...，m
Fifth step, weight updating
Based on network prediction error e k Updating network connection weight w ij 、w jk ；
w jk ＝w jk +ηH j e k ，j＝1，2，...，l；k＝1，2，...，m
Wherein eta is the learning rate;
sixth step, updating threshold value
Based on network prediction error e k Updating the network node threshold:
b k ＝b k +e k ，k＝1，2，...，m
seventh, judging whether the algorithm iteration is finished, if not, returning to the second step; so as to obtain a mathematical model between each characteristic value of the flame and the operation parameters of the pulverized coal furnace;
(1) mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(3) mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(4) mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)。
further, in the step 4:
the flame color characteristic value fc refers to extracting the L value of each pixel point of the flame body by using an LAB color model, and averaging to obtain the flame color characteristic value fc, wherein the larger the color characteristic value is, the brighter the flame is, and the higher the pulverized coal combustion efficiency is; the calculation formula of the flame color characteristic value fc is as follows:
wherein k represents the number of pixels of flame; lxi, i=1, 2,3 … k denotes the L value of the i-th pixel;
the flame center characteristic value fctr refers to a calculation formula for calculating the centroid by using a finite point set, wherein the calculation formula is as follows:
where (xi, yi), i=1, 2,3, … k represents the flame ith pixel point coordinate; k represents the number of pixels of flame;
calculating to obtain the center c (x, y) of the flame, and calculating the linear distance between the flame center and the theoretical flame center c0 (x 0, y 0), namely, the flame center characteristic value fctr is obtained, wherein the larger the flame center characteristic value is, the more the flame deviates from the theoretical flame center, and the lower the pulverized coal combustion efficiency is; the flame center c (x, y) and the theoretical flame center c0 (x 0, y 0) (as shown in fig. 3), the flame center characteristic value fctr is calculated as follows:
the flame shape characteristic value fs is that points (xoli, yoli) on the outer edge of the left side of the flame are taken, center symmetry mapping is carried out by taking the flame center c (x, y) as a center, mapping points (xoli ', yoli') are obtained, and the straight line distance between the points (xoli, yori) on the outer edge of the right side of the flame and the mapping points (xoli ', yoli') is calculated; similarly, taking points (xili, yili) on the inner edge of the left side of the flame, carrying out center symmetry mapping by taking the flame center c (x, y) as a center to obtain mapped points (xili ', yili'), and calculating the linear distance between the points (xiri, yiri) on the inner edge of the right side of the flame and the mapped points (xili ', yili'); averaging all the straight line distances to obtain a flame shape characteristic value fs, wherein the larger the flame shape characteristic value is, the worse the flame symmetry is, the worse the pulverized coal combustion stability is, and the lower the pulverized coal combustion efficiency is; the calculation formula of the flame shape characteristic value fs is as follows:
where xori, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the outer edge of the right side of the flame; yori, i=1, 2,3 … p, indicates the ordinate of the i-th pixel point on the outer edge of the right side of the flame; xoli', i=1, 2,3 … p represents the abscissa of the i-th pixel on the outer edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y); yoli', i=1, 2,3 … p represents the ordinate of the ith pixel point on the outer edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical; xiri, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the inner edge of the right side of the flame; yiri, i=1, 2,3 … p denotes the ordinate of the i-th pixel point on the inner edge on the right side of the flame; xili', i=1, 2,3 … p represents the abscissa of the i-th pixel on the inner edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y); yiri', i=1, 2,3 … p represents the ordinate of the ith pixel point on the inner edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical; p represents the number of pixel points at the left edge of the flame;
the characteristic value ff of the flame fullness is the proportion of the flame body to the cross section of the hearth, and the larger the characteristic value of the flame fullness is, the higher the combustion efficiency of the pulverized coal is, and the heat absorption of a heating surface of the hearth is facilitated; the calculation formula of the flame fullness characteristic value ff is as follows:
ff＝k/γ
wherein k represents the number of pixels of flame; gamma represents the number of pixels of the cross section of the hearth;
and carrying out correlation analysis on the coal powder furnace efficiency and the coal powder furnace efficiency, respectively endowing corresponding weight coefficients a1, a2, a3 and a4, and converting the problem of improving the coal powder furnace efficiency into the characteristic problem of improving the flame.
Further, the method for establishing the coal powder furnace optimization model in the step 5 comprises the following steps:
1) Determining an optimization target: when meeting the steam production requirement, maximizing the characteristic value of the flame;
2) Determining constraint conditions:
(1) a mathematical model between each characteristic value of flame and an operating parameter of a pulverized coal furnace, comprising:
mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) the running load of the pulverized coal furnace;
(3) design parameters of the pulverized coal furnace;
(4) actual running characteristics of the pulverized coal furnace, etc.;
3) Determining an optimization variable:
(1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the induced draft fan outlet pressure outfan_p_out.
Further, the method for performing optimization calculation by using the queuing competition algorithm in the step 7 includes;
optimization target: maximizing flame characteristics when meeting steam production requirements, i.e.
max(a1×fc+a2×fctr+a3×fs+a4×ff)
Wherein a1, a2, a3 and a4 are weight coefficients, fc is a flame color characteristic value, fctr is a flame center characteristic value, fs is a flame shape characteristic value, and ff is a flame fullness characteristic value;
optimizing variables: (1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the outlet pressure outfan_p_out of the induced draft fan;
the steps of the queuing competition algorithm are as follows:
1) t=1, v candidate solutions representing v families are generated in the independent variable range by a random method, and the random method is as follows:
wherein x is 1 i，j In evolution 1, the value of the j variable, lb, of the i family in the queue j Is the lower limit of the independent variable, ub j R is a random number, which is the upper limit of the independent variable;
2) Calculating the adaptation value of each family;
3) According to the adaptation value of each family, v families are arranged into a queue, and the families are arranged in an ascending order;
4) Judging whether an iteration termination condition is met, if so, selecting an individual arranged at the first position as an optimal solution, and taking a corresponding adaptive value as an optimal value, otherwise, turning to the step 5);
5) Sequentially allocating the search spaces corresponding to the families from small to large according to the positions in the queue; the search space with the smallest division of the families in front of the queue and the search space with the largest division of the families in the rear of the queue are allocated as follows:
in lb t i，j And ub t i，j Respectively represent the t generation of a subfamily groupIn the chemical process, the lower limit and the upper limit of the search space allocated by the j variable of the i family in the queue; x is x t i，j Is the value of the jth variable of the ith family in the queue in the t-th generation evolution; delta t j The length of the value interval of the jth variable in the t generation;
6) Each family produces epsilon offspring by asexual propagation in its search space and competes with the parent, leaving only one of the most excellent individuals to compete for the family status of the family in the next round, the propagation method being similar to that in 1);
7) The optimal individuals in each family constitute a new family, turning to step 2), t=t+1.
In general, the above technical solutions conceived by the present invention, compared with the prior art, enable the following beneficial effects to be obtained:
(1) The method for optimizing the combustion of the pulverized coal furnace based on flame analysis optimization converts the problem of 'each operation parameter and coal quality data-pulverized coal furnace efficiency' of the pulverized coal furnace into the problem of 'each operation parameter-flame model-pulverized coal furnace efficiency', overcomes the influence of the lack of real-time coal quality data on model precision in the problem of 'each operation parameter and coal quality data-pulverized coal furnace efficiency', does not need to know the coal quality of the pulverized coal furnace currently combusted, and ensures the highest efficiency of the pulverized coal furnace by controlling the optimal flame characteristic value.
(2) According to the pulverized coal furnace combustion optimization method based on flame analysis optimization, the flame characteristic analysis and the queue competition algorithm are used for optimizing energy conservation, the furnace shutdown transformation is not needed, and the normal operation of the pulverized coal furnace is not influenced in the system implementation process; the system does not need to reform equipment in the implementation process, so that the investment cost is reduced; the flame model has high precision, as shown in fig. 4, the model fitting R value is 0.9940, and the error is less than 2%; the system uses the queued competition algorithm for optimizing, solves the problem that other optimizing algorithms are easy to fall into local optimization, and has reliable optimizing result and high optimizing benefit.
Drawings
FIG. 1 is a system flow chart of a preferred embodiment of the present invention;
FIG. 2 is a flow chart of flame signature extraction in accordance with a preferred embodiment of the present invention;
FIG. 3 is a diagram of flame signature analysis in accordance with a preferred embodiment of the present invention;
FIG. 4 is a graph showing actual values versus calculated values for a preferred embodiment of the present invention;
FIG. 5 is a graph showing the result of optimizing the boiler efficiency according to the preferred embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
The invention relates to a pulverized coal furnace combustion optimizing method and system based on flame analysis and queue competition algorithm optimizing, which aims to solve the problem that a pulverized coal furnace is difficult to operate in an optimal state by means of manual operation. The system collects the operation data and flame video data of the pulverized coal furnace in an OPC technology and other modes, extracts the characteristics of flames, establishes a mathematical model of the flame characteristics and the operation parameters of the pulverized coal furnace, performs optimization calculation by using a queuing competition algorithm to obtain the optimal operation parameters of the pulverized coal furnace, and writes the optimization calculation result into a DCS system in real time by using the OPC technology to realize closed-loop optimization of the operation of the pulverized coal furnace.
The workflow of the system is shown in figure 1. The method comprises the following steps:
(1) Data acquisition
The OPC technology is used for collecting the operation data of the pulverized coal furnace in real time and storing the operation data into a database according to a specified format. The collected data mainly comprises: the main steam flow b_s_m of the pulverized coal furnace, the water supply flow b_w_m of the pulverized coal furnace, the rotating speed c_w of the pulverized coal furnace, the inlet pressure airhead_p_in of the air preheater, the outlet pressure airhead_p_out of the air preheater, the inlet pressure outfan_p_in of the induced draft fan and the outlet pressure outfan_p_out of the induced draft fan.
(2) Data preprocessing
(1) And eliminating abnormal values, and deleting NAN empty data, the value of 0 and abnormal value data which violate normal conditions in the running data of the pulverized coal furnace system.
(2) Steady-state analysis because the coal powder furnace has severe fluctuation in each operation parameter when the coal powder furnace is in variable load (i.e. unsteady state) operation, the real operation level of the coal powder furnace is difficult to reflect by the data at the moment, so that the operation data of the coal powder furnace system is subjected to steady analysis by adopting a sliding window method, abnormal data of unsteady state operation of the coal powder furnace is removed, and the operation data in a steady state is obtained.
The main steam flow of the pulverized coal furnace is an important factor reflecting whether the operation condition of a unit is stable, the main steam flow of the pulverized coal furnace is taken as a standard for measuring the operation state of the pulverized coal furnace, a sliding window method is used for carrying out steady-state analysis on the operation data of the pulverized coal furnace according to a time sequence, the operation data in a period of time is taken as a window from the starting time position of the data, the fluctuation condition of the data in the window is calculated, and if the fluctuation of the main steam flow of the pulverized coal furnace is large, the data in the window is considered to be in an unsteady state and the data is not reserved; otherwise, the data is in a steady state, and the data is reserved. And sliding from the starting time point to the ending time point by the process to obtain all steady-state data of the pulverized coal furnace. The calculation formula is as follows:
wherein: delta represents the standard deviation of the main steam flow of the boiler in the window; t represents a start time; n represents the width of the sliding window;representing the average value of the main steam flow of the boiler from t to t+N-1; x is x t A value representing a t-th boiler main steam flow; lambda is the primary steam flow fluctuation range.
(3) Correlation analysis and variable reduction
And carrying out correlation analysis on each operation variable and the efficiency of the pulverized coal furnace by using a correlation coefficient method, sequencing according to the strength of the correlation, reserving main operation variables with strong correlation, removing variables with weak correlation, and realizing variable reduction. The correlation coefficient is a statistical index of the degree of the relation between the reaction variables, and the value interval of the correlation coefficient is between 1 and-1. 1 represents a complete linear correlation of the two variables, -1 represents a complete negative correlation of the two variables, and 0 represents an uncorrelation of the two variables. The closer the data is to 0, the weaker the correlation is.
(3) Flame image acquisition
And collecting flame images of combustion of the pulverized coal furnace through an industrial camera, and storing the flame images in a storage medium.
(4) Flame model
1) Flame feature extraction
The characteristic values of the flame are extracted, including a flame color characteristic value fc, a flame center characteristic value fctr, a flame shape characteristic value fs and a flame fullness characteristic value ff. A flame signature extraction flow chart is shown in fig. 2.
(1) Flame color characteristic value fc
And extracting the L value of each pixel point of the flame body by using the LAB color model, and averaging to obtain a flame color characteristic value fc, wherein the larger the color characteristic value is, the brighter the flame is, and the higher the pulverized coal combustion efficiency is. The calculation formula of the flame color characteristic value fc is as follows:
wherein k represents the number of pixels of flame; lxi, i=1, 2,3 … k denotes the L value of the i-th pixel.
(2) Flame center characteristic value fctr
The method is characterized in that a calculation formula for calculating the centroid by using a finite point set is used for calculating the linear distance between the flame center and the theoretical flame center, and the calculation formula is as follows:
where (xi, yi), i=1, 2,3, … k represents the flame ith pixel point coordinate; k represents the number of pixels of the flame.
The center c (x, y) of the flame is obtained through calculation, the linear distance between the flame center and the theoretical flame center c0 (x 0, y 0) is calculated, namely, the flame center characteristic value fctr is obtained, the larger the flame center characteristic value is, the more the flame deviates from the theoretical flame center, the lower the pulverized coal combustion efficiency is, and meanwhile, the potential safety hazards or safety accidents such as high Wen Jiezha, uneven steam temperature of a water-cooled wall and uneven flow of the water-cooled wall can be possibly caused by the fact that the flame center characteristic value is too large. The flame center c (x, y) and the theoretical flame center c0 (x 0, y 0) (as shown in fig. 3), the flame center characteristic value fctr is calculated as follows:
(3) characteristic value fs of flame shape
As shown in fig. 3, taking points (xoli, yoli) on the outer edge of the left side of the flame, carrying out central symmetry mapping by taking the flame center c (x, y) as a center to obtain mapped points (xoli ', yoli'), and calculating the linear distance between the points (xoli, yori) on the outer edge of the right side of the flame and the mapped points (xoli ', yoli'); similarly, the points (xili, yili) on the inner edge of the flame left side are taken, the center symmetry mapping is carried out by taking the flame center c (x, y) as the center, the mapping points (xili ', yili') are obtained, and the straight line distance between the points (xiri, yiri) on the inner edge of the flame right side and the mapping points (xili ', yili') is calculated. And averaging all the straight line distances to obtain a flame shape characteristic value fs, wherein the larger the flame shape characteristic value is, the worse the flame symmetry is, the worse the pulverized coal combustion stability is, and the lower the pulverized coal combustion efficiency is. The calculation formula of the flame shape characteristic value fs is as follows:
where xori, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the outer edge of the right side of the flame; yori, i=1, 2,3 … p, indicates the ordinate of the i-th pixel point on the outer edge of the right side of the flame; xoli', i=1, 2,3 … p represents the abscissa of the i-th pixel on the outer edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y); yoli', i=1, 2,3 … p represents the ordinate of the ith pixel point on the outer edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical; xiri, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the inner edge of the right side of the flame; yiri, i=1, 2,3 … p denotes the ordinate of the i-th pixel point on the inner edge on the right side of the flame; xili', i=1, 2,3 … p represents the abscissa of the i-th pixel on the inner edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y); yiri', i=1, 2,3 … p represents the ordinate of the ith pixel point on the inner edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical; p represents the number of pixels on the left edge of the flame.
(4) Flame fullness characteristic value ff
The ratio of the flame body to the cross section of the hearth is the characteristic value ff of the flame fullness, and the larger the characteristic value of the flame fullness is, the higher the pulverized coal combustion efficiency is, and the heat absorption of the heating surface of the hearth is facilitated. The calculation formula of the flame fullness characteristic value ff is as follows:
ff＝k/γ
wherein k represents the number of pixels of flame; gamma represents the number of pixels of the cross section of the hearth.
And carrying out correlation analysis on the coal powder furnace efficiency and the coal powder furnace efficiency, respectively endowing corresponding weight coefficients a1, a2, a3 and a4, and converting the problem of improving the coal powder furnace efficiency into the characteristic problem of improving the flame.
2) Establishing a mathematical model between flame characteristics and coal powder furnace operation parameters
The BP neural network algorithm is used for establishing a mathematical model between each characteristic value of flame and the operation parameter of the pulverized coal furnace, and the steps are as follows:
first step, network initialization
The input layer is the operation parameters of the pulverized coal furnace, if the pulverized coal furnace has n operation parameters, the input layer has n neurons, and the input vector is X= (X) 1 ,x 2 ,…,x n ) T The method comprises the steps of carrying out a first treatment on the surface of the The output layer is each characteristic value of the flame, if the flame has m characteristic values, the output layer has m neurons, and the output vector is Y= (Y) 1 ,y 2 ,…,y m ) T The method comprises the steps of carrying out a first treatment on the surface of the Determining the number of nodes in network input layer and hiding according to system input-output series (X, Y)Node number l, output layer node number, initializing connection weight of input layer and hidden layer as w ij The connection weight of the hidden layer and the output layer is w jk Initializing the threshold value of each neuron of the hidden layer to alpha j (j=1, 2, …, l) with output layer each neuron threshold b k (k=1, 2, …, m), given a learning rate and a neuron excitation function;
second step, hidden layer output calculation
According to the input vector X, the connection weight w between the input layer and the hidden layer ij Hidden layer threshold a j (j=1, 2, …, l) calculating the hidden layer output H j ：
Wherein l is the number of hidden layer nodes, and f is a hidden layer excitation function;
third step, output layer output calculation
Outputting H according to hidden layer j Connection weight w of hidden layer and output layer jk And output layer individual neuron threshold b k (k=1, 2, …, m), calculating the predicted output O of the BP neural network k ：
Fourth, error calculation
Output O from network prediction k And desired output Y k Calculating a network prediction error e k ：
e k ＝Y k -O k ，k＝1，2，...，m
Fifth step, weight updating
Based on network prediction error e k Updating network connection weight w ij 、w jk ；
w jk ＝w jk +ηH j e k ，j＝1，2，...，l；k＝1，2，...，m
Wherein eta is the learning rate;
sixth step, updating threshold value
Based on network prediction error e k Updating the network node threshold:
b k ＝b k +e k ，k＝1，2，...，m
seventh, judging whether the algorithm iteration is finished, if not, returning to the second step; so as to obtain a mathematical model between each characteristic value of the flame and the operation parameters of the pulverized coal furnace;
(1) mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(3) mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(4) mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(5) Optimizing model of pulverized coal furnace
According to the relation between the flame characteristics and the efficiency of the pulverized coal furnace and between the flame characteristics and the operating variables of the pulverized coal furnace, the problem of improving the efficiency of the pulverized coal furnace is converted into the problem of improving the flame characteristic parameters, and a pulverized coal furnace optimization model is established.
1) Determining an optimization target: the characteristic value of the flame is maximized when the steam production requirement is met.
2) Determining constraint conditions:
(1) a mathematical model between each characteristic value of flame and an operating parameter of a pulverized coal furnace, comprising:
mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) the running load of the pulverized coal furnace;
(3) design parameters of the pulverized coal furnace;
(4) the actual running characteristics of the pulverized coal furnace, etc.
3) Determining an optimization variable:
(1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the induced draft fan outlet pressure outfan_p_out.
(6) Collecting real-time operation data of unit and analyzing
Because the operation parameters have larger fluctuation when the pulverized coal furnace is in variable load (i.e. unsteady state) operation, the real operation level of the pulverized coal furnace is difficult to reflect by the data at the moment, so that the operation data of the pulverized coal furnace is subjected to steady state analysis by adopting a sliding window method, and the current operation state of the pulverized coal furnace is obtained.
The main steam flow of the pulverized coal furnace is an important factor reflecting whether the running condition of the unit is stable, the main steam flow of the pulverized coal furnace is taken as a standard for measuring the running state of the unit, a sliding window method is used for carrying out steady-state analysis on the running data of the pulverized coal furnace according to a time sequence, the running data in a period of time is taken as a window from the starting time position of the data, the fluctuation condition of the data in the window is calculated, and if the fluctuation of the main steam flow of the pulverized coal furnace is large, the data in the window is considered to be in an unsteady state; otherwise, the data is in a steady state, and when the pulverized coal furnace is in a steady state, optimization calculation is carried out.
(7) Optimization calculation by using queuing competition algorithm
Optimization target: maximizing flame characteristics when meeting steam production requirements, i.e.
max(a1×fc+a2×fctr+a3×fs+a4×ff)
Wherein a1, a2, a3 and a4 are weight coefficients, fc is a flame color characteristic value, fctr is a flame center characteristic value, fs is a flame shape characteristic value, and ff is a flame fullness characteristic value;
optimizing variables:
(1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the induced draft fan outlet pressure outfan_p_out.
The steps of the queuing competition algorithm are as follows:
1) t=1, v candidate solutions representing v families are generated in the independent variable range by a random method, and the random method is as follows:
wherein x is 1 i，j In evolution 1, the value of the j variable, lb, of the i family in the queue j Is the lower limit of the independent variable, ub j R is a random number, which is the upper limit of the independent variable;
2) Calculating the adaptation value of each family;
3) According to the adaptation value of each family, v families are arranged into a queue, and the invention is arranged in an ascending order;
4) Judging whether an iteration termination condition is met, if so, selecting an individual arranged at the first position as an optimal solution, and taking a corresponding adaptive value as an optimal value, otherwise, turning to the step 5);
5) Sequentially allocating the search spaces corresponding to the families from small to large according to the positions in the queue; the search space with the smallest division of the families in front of the queue and the search space with the largest division of the families in the rear of the queue are allocated as follows:
in lb t i，j And ub t i，j Respectively representing the lower limit and the upper limit of a search space allocated by the jth variable of the ith family in the queue in the t generation evolution of a certain subfamily group; x is x t i，j Is the value of the jth variable of the ith family in the queue in the t-th generation evolution; delta t j The length of the value interval of the jth variable in the t generation;
6) Each family produces epsilon offspring by asexual propagation in its search space and competes with the parent, leaving only one of the most excellent individuals to compete for the family status of the family in the next round, the propagation method being similar to that in 1);
7) The optimal individuals in each family constitute a new family, turning to step 2), t=t+1.
(8) Writing DCS system
And (3) an OPC technology is used for writing an optimization scheme obtained by optimizing and calculating a queuing competition algorithm into the DCS in real time, so that closed-loop optimization of the operation of the pulverized coal furnace is realized.
(9) Deep learning automatic calibration
According to the condition that the pulverized coal furnace operates according to an optimized scheme, the system continuously performs deep learning, and the accuracy of a flame model is continuously improved.
The efficiency of the pulverized coal furnace is improved by more than 1%, and the optimization calculation result is shown in figure 5; the system can learn according to the actual running condition of the optimization result, the precision of the flame model is continuously improved, and the model upgrading process does not need manual intervention; before the system starts the optimization calculation, the current running state of the pulverized coal furnace can be judged, and if the pulverized coal furnace is in unsteady state or abnormal running, the optimization calculation is not performed, so that the safe and stable running of the pulverized coal furnace is ensured.
In addition, it should be noted that the same symbol parameters in different formulas in the material of the present invention have the same meaning.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (3)
1. The pulverized coal furnace combustion optimizing method based on flame analysis optimizing is characterized by comprising the following steps of:
step 1, storing operation data of a pulverized coal furnace acquired by an OPC method into a database; the acquired data includes: the main steam flow b_s_m of the pulverized coal furnace, the water supply flow b_w_m of the pulverized coal furnace, the rotating speed c_w of the pulverized coal feeder, the inlet pressure airhead_p_in of the air preheater, the outlet pressure airhead_p_out of the air preheater, the inlet pressure outfan_p_in of the induced draft fan and the outlet pressure outfan_p_out of the induced draft fan;
step 2, taking the database in the step 1 as a sample library, and preprocessing data by removing abnormal values, steady-state analysis, correlation analysis and variable reduction to obtain data for establishing a mathematical model;
step 3, collecting flame images of combustion of the pulverized coal furnace through an industrial camera, and storing the flame images into a storage medium;
step 4, establishing a mathematical model between flame characteristics and coal powder furnace operation parameters by extracting the characteristic values of the flame images obtained in the step 3, wherein the characteristic values of the flame images comprise a flame color characteristic value fc, a flame center characteristic value fctr, a flame shape characteristic value fs and a flame fullness characteristic value ff;
the method for establishing the mathematical model between the flame characteristic and the coal powder furnace operation parameter in the step 4 is to establish the mathematical model between each characteristic value of the flame and the coal powder furnace operation parameter by using BP neural network algorithm, and comprises the following steps:
first step, network initialization
The input layer is the operation parameters of the pulverized coal furnace, if the pulverized coal furnace has n operation parameters, the input layer has n neurons, and the input vector is X= (X) 1 ,x 2 ,…,x n ) T The method comprises the steps of carrying out a first treatment on the surface of the The output layer is each characteristic value of the flame, if the flame has m characteristic values, the output layer has m neurons, and the output vector is Y= (Y) 1 ,y 2 ,…,y m ) T The method comprises the steps of carrying out a first treatment on the surface of the Determining the number of network input layer nodes and the number of hidden nodes l according to the system input/output series (X, Y), outputting the number of layer nodes, and initializing the connection weight of the input layer and the hidden layer as w ij The connection weight of the hidden layer and the output layer is w jk Initializing the threshold value of each neuron of the hidden layer to alpha j (j=1, 2, …, l) with output layer each neuron threshold b k (k=1, 2, …, m), given a learning rate and a neuron excitation function;
second step, hidden layer output calculation
According to the input vector X, the connection weight w between the input layer and the hidden layer ij Hidden layer threshold alpha j (j=1, 2, …, l) calculating the hidden layer output H j ：
Wherein l is the number of hidden layer nodes, and f is a hidden layer excitation function;
third step, output layer output calculation
Outputting H according to hidden layer j Connection weight w of hidden layer and output layer jk And output layer individual neuron threshold b k (k=1, 2, …, m), calculating the predicted output O of the BP neural network k ：
Fourth, error calculation
Output O from network prediction k And desired output Y k Calculating a network prediction error e k ：
e k ＝Y k -O k ，k＝1，2，...，m
Fifth step, weight updating
Based on network prediction error e k Updating network connection weight w ij 、w jk ；
w jk ＝w jk +ηH j e k ，j＝1，2，...，l；k＝1，2，...，m
Wherein eta is the learning rate;
sixth step, updating threshold value
Based on network prediction error e k Updating a network node threshold;
b k ＝b k +e k ，k＝1，2，...，m
seventh, judging whether the algorithm iteration is finished, if not, returning to the second step; so as to obtain a mathematical model between each characteristic value of the flame and the operation parameters of the pulverized coal furnace;
(1) mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(3) mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(4) mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)；
the flame color characteristic value fc refers to extracting the L value of each pixel point of the flame body by using an LAB color model, and averaging to obtain the flame color characteristic value fc, wherein the larger the color characteristic value is, the brighter the flame is, and the higher the pulverized coal combustion efficiency is; the calculation formula of the flame color characteristic value fc is as follows:
wherein k represents the number of pixels of flame; lxi, i=1, 2,3 … k denotes the L value of the i-th pixel;
the flame center characteristic value fctr refers to a calculation formula for calculating the centroid by using a finite point set, wherein the calculation formula is as follows:
where (xi, yi), i=1, 2,3, … k represents the flame ith pixel point coordinate; k represents the number of pixels of flame;
calculating to obtain the center c (x, y) of the flame, and calculating the linear distance between the flame center and the theoretical flame center c0 (x 0, y 0), namely, the flame center characteristic value fctr is obtained, wherein the larger the flame center characteristic value is, the more the flame deviates from the theoretical flame center, and the lower the pulverized coal combustion efficiency is;
the flame center c (x, y) and the theoretical flame center c0 (x 0, y 0) are calculated as follows:
the flame shape characteristic value fs is that points (xoli, yoli) on the outer edge of the left side of the flame are taken, center symmetry mapping is carried out by taking the flame center c (x, y) as a center, mapping points (xoli ', yoli') are obtained, and the straight line distance between the points (xoli, yori) on the outer edge of the right side of the flame and the mapping points (xoli ', yoli') is calculated; similarly, taking points (xili, yili) on the inner edge of the left side of the flame, carrying out center symmetry mapping by taking the flame center c (x, y) as a center to obtain mapped points (xili ', yili'), and calculating the linear distance between the points (xiri, yiri) on the inner edge of the right side of the flame and the mapped points (xili ', yili'); averaging all the straight line distances to obtain a flame shape characteristic value fs, wherein the larger the flame shape characteristic value is, the worse the flame symmetry is, the worse the pulverized coal combustion stability is, and the lower the pulverized coal combustion efficiency is; the calculation formula of the flame shape characteristic value fs is as follows:
where xori, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the outer edge of the right side of the flame;
yori, i=1, 2,3 … p, indicates the ordinate of the i-th pixel point on the outer edge of the right side of the flame;
xoli', i=1, 2,3 … p represents the abscissa of the i-th pixel on the outer edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y);
yoli', i=1, 2,3 … p represents the ordinate of the ith pixel point on the outer edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical;
xiri, i=1, 2,3 … p represents the abscissa of the i-th pixel point on the inner edge of the right side of the flame;
yiri, i=1, 2,3 … p denotes the ordinate of the i-th pixel point on the inner edge on the right side of the flame;
xili', i=1, 2,3 … p represents the abscissa of the i-th pixel on the inner edge of the left side of the flame after being mapped symmetrically about the flame center c (x, y);
yiri', i=1, 2,3 … p represents the ordinate of the ith pixel point on the inner edge of the left side of the flame after being mapped with the center of the flame c (x, y) being symmetrical;
p represents the number of pixel points at the left edge of the flame;
the characteristic value ff of the flame fullness is the proportion of the flame body to the cross section of the hearth, and the larger the characteristic value of the flame fullness is, the higher the combustion efficiency of the pulverized coal is, and the heat absorption of a heating surface of the hearth is facilitated; the calculation formula of the flame fullness characteristic value ff is as follows:
ff＝k/γ
wherein k represents the number of pixels of flame; gamma represents the number of pixels of the cross section of the hearth;
carrying out correlation analysis on the coal powder furnace efficiency and the coal powder furnace efficiency, and respectively endowing corresponding weight coefficients a1, a2, a3 and a4;
step 5, establishing a pulverized coal furnace optimization model according to the relation between flame characteristics and pulverized coal furnace efficiency and between the flame characteristics and various operation variables of the pulverized coal furnace;
the method for establishing the coal powder furnace optimization model in the step 5 comprises the following steps:
1) Determining an optimization target: when meeting the steam production requirement, maximizing the characteristic value of the flame;
2) Determining constraint conditions:
(1) a mathematical model between each characteristic value of flame and an operating parameter of a pulverized coal furnace, comprising:
mathematical model between flame colour characteristic value fc and pulverized coal furnace operating parameters:
fc＝f1(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame center characteristic fctr and pulverized coal furnace operating parameters:
fctr＝f2(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame shape characteristic value fs and pulverized coal furnace operating parameters:
fs＝f3(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
mathematical model between flame fullness characteristic value ff and pulverized coal furnace operating parameters:
ff＝f4(b_s_m,b_w_m,c_w,airheat_p_in,airheat_p_out,outfan_p_in,outfan_p_out)
(2) the running load of the pulverized coal furnace;
(3) design parameters of the pulverized coal furnace;
(4) actual running characteristics of the pulverized coal furnace, etc.;
3) Determining an optimization variable:
(1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the outlet pressure outfan_p_out of the induced draft fan;
step 6, collecting real-time operation data of the unit, performing steady-state analysis on the operation data of the pulverized coal furnace by adopting a sliding window method, judging whether the operation data is in a steady state, and if the operation data is in the steady state, judging that the current operation state data can be used for subsequent optimization calculation;
step 7, when the steam production requirement is met, the characteristic value of the flame is maximized as an optimization target, state data of the pulverized coal furnace in the steady state in the step 6 is used as input, and an array competition algorithm is used for optimization calculation;
step 8, an OPC method is used for optimizing and calculating an optimization scheme obtained by a queuing competition algorithm, and the optimization scheme is written into a DCS system in real time to realize closed-loop optimization of the operation of the pulverized coal furnace;
and 9, deep learning automatic calibration, wherein the system continuously and deeply learns according to the running condition of the pulverized coal furnace according to an optimized scheme, and the accuracy of the flame model is continuously improved.
2. The method for optimizing the combustion of the pulverized coal furnace based on flame analysis and optimization according to claim 1, wherein the method for reducing the correlation analysis and the variables in the step 2 is as follows:
and carrying out correlation analysis on each operation variable and the efficiency of the pulverized coal furnace by using a correlation coefficient method, removing variables with weak correlation, realizing approximate subtraction of the variables, sequencing according to the strength of the correlation, and reserving main operation variables with strong correlation.
3. The method for optimizing the combustion of the pulverized coal furnace based on flame analysis optimizing according to claim 1, wherein the method for optimizing calculation in the step 7 by using a queuing competition algorithm comprises the following steps of;
optimization target: maximizing flame characteristics when meeting steam production requirements, i.e.
max(a1×fc+a2×fctr+a3×fs+a4×ff)
Wherein a1, a2, a3 and a4 are weight coefficients, fc is a flame color characteristic value, fctr is a flame center characteristic value, fs is a flame shape characteristic value, and ff is a flame fullness characteristic value;
optimizing variables: (1) main steam flow b_s_m of the pulverized coal furnace;
(2) the water supply flow b_w_m of the pulverized coal furnace;
(3) the rotational speed c_w of the powder feeder;
(4) air preheater inlet pressure airhead_p_in;
(5) air preheater outlet pressure airhead_p_out;
(6) the inlet pressure outfan_p_in of the induced draft fan;
(7) the outlet pressure outfan_p_out of the induced draft fan;
the steps of the queuing competition algorithm are as follows:
1) t=1, and v candidate solutions are generated within the independent variable range by adopting a random method, wherein v candidate solutions represent v families, and initial subfamily groups of the node are formed, and the random method is as follows:
wherein x is 1 i,j In evolution 1, the value of the j variable, lb, of the i family in the queue j Is the lower limit of the independent variable, ub j R is a random number, which is the upper limit of the independent variable;
2) Calculating the adaptation value of each family;
3) According to the adaptation value of each family, v families are arranged into a queue, and are arranged in ascending order;
4) Judging whether an iteration termination condition is met, if so, selecting an individual arranged at the first position as an optimal solution, and taking a corresponding adaptive value as an optimal value, otherwise, turning to the step 5);
5) Sequentially allocating the search spaces corresponding to the families from small to large according to the positions in the queue; the search space with the smallest division of the families in front of the queue and the search space with the largest division of the families in the rear of the queue are allocated as follows:
in lb t i,j And ub t i,j Respectively represent the t generation of a subfamily groupIn evolution, the lower and upper limits of the search space allocated by the j variable of the i family in the queue; x is x t i,j Is the j variable value of the i family in the queue in the t generation evolution; delta t j The length of the value interval of the jth variable in the t generation;
6) Each family produces epsilon offspring by asexual propagation in its search space and competes with the parent, leaving only one of the most excellent individuals to compete for the family status of the family in the next round, the propagation method being similar to that in 1);
7) The optimal individuals in each family constitute a new family, turning to step 2), t=t+1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111008401.8A CN114046533B (en) | 2021-08-31 | 2021-08-31 | Pulverized coal furnace combustion optimization method based on flame analysis optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111008401.8A CN114046533B (en) | 2021-08-31 | 2021-08-31 | Pulverized coal furnace combustion optimization method based on flame analysis optimization |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114046533A CN114046533A (en) | 2022-02-15 |
| CN114046533B true CN114046533B (en) | 2024-03-12 |
Family
ID=80204979
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111008401.8A Active CN114046533B (en) | 2021-08-31 | 2021-08-31 | Pulverized coal furnace combustion optimization method based on flame analysis optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114046533B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117433036B (en) * | 2023-11-10 | 2024-06-07 | 唐山三友化工股份有限公司 | Pulverized coal furnace combustion optimization control system and method based on flame intensity detection |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846305A (en) * | 2017-01-11 | 2017-06-13 | 华北电力大学 | A kind of boiler combustion stability monitoring method based on many characteristics of image of flame |
| CN107025652A (en) * | 2017-05-05 | 2017-08-08 | 太原理工大学 | A kind of flame detecting method based on kinetic characteristic and color space time information |
| CN107191914A (en) * | 2017-07-17 | 2017-09-22 | 武汉智凯科技有限公司 | Boiler on-line tuning system and method based on as-fired coal information and fire defector |
| CN107729913A (en) * | 2017-08-25 | 2018-02-23 | 徐州科融环境资源股份有限公司 | A kind of boiler furnace Situation Awareness method based on multiple features fusion cluster |
| CN109300110A (en) * | 2018-08-24 | 2019-02-01 | 哈尔滨理工大学 | A kind of forest fire image detecting method based on improvement color model |
| CN112628712A (en) * | 2021-01-11 | 2021-04-09 | 大唐黄岛发电有限责任公司 | Secondary air closed-loop optimization control system based on air door resistance coefficient |
| CN113222244A (en) * | 2021-05-10 | 2021-08-06 | 刘旸 | Online boiler combustion optimization method based on flame combustion image judgment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2107312A1 (en) * | 2008-04-01 | 2009-10-07 | Siemens Aktiengesellschaft | Pilot combustor in a burner |
-
2021
- 2021-08-31 CN CN202111008401.8A patent/CN114046533B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846305A (en) * | 2017-01-11 | 2017-06-13 | 华北电力大学 | A kind of boiler combustion stability monitoring method based on many characteristics of image of flame |
| CN107025652A (en) * | 2017-05-05 | 2017-08-08 | 太原理工大学 | A kind of flame detecting method based on kinetic characteristic and color space time information |
| CN107191914A (en) * | 2017-07-17 | 2017-09-22 | 武汉智凯科技有限公司 | Boiler on-line tuning system and method based on as-fired coal information and fire defector |
| CN107729913A (en) * | 2017-08-25 | 2018-02-23 | 徐州科融环境资源股份有限公司 | A kind of boiler furnace Situation Awareness method based on multiple features fusion cluster |
| CN109300110A (en) * | 2018-08-24 | 2019-02-01 | 哈尔滨理工大学 | A kind of forest fire image detecting method based on improvement color model |
| CN112628712A (en) * | 2021-01-11 | 2021-04-09 | 大唐黄岛发电有限责任公司 | Secondary air closed-loop optimization control system based on air door resistance coefficient |
| CN113222244A (en) * | 2021-05-10 | 2021-08-06 | 刘旸 | Online boiler combustion optimization method based on flame combustion image judgment |
Non-Patent Citations (1)
| Title |
|---|
| 并行列队竞争算法及应用;张和烽;武汉理工大学学报;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114046533A (en) | 2022-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107726358B (en) | Boiler Combustion Optimization System based on CFD numerical simulations and intelligent modeling and method | |
| US8135653B2 (en) | Power plant control device which uses a model, a learning signal, a correction signal, and a manipulation signal | |
| CN107203687B (en) | Multi-target cooperative intelligent optimization control method for desulfurization process of absorption tower | |
| CN110030843B (en) | Heat accumulating type aluminum melting furnace parameter optimization setting method based on improved whale optimization algorithm | |
| CN109492319B (en) | Soft measurement method for oxygen content of boiler flue gas of thermal power plant | |
| CN114046533B (en) | Pulverized coal furnace combustion optimization method based on flame analysis optimization | |
| CN110097929A (en) | A kind of blast furnace molten iron silicon content on-line prediction method | |
| CN110413601B (en) | Generator data screening method based on Gauss naive Bayes and prediction error method | |
| CN115495984A (en) | Industrial boiler combustion optimization method based on deep cycle neural network and evolutionary computation | |
| CN113837464A (en) | Load prediction method of cogeneration boiler based on CNN-LSTM-Attention | |
| CN110059821A (en) | Neural network optimization, device, computer equipment and storage medium | |
| CN117034767A (en) | Ceramic roller kiln temperature prediction method based on KPCA-GWO-GRU | |
| CN110207094A (en) | IQGA-SVR boiler heating surface fouling characteristics discrimination method based on principal component analysis | |
| CN111881120A (en) | Intelligent operation optimization method for boiler | |
| CN107742000B (en) | Boiler combustion oxygen content modeling method | |
| CN117454765A (en) | Copper smelting furnace spray gun service life prediction method based on IPSO-BP neural network | |
| CN116951398A (en) | Model optimization control method and system for reducing left and right steam temperature deviation of boiler reheater | |
| CN112270121A (en) | Temperature anomaly detection method for optimizing SVDD (singular value decomposition) by improving whale algorithm and application | |
| CN116484675A (en) | Crack propagation life prediction method and system for ship engine blade | |
| CN111428906B (en) | Industrial boiler steam volume prediction method based on image transformation | |
| CN115809744A (en) | Fly ash carbon content prediction method based on genetic simulated annealing parameter optimization | |
| CN116307507A (en) | Multi-target particle swarm coal blending method based on superheater wall temperature prediction | |
| CN113947332B (en) | Underground engineering comprehensive security capability assessment method and system | |
| CN112396101A (en) | Intelligent image analysis and decision-making method based on firepower plant hearth flame television | |
| CN113759709A (en) | Method and device for training strategy model, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |