CN112163376A - Extreme random tree furnace temperature prediction control method based on longicorn stigma search - Google Patents
Extreme random tree furnace temperature prediction control method based on longicorn stigma search Download PDFInfo
- Publication number
- CN112163376A CN112163376A CN202011072976.1A CN202011072976A CN112163376A CN 112163376 A CN112163376 A CN 112163376A CN 202011072976 A CN202011072976 A CN 202011072976A CN 112163376 A CN112163376 A CN 112163376A
- Authority
- CN
- China
- Prior art keywords
- longicorn
- prediction
- iteration
- node
- value
- 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.)
- Granted
Links
- 241001481710 Cerambycidae Species 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000002485 combustion reaction Methods 0.000 claims abstract description 16
- 238000010845 search algorithm Methods 0.000 claims abstract description 12
- 230000000694 effects Effects 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 31
- 238000012360 testing method Methods 0.000 claims description 21
- 238000003066 decision tree Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 7
- 238000002790 cross-validation Methods 0.000 claims description 6
- 238000005096 rolling process Methods 0.000 claims description 5
- 238000012937 correction Methods 0.000 claims description 4
- 230000009191 jumping Effects 0.000 claims description 4
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 3
- 229910052760 oxygen Inorganic materials 0.000 claims description 3
- 239000001301 oxygen Substances 0.000 claims description 3
- 238000011158 quantitative evaluation Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 239000002912 waste gas Substances 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims description 2
- 239000002737 fuel gas Substances 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 238000004088 simulation Methods 0.000 claims description 2
- 230000007717 exclusion Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 12
- 230000009286 beneficial effect Effects 0.000 description 4
- 239000007789 gas Substances 0.000 description 4
- 239000010815 organic waste Substances 0.000 description 3
- 239000012855 volatile organic compound Substances 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000003776 cleavage reaction Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004043 dyeing Methods 0.000 description 1
- 238000003933 environmental pollution control Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000007800 oxidant agent Substances 0.000 description 1
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 239000003973 paint Substances 0.000 description 1
- 230000001172 regenerating effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000007017 scission Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Feedback Control In General (AREA)
Abstract
An extreme random tree furnace temperature prediction control method based on longicorn stigma search belongs to the field of furnace temperature prediction and control in an industrial combustion process. By establishing a regression relationship between key variables influencing the temperature and the temperature of the combustion chamber, the predicted value of the furnace temperature in the combustion chamber is obtained in real time, and the method enables the prediction accuracy of the furnace temperature to reach +/-1.5 ℃. Considering that the control precision of the original system is not high, a search algorithm of the longicorn stigma is designed at the same time, a secondary performance index function commonly used in prediction control is selected as a fitness function of the longicorn, and the optimal control quantity is obtained through searching through an olfactory search mechanism of the algorithm, so that the control effect of the whole system is better.
Description
Technical Field
The invention belongs to the field of furnace temperature prediction and control in an industrial combustion process, and particularly relates to an extreme random tree prediction control method based on longicorn stigma search.
Background
With the increasing strictness of environmental pollution control, the effective treatment of low-concentration Volatile Organic Compounds (VOCs) discharged from petrochemical, paper, paint, printing and dyeing and other industrial production has received high domestic attention. At present, the most common equipment for treating VOCs in China is a Regenerative Thermal Oxidizer (RTO). The conventional RTO adopts simple PID control for the temperature in the combustion chamber, while the oxidation furnace is essentially a nonlinear system with large inertia, large hysteresis and multiple variables, and when the working condition in the combustion chamber changes, the temperature fluctuation has great influence on the conversion rate of the organic waste gas combustion reaction, so that the furnace temperature in the combustion chamber needs to be accurately controlled.
The predictive control has good control effect and strong robustness, and is widely applied to a nonlinear system. The method has the main idea that the system is modeled, the future output state of the system is predicted, the predicted value is subjected to feedback correction and then is compared with reference input, quadratic performance indexes are applied to rolling optimization, the optimal control quantity is obtained, and the whole cycle is completed. The classical predictive control algorithm generally establishes a model of a system by obtaining step or impulse response parameters of the system to obtain an approximately linear model, and for a nonlinear system with more interference, the problem of low accuracy of the system model exists.
In recent years, methods of machine learning have been developed rapidly and widely used in model prediction. The key parameter in the combustion control system is temperature, and accurate prediction and control of the temperature are important indexes. The extreme random tree (ET) algorithm has good generalization and nonlinear modeling capability and can flexibly process various types of data.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a combustion chamber furnace temperature prediction method based on an extreme random tree algorithm, which is used for obtaining a predicted value of the furnace temperature in a combustion chamber in real time by establishing a regression relation between a key variable influencing the temperature and the combustion chamber temperature, and the method enables the prediction accuracy of the furnace temperature to reach +/-1.5 ℃. Considering that the control precision of the original system is not high, a Beatle Antenna Search (BAS) algorithm is designed at the same time, a secondary performance index function commonly used in prediction control is selected as a fitness function of the Beatle, and the optimal control quantity is obtained through searching through an olfactory Search mechanism of the algorithm, so that the control effect of the whole system is better.
The technical scheme adopted by the invention is as follows: the furnace temperature prediction part adopts an extreme random tree model prediction algorithm, the control quantity optimization part adopts a longicorn beard search algorithm, and the method is mainly realized by the following steps:
Step 3, training all training sets by using an extreme random tree ET algorithm, and establishing a regression prediction model;
and 4, carrying out simulation prediction on the test set by using the established regression prediction model to obtain a predicted value of the hearth temperature, introducing a feedback correction link for overcoming the deviation between the predicted output and the actual output, and correcting the predicted furnace temperature.
And 5, taking the set temperature of the hearth, the corrected predicted temperature and the historical control quantity as input, solving an objective function J according to a longicorn stigma search algorithm to obtain an optimal fuel gas control quantity u (k), and realizing rolling optimization of the control quantity.
The invention has the beneficial effects that: the invention realizes furnace temperature prediction by modeling the nonlinear relation between the furnace temperature and the valve opening degree in the organic waste gas combustion process based on the extreme random tree algorithm, and provides a method for predicting and controlling the furnace temperature in the organic waste gas combustion process by carrying out online rolling optimization on the control variable through the longicorn stigma search algorithm.
On the basis of the above technical solution, the solution will be described in further detail.
The cross validation method in the step 2 comprises the following specific steps:
step 2.1 mixingThe initial data set O is hierarchically sampled for k times to obtain k mutually exclusive subsets with similar sizes,
step 2.2 use the union of k-1 mutually exclusive subsets as training set each time, the remaining one as test set. And obtaining k groups of training sets and test sets in total, so as to carry out k times of modeling and finally output the mean value of the test result.
The beneficial effect of above-mentioned scheme is: by hierarchical sampling, so that each subset DiThe consistency of the data distribution is maintained as much as possible. A cross-validation method based on a random function is adopted, and a data set is divided into a training set and a testing set by setting the proportion of the testing set in the training set and the seeds of a random tree. The randomness of the training set and the testing set is guaranteed, the accuracy of training and prediction can be improved, over-fitting and under-fitting are avoided, and the reliability of a temperature predicted value is guaranteed.
Further, the ET algorithm training process described in step 3 is as follows:
for the acquired dataset Ω { (A)1,y1),(A2,y2)…(Ai,yi)…(AN,yN) In which A isiIs a 1 x 5 dimensional row vector, input for the modeling sample, yiIs AiThe real output value of the corresponding sample, i is 1,2 … N, N is the number of sample groups, and the data is trained based on the extreme random tree algorithm, and the specific training process is as follows:
step 3.1 gives a training data set. Selecting N in the acquired data set omega by a cross verification method1Omega corresponding to group data1The data set is trained as a training set. Setting a maximum iteration number T, preferably 100, and setting the initialization iteration number T to be 1;
and 3.2, randomly dividing the node data set. For the t-th iteration, from training set N1Randomly selecting sigma characteristic sequences { a1,a2,…aσTaking it as initial node, and randomly selecting a corresponding group from the selected featuresSequence s1,s2,…,sσAnd f, taking the value range of the threshold as a threshold of the characteristic attribute, wherein the value range of the threshold is between the maximum value and the minimum value of the characteristic. A in the characteristic sequencexIs greater than sxThe x is 1,2, …, the samples of the sigma are assigned to the left node of the branch, and the rest samples are assigned to the right node, so that the random division under the node is realized;
and 3.3, acquiring the optimal splitting value under the random characteristic. The root mean square error is used as a quantitative evaluation criterion for the numerical characteristics of the system. And traversing all the characteristic values on the left node and the right node, calculating the root mean square error of each characteristic value on the corresponding node, and selecting the characteristic value corresponding to the minimum root mean square error as a split point. By the feature ai∈{a1,a2,…aσFor example, after being divided in step 3.2, the samples at the left and right sides of the feature are respectively ail＝(α1,α2,…,αi,…,αp)，air＝(β1,β2,…,βj,…,βq) Wherein: alpha is alphaiFor the samples assigned to the left, i ═ 1,2, … p; beta is ajFor the samples assigned to the right, j ═ 1,2, … q; p and q are the number of samples on the left and right sides, respectively. Taking the left node as an example, the Root Mean Square Error (RMSE) corresponding to each characteristic valueiComprises the following steps:
wherein
Selecting the feature with the minimum root mean square error for splitting;
and 3.4, judging whether the splitting end condition is met or not, if so, jumping to the step 3.5, otherwise, updating the iteration times, executing t which is t +1, and repeating the steps 3.2-3.3. The cleavage termination conditions are as follows (any condition may be satisfied):
(1) the maximum depth of the tree (max _ depth), preferably 50;
(2) the minimum number of samples (min _ samples _ split) required for internal node subdivision is preferably 2;
(3) the minimum number of samples (min _ samples _ leaf) of a leaf node is preferably 1.
Step 3.5, the splitting is finished, and the decision tree f obtained by the t iteration is obtainedt。
Step 3.6, if T is T, go to step 3.7, otherwise, T is T +1, go to step 3.2;
step 3.7 combine the generated T decision trees by using a mean value method to obtain a prediction model
Comprises the following steps:
inputting the test set into a prediction model to obtain a prediction output
N2Is the number of groups of the test set, and measures the prediction effect by using the corresponding evaluation index.
The beneficial effect of above-mentioned scheme is: as the characteristics of the algorithm are randomly selected, the splitting threshold is also randomly set, and all samples are adopted for establishing each decision tree, the randomness of training based on the extreme random tree is stronger, the prediction effect of the trained regression prediction model is better, and the accuracy of the obtained furnace temperature prediction value is higher.
The longicorn stigma search algorithm in the step 5 comprises the following specific steps:
step 5.1 gives the corresponding objective function:
wherein: t isrIs input into a reference end; t ispOutputting for the corrected prediction; u is a control variable; lambda is a weight coefficient (0 < lambda < 1); d is the predicted step number.
Define e as the model error vector:
e＝[Tr(k+1)-Tp(k+1),…,Tr(k+d)-Tp(k+d)]T
define Δ u as the increment of the control quantity:
Δu＝[u(k)-u(k-1),…,u(k+d-1)-u(k+d-2)]T
according to the above two formulae, f is defined as:
f＝[eT λΔuT]T。
screening a d-dimensional control vector u*Let the objective function J reach the minimum value, and obtain the mathematical formula when J is the minimum value through the definitional formula of f: min j ═ fT×f。
And 5.2, reading data after the execution of the extreme random tree ET algorithm at the last moment, substituting the data into the objective function, and initializing parameters of the longicorn stigma search algorithm. The method mainly comprises the following steps: the dimension of the problem is D and the distance between two whiskers is D0The initial step size is, the attenuation coefficient of the distance between two whiskers is eta _ D0The attenuation coefficient of the step length is eta _, the iteration number is n, and the centroid coordinate is as follows:
wherein: u. of0Is a random initial value in (0,1) for the quantity to be optimized in the objective function; rands () represents a randomly generated vector.
Step 5.3 if the left whisker coordinate is expressed as uleftThe coordinates of the right whisker are expressed as urightThen, the relationship between the left and right whiskers can be expressed as: u. ofleft-uright＝D0Dir, dir represents a normalized random vector, and after l iterations, the two whiskers of coordinates are:
wherein: u. of0 lIs the centroid coordinate at the first iteration; d0 lThe distance between two whiskers in the first iteration is shown;
substituting the objective function to determine the leftOdor intensity of the right two whiskers: j. the design is a squareleft＝J(uleft)、Jright＝J(uright) And comparing the sizes of the two.
If Jleft＜JrightThen, the longicorn advances to the left direction, and the centroid coordinate is:
u0 l+1＝u0 l+l·dir(Jleft-Jright)
if Jleft＞JrightThen, the longicorn advances to the right direction, and the centroid coordinate is:
u0 l+1＝u0 l-l·dir(Jleft-Jright)
the two equations above can be uniformly written using the sign function sign:
u0 l+1＝u0 l-l·dir·sign(Jleft-Jright)
step 5.4, the odor intensity after the (l +1) th iteration is determined, and the step length and the distance between the two whiskers at the moment are updated:
l+1＝eta_·l
D0 l+1＝eta_D0·D0 l
wherein:l+1the moving step length of the longicorn at the l +1 iteration; d0 l+1The distance between two whiskers of the longicorn at the l +1 iteration.
Step 5.5, judging whether the iteration cutoff condition is reached: and when a certain iteration number n is 5000, or the barycentric coordinate is continuously unchanged for 200 times. If the iteration cutoff condition is met, outputting a global optimal value u*Namely, the optimal control quantity is obtained,
if not, the steps 5.2 to 5.4 are required to be repeated until the conditions are met.
The beneficial effect of above-mentioned scheme is: as the characteristics of the algorithm are randomly selected, the splitting threshold is also randomly set, and all samples are adopted for establishing each decision tree, the randomness of training based on the extreme random tree is stronger, the prediction effect of the trained regression prediction model is better, and the accuracy of the obtained furnace temperature prediction value is higher.
Drawings
FIG. 1 is a block diagram of an extreme random tree furnace temperature predictive control system based on longicorn whisker search.
FIG. 2 is an overall flow chart of extreme random tree predictive control based on longicorn whisker search.
Fig. 3 is a flow chart of the principle of the extreme random tree algorithm.
Fig. 4 is a schematic flow chart of a longicorn stigma search algorithm.
Fig. 5 is a comparison graph of the prediction error effect of the 4 algorithms. Wherein (a) is an RF prediction error map; (b) predicting an error map for the KNN; (c) predicting an error map for the ET; (d) an error map is predicted for the SVR.
FIG. 6 is a control strategy comparison graph.
Detailed Description
The following further describes the embodiments of the present invention with reference to the drawings.
In the attached figure 1, the prediction model is generated by an extreme random tree algorithm, and the input comprises gas control quantity u (k), temperature T (k) and M (k), wherein M (k) is a vector group with 1 x 3 dimensions and consists of air volume, waste gas concentration and oxygen content. Output of the prediction model Tm(k +1) is subjected to a feedback correction to obtain Tp(k +1) is input to an optimization stage, which performs a rolling optimization of the control variables based on a search for the longicorn stigma, where TrAnd (k +1) is input at the reference end.
The specific process of modeling the furnace temperature based on the extreme random tree algorithm optimized by the longicorn stigma comprises the following steps:
step 3, dividing the features by taking the root mean square error as a quantitative evaluation criterion, and selecting the features with the minimum root mean square error for division;
and 4, judging whether the splitting end condition is met, if so, jumping to the step 5, otherwise, jumping to the step 3. The conditions satisfying the end of splitting are set as follows: the maximum depth of the tree (max _ depth) is 50; the internal node subdivides the required minimum sample number (min _ samples _ split) to 2; the leaf node minimum sample number (min _ samples _ leaf) is 1.
Step 5, finishing splitting, and obtaining a decision tree f obtained by the t iterationt。
step 7, combining the generated T decision trees by using a mean value method to obtain a prediction model
Comprises the following steps:
step 8 gives the corresponding objective function:
obtaining optimal u by utilizing longicorn stigma search algorithm*。
Example (b):
by adopting the extreme random tree furnace temperature prediction control method based on longicorn stigma search, field data of 5 days are randomly derived from a database, and the fact that the variation range of natural gas flow and furnace temperature is large in 8-month-10-day 2018 is found, and almost all possible working conditions on the field are covered. General day32600 groups of data are collected in total, and the sampling period is T-0.1 s. 3000 groups are randomly selected from the 32600 groups of data, and 2700 groups are selected as a training set omega of an extreme random tree algorithm through a cross verification method1＝{(A1,y1),(A2,y2)…(A2700,y2700)}，AiIs a row vector of 1 x 5 dimensions, a set of input quantities, y, for the modeling sampleiIs AiThe true output of the corresponding modeled sample, i ═ 1,2, …, 2700; the remaining 300 groups were used as test set omega2＝{(B1,y1),(B2,y2)…(B300,y300)}，BiIs a row vector of 1 x 5 dimensions, a set of input quantities, y, for the modeling sampleiIs BiThe corresponding real output of the modeled sample, i ═ 1,2, …, 300. The specific implementation mode is as follows:
initializing each parameter, setting the maximum iteration number T as 100, and setting the maximum characteristic number
The maximum depth of the tree (max _ depth) is 50; the internal node subdivides the required minimum sample number (min _ samples _ split) to 2; the leaf node minimum sample number (min _ samples _ leaf) is 1. Randomly selecting a feature from the input training set A as an initial node, and splitting according to the detailed step 3 to obtain a regression prediction model. The test data set is input, the model prediction effect is over-measured, and as can be seen from the attached figure 5, compared with SVR, KNN and RF, the extreme random tree furnace temperature prediction control method based on the longicorn stigma search has smaller prediction error and highest prediction precision.
And reading related data after the ET algorithm is executed at the last moment, substituting the data into the objective function, and initializing parameters of the longicorn stigma search algorithm. Setting the problem dimension d to be 3, the initial step size to be 30, the distance C between the initial step size and the two whiskers to be 4, the attenuation coefficient eta to be 0.95, and the iteration number n to be 5000. The longicorn is iteratively searched in the global scope, and finally the optimal Control quantity is output, as can be seen from the attached drawing 6, for a step signal randomly given by the system, the Control effect of the extreme random tree Predictive Control algorithm based on the longicorn whisker search is superior to that of Generalized Predictive Control (GPC) which is put into production on site, and under the condition that the temperature is mutated, the system can respond to the variable quantity more quickly by the extreme random tree Predictive Control method based on the longicorn whisker search, and has shorter rise time, smaller overshoot and smaller steady-state error.
The present invention is not intended to be limited to the particular embodiments shown above, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. An extreme random tree furnace temperature prediction control method based on longicorn stigma search is characterized by comprising the following steps:
step 1, collecting control input and output data of past and current moments in a combustion process as an original data set O;
step 2, dividing the original data set O into a training set N by adopting a cross validation mode1And test set N2；
Step 3, training all training sets by using an extreme random tree ET algorithm, and establishing a regression prediction model;
step 4, carrying out simulation prediction on the test set by using the established regression prediction model to obtain a predicted value of the hearth temperature, introducing a feedback correction link, and correcting the furnace temperature obtained by prediction;
and 5, taking the set temperature of the hearth, the corrected predicted temperature and the historical control quantity as input, solving an objective function J according to a longicorn stigma search algorithm to obtain an optimal fuel gas control quantity u (k), and realizing rolling optimization of the control quantity.
2. The extreme random tree furnace temperature prediction control method based on longicorn stigma search as claimed in claim 1, wherein the specific steps of the cross validation method in step 2 are as follows:
step 2.1, the original data set O is subjected to k times of layered sampling to obtain k mutual exclusions with similar sizesThe subset of the plurality of sets of data,
step 2.2, using the union set of k-1 mutually exclusive subsets as a training set each time, and using the rest subsets as a test set; and obtaining k groups of training sets and test sets in total, so as to carry out k times of modeling and finally output the mean value of the test result.
3. The method of claim 1, wherein Ω { (A) is used for the collected data set1,y1),(A2,y2)…(Ai,yi)…(AN,yN) In which A isiIs a 1 x 5 dimensional row vector, input for the modeling sample, yiIs AiThe real output value of the corresponding sample, i is 1,2 … N, N is the number of sample groups, and the data is trained based on the extreme random tree algorithm, and the specific training process is as follows:
step 3.1, giving a training data set: selecting N in the acquired data set omega by a cross verification method1Omega corresponding to group data1As a training set, training the data set; setting a maximum iteration time T, and initializing the iteration time T to be 1;
step 3.2, randomly dividing the node data set: for the t-th iteration, from training set N1Randomly selecting sigma characteristic sequences { a1,a2,…aσAs the initial node, and randomly selecting a corresponding set of sequences s from the selected features1,s2,…,sσTaking the value range of the threshold value as the threshold value of the characteristic attribute, wherein the value range of the threshold value is between the maximum value and the minimum value of the characteristic; a in the characteristic sequencexIs greater than sxThe x is 1,2, …, the samples of the sigma are assigned to the left node of the branch, and the rest samples are assigned to the right node, so that the random division under the node is realized;
step 3.3, obtaining the optimal splitting value under random characteristics: taking root mean square error as a quantitative evaluation criterion for the numerical characteristics of the system; traversing all characteristic values on the left node and the right node, calculating the root mean square error of each characteristic value on the corresponding node, and selecting the characteristic value corresponding to the minimum root mean square error as a split point;
step 3.4, judging whether the splitting end condition is met, if so, jumping to step 3.5, otherwise, updating the iteration times, executing t which is t +1, and repeating the steps 3.2-3.3; the splitting termination condition may satisfy any one of the following conditions:
(1) maximum depth of tree max _ depth;
(2) the internal node is subdivided into the required minimum sample number min _ samples _ split;
(3) the minimum sample number min _ samples _ leaf of the leaf node;
step 3.5, the splitting is finished, and the decision tree f obtained by the t iteration is obtainedt；
Step 3.6, if T is T, go to step 3.7, otherwise, T is T +1, go to step 3.2;
step 3.7, combining the generated T decision trees by using a mean value method to obtain a prediction model
Comprises the following steps:
inputting the test set into a prediction model to obtain a prediction output
N2Is the number of groups of the test set, and measures the prediction effect by using the corresponding evaluation index.
4. The method for controlling furnace temperature prediction of extreme random tree based on longicorn silk search as claimed in claim 3, wherein in step 3.3, when the selected characteristic is ai∈{a1,a2,…aσIn the case of (1), after being divided in the step 3.2, samples on the left and right sides of the feature are respectively ail＝(α1,α2,…,αi,…,αp)，air＝(β1,β2,…,βj,…,βq) Wherein: alpha is alphaiFor the samples assigned to the left, i ═ 1,2, … p; beta is ajFor the samples assigned to the right, j ═ 1,2, … q; p and q are the number of samples on the left side and the right side respectively; the Root Mean Square Error (RMSE) corresponding to each characteristic value of the left nodeiComprises the following steps:
wherein
Selecting the feature with the minimum root mean square error for splitting; the right node is computed in the same way as the left node.
5. The extreme random tree furnace temperature predictive control method based on longicorn stigma search as claimed in claim 1, wherein the concrete steps of the longicorn stigma search algorithm of step 5 are as follows:
step 5.1, a corresponding objective function is given:
wherein: t isrIs input into a reference end; t ispOutputting for the corrected prediction; u is a control variable; lambda is a weight coefficient, and lambda is more than 0 and less than 1; d is the predicted step number;
define e as the model error vector:
e＝[Tr(k+1)-Tp(k+1),…,Tr(k+d)-Tp(k+d)]T
define Δ u as the increment of the control quantity:
Δu＝[u(k)-u(k-1),…,u(k+d-1)-u(k+d-2)]T
according to the above two formulae, f is defined as:
f＝[eT λΔuT]T；
screening a d-dimensional control vector u*Let the objective function J reach the minimum value, and obtain the mathematical formula when J is the minimum value through the definitional formula of f: min J ═ fT×f；
Step 5.2, reading data after the execution of the extreme random tree ET algorithm at the last moment, substituting the data into a target function, and initializing parameters of the longicorn stigma search algorithm; the parameter initialization comprises the following steps: the dimension of the problem is D and the distance between two whiskers is D0The initial step size is, the attenuation coefficient of the distance between two whiskers is eta _ D0The attenuation coefficient of the step length is eta _, the iteration number is n, and the centroid coordinate is as follows:
wherein: u. of0Is a random initial value in (0,1) for the quantity to be optimized in the objective function; rands () represents a randomly generated vector;
step 5.3, let the left whisker coordinate be uleftThe coordinates of the right whisker are expressed as urightThen, the relationship between the left and right whiskers can be expressed as: u. ofleft-uright＝D0Dir, dir represents a normalized random vector, and after l iterations, the two whiskers of coordinates are:
wherein: u. of0 lIs the centroid coordinate at the first iteration; d0 lThe distance between two whiskers in the first iteration is shown;
substituting the objective function to determine the odor intensity of the left and right whiskers: j. the design is a squareleft＝J(uleft)、Jright＝J(uright) And comparing the sizes of the two;
if Jleft＜JrightThen, the longicorn advances to the left direction, and the centroid coordinate is:
u0 l+1＝u0 l+l·dir(Jleft-Jright)
if Jleft＞JrightThen, the longicorn advances to the right direction, and the centroid coordinate is:
u0 l+1＝u0 l-l·dir(Jleft-Jright)
the two equations above can be uniformly written using the sign function sign:
u0 l+1＝u0 l-l·dir·sign(Jleft-Jright)
and 5.4, determining the odor intensity after the (l +1) th iteration, and updating the step length and the distance between the two whiskers:
l+1＝eta_·l
D0 l+1＝eta_D0·D0 l
wherein:l+1the moving step length of the longicorn at the l +1 iteration; d0 l+1The distance between two whiskers of the longicorn at the l +1 iteration;
step 5.5, judging whether the iteration cutoff condition is reached: certain iteration times are achieved, or the coordinates of the mass center are continuously unchanged; if the iteration cutoff condition is met, outputting a global optimal value u*If the control quantity is not the optimal control quantity, the step 5.2-step 5.4 are required to be repeated until the condition is met.
6. The method for controlling furnace temperature prediction of an extreme random tree based on longicorn silk search as claimed in claim 1, wherein the data collected in step 1 comprises: waste gas concentration, oxygen content, air quantity, proportional valve opening degree and temperature.
7. The method for controlling furnace temperature prediction of an extreme random tree based on longicorn silk search as claimed in claim 1, wherein in step 3.1, the maximum number of iterations is 100.
8. The method for controlling furnace temperature prediction of an extreme random tree based on longicorn stigma search as claimed in claim 1, wherein in step 3.4, the splitting end condition satisfies any one of the following conditions:
(1) the maximum depth max _ depth of the tree is 50;
(2) the internal node is subdivided into the required minimum sample number min _ samples _ split ═ 2;
(3) the minimum number of samples min _ samples _ leaf of the leaf node is 1.
9. The extreme random tree furnace temperature predictive control method based on longicorn silk search as claimed in claim 1, wherein in step 5.5, the number of iterations reached is 5000, or the centroid coordinate is continuously cut off for 200 times of unchanged iteration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011072976.1A CN112163376B (en) | 2020-10-09 | 2020-10-09 | Extreme random tree furnace temperature prediction control method based on longhorn beetle whisker search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011072976.1A CN112163376B (en) | 2020-10-09 | 2020-10-09 | Extreme random tree furnace temperature prediction control method based on longhorn beetle whisker search |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112163376A true CN112163376A (en) | 2021-01-01 |
| CN112163376B CN112163376B (en) | 2024-03-12 |
Family
ID=73866416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011072976.1A Active CN112163376B (en) | 2020-10-09 | 2020-10-09 | Extreme random tree furnace temperature prediction control method based on longhorn beetle whisker search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112163376B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113069124A (en) * | 2021-03-09 | 2021-07-06 | 浙江工业大学 | Electrocardio monitoring method based on CNN-ET model |
| CN114719333A (en) * | 2022-04-14 | 2022-07-08 | 兰州理工大学 | Multi-inverter electromagnetic heating system and output power prediction control method thereof |
| CN115028301A (en) * | 2022-05-31 | 2022-09-09 | 河北工程大学 | Intelligent purified swimming pool water recycling system and method |
| CN115186582A (en) * | 2022-07-05 | 2022-10-14 | 科大智能物联技术股份有限公司 | Steel rolling heating furnace control method based on machine learning model |
| CN116625134A (en) * | 2023-07-24 | 2023-08-22 | 苏州弘皓光电科技有限公司 | Electric furnace temperature monitoring control method and system based on 5G technology |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101457264A (en) * | 2008-12-29 | 2009-06-17 | 杭州电子科技大学 | Blast furnace temperature optimization control method |
| US20120150507A1 (en) * | 2009-06-24 | 2012-06-14 | Abb Research Ltd | Estimating initial states of a system model for controlling an industrial process |
| CN109473148A (en) * | 2018-10-26 | 2019-03-15 | 武汉工程大学 | A kind of ion concentration prediction technique, device and computer storage medium |
| CN109709795A (en) * | 2018-12-24 | 2019-05-03 | 东华大学 | A kind of PID controller parameter setting method based on longicorn palpus searching algorithm |
| CN110175195A (en) * | 2019-04-23 | 2019-08-27 | 哈尔滨工业大学 | Mixed gas detection model construction method based on extreme random tree |
| CN110532674A (en) * | 2019-08-27 | 2019-12-03 | 东北电力大学 | A kind of coal-fired power station boiler fire box temperature measurement method |
| WO2020000248A1 (en) * | 2018-06-27 | 2020-01-02 | 大连理工大学 | Space reconstruction based method for predicting key performance parameters of transition state acceleration process of aircraft engine |
| CN110766137A (en) * | 2019-10-18 | 2020-02-07 | 武汉大学 | Power electronic circuit fault diagnosis method based on longicorn whisker optimized deep confidence network algorithm |
| CN110783962A (en) * | 2019-04-17 | 2020-02-11 | 上海电机学院 | Control method of virtual synchronous generator grid-connected inverter |
| CN110852902A (en) * | 2019-11-14 | 2020-02-28 | 合肥工业大学 | Photovoltaic power generation power prediction method based on BAS-BP |
| CN111158237A (en) * | 2019-12-25 | 2020-05-15 | 南京理工大学 | Industrial furnace temperature multi-step prediction control method based on neural network |
| CN111173573A (en) * | 2020-01-08 | 2020-05-19 | 上海电力大学 | Identification method for power object model of steam turbine regulating system |
| CN111562744A (en) * | 2020-06-15 | 2020-08-21 | 安徽工业大学 | Boiler combustion implicit generalized predictive control method based on PSO algorithm |
-
2020
- 2020-10-09 CN CN202011072976.1A patent/CN112163376B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101457264A (en) * | 2008-12-29 | 2009-06-17 | 杭州电子科技大学 | Blast furnace temperature optimization control method |
| US20120150507A1 (en) * | 2009-06-24 | 2012-06-14 | Abb Research Ltd | Estimating initial states of a system model for controlling an industrial process |
| WO2020000248A1 (en) * | 2018-06-27 | 2020-01-02 | 大连理工大学 | Space reconstruction based method for predicting key performance parameters of transition state acceleration process of aircraft engine |
| CN109473148A (en) * | 2018-10-26 | 2019-03-15 | 武汉工程大学 | A kind of ion concentration prediction technique, device and computer storage medium |
| CN109709795A (en) * | 2018-12-24 | 2019-05-03 | 东华大学 | A kind of PID controller parameter setting method based on longicorn palpus searching algorithm |
| CN110783962A (en) * | 2019-04-17 | 2020-02-11 | 上海电机学院 | Control method of virtual synchronous generator grid-connected inverter |
| CN110175195A (en) * | 2019-04-23 | 2019-08-27 | 哈尔滨工业大学 | Mixed gas detection model construction method based on extreme random tree |
| CN110532674A (en) * | 2019-08-27 | 2019-12-03 | 东北电力大学 | A kind of coal-fired power station boiler fire box temperature measurement method |
| CN110766137A (en) * | 2019-10-18 | 2020-02-07 | 武汉大学 | Power electronic circuit fault diagnosis method based on longicorn whisker optimized deep confidence network algorithm |
| CN110852902A (en) * | 2019-11-14 | 2020-02-28 | 合肥工业大学 | Photovoltaic power generation power prediction method based on BAS-BP |
| CN111158237A (en) * | 2019-12-25 | 2020-05-15 | 南京理工大学 | Industrial furnace temperature multi-step prediction control method based on neural network |
| CN111173573A (en) * | 2020-01-08 | 2020-05-19 | 上海电力大学 | Identification method for power object model of steam turbine regulating system |
| CN111562744A (en) * | 2020-06-15 | 2020-08-21 | 安徽工业大学 | Boiler combustion implicit generalized predictive control method based on PSO algorithm |
Non-Patent Citations (8)
| Title |
|---|
| XIAOJUN WANG: "Ladle furnace temperature prediction model based on large-scale data with random forest", pages 770 - 774 * |
| XINGRUI CHEN.ET AL: "Load Forecasting Based On Elman Neural Network Optimized By Beetle Antennae Search Optimization", pages 1 - 6 * |
| 应健锋等: "一种多Ｘ值输入下测试覆盖率损失的预测方法", vol. 37, no. 4, pages 12 - 18 * |
| 李大中等: "一种利用混合算法选择变量的天牛须优化神经网络风速预测方法", vol. 35, no. 12, pages 28 - 33 * |
| 瞿红春等: "基于IBAS-Elman网络的滚动轴承故障诊断研究", vol. 48, no. 16, pages 201 - 205 * |
| 薛美盛等: "神经网络预测控制在加热炉炉温控制中 的仿真研究", vol. 45, no. 8, pages 590 - 594 * |
| 薛美盛等: "神经网络预测控制在加热炉炉温控制中的仿真研究", vol. 45, no. 8, pages 590 - 594 * |
| 韩瑞等: "基于Boruta和极端随机树方法的森林蓄积量估测", no. 4, pages 127 - 133 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113069124A (en) * | 2021-03-09 | 2021-07-06 | 浙江工业大学 | Electrocardio monitoring method based on CNN-ET model |
| CN114719333A (en) * | 2022-04-14 | 2022-07-08 | 兰州理工大学 | Multi-inverter electromagnetic heating system and output power prediction control method thereof |
| CN115028301A (en) * | 2022-05-31 | 2022-09-09 | 河北工程大学 | Intelligent purified swimming pool water recycling system and method |
| CN115028301B (en) * | 2022-05-31 | 2024-04-26 | 河北工程大学 | Intelligent purifying swimming pool water recycling system and method |
| CN115186582A (en) * | 2022-07-05 | 2022-10-14 | 科大智能物联技术股份有限公司 | Steel rolling heating furnace control method based on machine learning model |
| CN116625134A (en) * | 2023-07-24 | 2023-08-22 | 苏州弘皓光电科技有限公司 | Electric furnace temperature monitoring control method and system based on 5G technology |
| CN116625134B (en) * | 2023-07-24 | 2023-10-24 | 苏州弘皓光电科技有限公司 | Electric furnace temperature monitoring control method and system based on 5G technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112163376B (en) | 2024-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112163376B (en) | Extreme random tree furnace temperature prediction control method based on longhorn beetle whisker search | |
| CN111738477B (en) | Power grid new energy consumption capability prediction method based on deep feature combination | |
| CN110674965A (en) | Multi-time step wind power prediction method based on dynamic feature selection | |
| Meddage et al. | Tree-based regression models for predicting external wind pressure of a building with an unconventional configuration | |
| CN114239400A (en) | Multi-working-condition process self-adaptive soft measurement modeling method based on local double-weighted probability hidden variable regression model | |
| CN114169434A (en) | Load prediction method | |
| CN117034767A (en) | Ceramic roller kiln temperature prediction method based on KPCA-GWO-GRU | |
| CN112765902A (en) | RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof | |
| Chhabra et al. | Optimizing design parameters of fuzzy model based cocomo using genetic algorithms | |
| CN116245019A (en) | Load prediction method, system, device and storage medium based on Bagging sampling and improved random forest algorithm | |
| CN113204918B (en) | Modeling method for cement kiln production energy consumption | |
| Marenbach et al. | Data-driven structured modelling of a biotechnological fed-batch fermentation by means of genetic programming | |
| CN110276478B (en) | Short-term wind power prediction method based on segmented ant colony algorithm optimization SVM | |
| CN112561200A (en) | Wind power station output hybrid prediction technology based on complete set empirical mode decomposition and improved ant colony optimization long-short term memory network | |
| CN115829099A (en) | Yellow wine fermentation prediction method and system based on ANFIS and random fractal search algorithm | |
| TWI735385B (en) | Method and sintering factory for predicting sulfur oxide | |
| CN114202063A (en) | Fuzzy neural network greenhouse temperature prediction method based on genetic algorithm optimization | |
| CN114818203A (en) | Reducer design method based on SWA algorithm | |
| CN111582567B (en) | Wind power probability prediction method based on hierarchical integration | |
| CN110245400B (en) | Identification method of oxygen object model of boiler combustion system | |
| CN109711040B (en) | Intelligent industrial design reinforcement learning algorithm based on search direction learning | |
| CN113610260A (en) | Method for predicting concentration of flue gas components in incineration process of municipal domestic waste | |
| CN112801356A (en) | Power load prediction method based on MA-LSSVM | |
| CN114282614B (en) | Medium-long runoff prediction method for optimizing CNN-GRU based on random forest and IFDA | |
| CN112086135B (en) | State analysis method, device and equipment |
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 |