CN112200735A - Temperature identification method based on flame image and control method of low-concentration gas combustion system - Google Patents
Temperature identification method based on flame image and control method of low-concentration gas combustion system Download PDFInfo
- Publication number
- CN112200735A CN112200735A CN202010984130.9A CN202010984130A CN112200735A CN 112200735 A CN112200735 A CN 112200735A CN 202010984130 A CN202010984130 A CN 202010984130A CN 112200735 A CN112200735 A CN 112200735A
- Authority
- CN
- China
- Prior art keywords
- temperature
- combustion flame
- combustion
- hearth
- flame image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000002485 combustion reaction Methods 0.000 title claims abstract description 149
- 238000000034 method Methods 0.000 title claims abstract description 62
- 239000011159 matrix material Substances 0.000 claims description 37
- 238000001914 filtration Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 14
- 238000002790 cross-validation Methods 0.000 claims description 11
- 238000010845 search algorithm Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 2
- 239000007789 gas Substances 0.000 description 14
- 238000012360 testing method Methods 0.000 description 7
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 3
- 229910002091 carbon monoxide Inorganic materials 0.000 description 3
- 230000000630 rising effect Effects 0.000 description 3
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 229910052760 oxygen Inorganic materials 0.000 description 2
- 239000001301 oxygen Substances 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009529 body temperature measurement Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N1/00—Regulating fuel supply
- F23N1/002—Regulating fuel supply using electronic means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K13/00—Thermometers specially adapted for specific purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- Radiation Pyrometers (AREA)
Abstract
The invention discloses a temperature identification method based on flame images and a control method of a low-concentration gas combustion system, wherein the temperature identification method comprises the following steps of: s1, acquiring a combustion video in the hearth, and capturing a picture to obtain a plurality of combustion flame images in the hearth, wherein the images correspond to the temperatures at corresponding moments one by one; s2, denoising; s3, extracting the characteristic quantity; and S4, performing data fitting by using the established corresponding relation between the characteristic quantity and the temperature, and then inputting the combustion flame image to be identified, namely directly extracting the temperature corresponding to the combustion flame image to be identified. The invention applies the image processing and recognition technology to the direct combustion system of the low-concentration gas, judges the combustion state of the hearth according to the temperature information obtained from the continuous pictures, and can realize the control of the air input according to the combustion state, thereby leading the combustion state of the hearth to be more reasonable.
Description
Technical Field
The invention relates to the technical field of boiler high-temperature measurement, in particular to a temperature identification method based on flame images and a control method of a low-concentration gas combustion system.
Background
The low-concentration gas direct combustion system needs to control the temperature to be between 900 ℃ and 1100 ℃, and when the temperature is lower than 800 ℃, the temperature protection system can be triggered, so that the safe and stable operation of the system is guaranteed. The temperature is too low, so that the newly-entered gas cannot be ensured to be ignited and combusted in time, and incomplete combustion and combustible gas accumulation are easy to generate. Once the gas is not ignited in time, an explosion of the combustion chamber occurs. In order to avoid this, it is necessary to grasp the state of gas combustion in the furnace in time, and the determination of the gas combustion state requires temperature.
The temperature measuring method mainly used at present is to install a temperature sensing device in a hearth and obtain temperature related information through the temperature sensing device. But these components are close to or in direct contact with the flame in a manner that greatly reduces the lifetime of the components. The temperature is identified by an image identification method, the temperature cannot be directly contacted with flame, meanwhile, the condition that the temperature is misread due to interference of components is reduced, but the research on the aspect is lacked at present.
Disclosure of Invention
The invention aims to solve the technical problems that a temperature identification method based on a flame image and a control method of a low-concentration gas combustion system are provided, and aims to solve the problems that components are lost and the temperature of a hearth cannot be accurately identified in the existing method.
In order to solve the technical problems, the invention adopts the following technical scheme:
a temperature identification method based on flame images comprises the following steps:
s1, acquiring a combustion video in the hearth, detecting the temperature in the hearth in real time through a temperature sensing detector, capturing the acquired combustion video in the hearth by taking the T time as the granularity to obtain a plurality of combustion flame images in the hearth, and corresponding the combustion flame images to the temperatures detected by the temperature sensing detector at corresponding moments one by one;
s2, denoising each combustion flame image by using a self-adaptive weighted mean-median filtering method;
s3, extracting characteristic quantities of the denoised combustion flame images of the combustion flame images, and then enabling the temperatures corresponding to the combustion flame images of the combustion flame images to correspond to the extracted characteristic quantities one by one;
and S4, performing data fitting by using the established corresponding relation between the characteristic quantity and the temperature, and then inputting the combustion flame image to be identified, namely directly extracting the temperature corresponding to the combustion flame image to be identified.
Further, in step S1, the specific operation method of screenshot is as follows:
the image acquisition is carried out by using a video Capture module in a cv2 package, and combustion flame images at different time intervals are obtained by changing delay time.
Further, the specific operation method of step S2 is as follows:
s21, converting the combustion flame image into an RGB three-channel image to obtain a R, G, B matrix of the RGB three-channel image, and then converting the R, G, B matrix into a gray matrix;
s22, denoising pulse noise and Gaussian noise in the combustion flame image noise by using a self-adaptive weighted mean-median filtering method based on the gray matrix corresponding to the combustion flame image, wherein the denoising calculation formula is shown as formula 1, formula 2 and formula 3:
the data to be denoised is represented by zxy, namely the data at the center of the window, mean represents the mean value of the window data, med represents the median value of the window data, and zxy' represents the corrected data.
Further, in step S3, a specific operation method of extracting the feature amount is as follows:
s31, cutting and dividing the combustion flame image along more than two transverse cutting lines and more than two longitudinal cutting lines, wherein the gray coefficient of the gray matrix corresponding to each intersection point of the more than two transverse cutting lines and the more than two longitudinal cutting lines is the characteristic quantity of the combustion flame image;
s32, extracting characteristic quantities of all the combustion flame images according to the steps to obtain characteristic quantities corresponding to different temperatures, wherein the coefficient matrix corresponding to the characteristic quantity of each combustion flame image and the coefficient matrix formed by the corresponding temperatures of all the combustion flame images are as shown in the formula 4:
wherein x is11、x12…xndThe gray scale coefficients corresponding to the respective intersections are shown, the subscript nd represents the number of rows and columns of the cutting lines, the input represents a coefficient matrix corresponding to the characteristic amount of the combustion flame image, wendu1、 wendu2…wendunThe temperatures corresponding to all the combustion flame images are indicated, and the output indicates a coefficient matrix formed by the corresponding temperatures of all the combustion flame images.
Further, the specific operation method of step S4 is as follows:
s41, taking the characteristic quantity corresponding to the combustion flame image as the input of the KNN algorithm, taking the temperature corresponding to the combustion flame image as the output of the KNN algorithm, and fitting the data;
and S42, inputting the combustion flame image to be recognized, and calling the fitted algorithm to directly obtain the temperature corresponding to the combustion flame image to be recognized.
Further, in the step S4, the fitting of the data includes the following steps:
optimizing the k parameter of the KNN algorithm by using a grid search algorithm and cross validation, setting the parameters of the cross validation, dividing the data into a training set and a validation set, and finding out a k value which enables the highest classification accuracy by using the data of the training set in the range of 1-11; the accuracy of the algorithm is verified using the data of the remaining verification set.
Further, the algorithm in step S41 is implemented as follows:
s411, arbitrarily selecting a point x, and expressing the coordinate of the point x as (x)1,x2);
S412, find the euclidean distance between the remaining points and x, as shown in equation 5 below:
s413, finding out k points nearest to X according to the k parameters of the KNN algorithm, and recording the k points as y1、y2…ykThe coordinates of the k points are (y)11,y12)、(y21,y22)…(yk1,yk2);
And S414, averaging the gray coefficients corresponding to the coordinates of the k points to obtain the prediction result of the x.
A control method of a low-concentration gas combustion system comprises the following steps:
w1, acquiring a plurality of continuous combustion flame images in a hearth, and obtaining the corresponding temperatures of the plurality of combustion flame images by using the flame image-based temperature identification method according to any one of claims 1 to 7;
w2, judging the combustion state of the hearth according to the temperature change trend corresponding to the obtained multiple combustion flame images, and finally adjusting the air inflow according to the combustion state of the hearth.
Further, the combustion state of the hearth is judged according to the temperature information corresponding to the multiple combustion flame images, a reasonable range of the air intake amount when the temperature is a certain amount is obtained according to the corresponding relation between the temperature and the air intake amount when the hearth normally combusts, the air intake amount is increased in the reasonable range when the temperature of the hearth is in an increasing state, the air intake amount is reduced in the reasonable range when the temperature of the hearth is in a decreasing state, and the air intake amount is maintained to be unchanged when the temperature of the hearth is unchanged.
The invention has the beneficial effects that:
according to the invention, by establishing the relationship among the combustion flame image, the temperature and the characteristic quantity, the temperature is brought into a KNN algorithm which takes the characteristic quantity as input and takes the temperature corresponding to the combustion flame image as output, the temperature information can be directly read from the image, and the identification method is ensured to have very high accuracy by converting the temperature information into a gray matrix, reducing noise and the like, so that the problems that components are lost and the temperature of a hearth cannot be accurately identified in the conventional method can be effectively solved.
The invention applies the image processing and recognition technology to the direct combustion system of the low-concentration gas, judges the combustion state of the hearth according to the temperature information obtained from the continuous pictures, and can realize the control of the air input according to the combustion state, thereby leading the combustion state of the hearth to be more reasonable.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the structures shown in the drawings without any creative work.
FIG. 1 is a flow chart of an implementation of an embodiment of the present invention;
FIG. 2 is a graph of temperature change over time (time on the abscissa and temperature ℃ on the ordinate);
FIG. 3 is a furnace image after adaptive weighted mean-median processing;
fig. 4 is a flow chart of implementation of feature quantity extraction;
FIG. 5 is a KNN algorithm implementation flow chart;
fig. 6 is the error of the verification set of KNN algorithm.
FIG. 7 is a schematic diagram of cross-validation.
Fig. 8 shows the predicted temperatures for 6 consecutive images.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings, wherein technical solutions in embodiments of the present invention are clearly and completely described, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making an invasive task, are within the scope of the present invention.
In order to provide a safer, more reliable and more accurate method for obtaining temperature to match the control of the low-concentration gas combustion system, the invention designs a temperature identification method based on flame images, and the method and the overall implementation flow chart of the control method of the low-concentration gas combustion system are shown in fig. 1:
the invention relates to a temperature identification method based on flame images, which comprises the following steps:
s1, acquiring a combustion video in the hearth, detecting the temperature in the hearth in real time through a temperature sensing detector, capturing the acquired combustion video in the hearth by taking the T time as the granularity to obtain a plurality of (at least two) combustion flame images in the hearth, and corresponding each combustion flame image to the temperature detected by the temperature sensing detector at the corresponding moment one by one, wherein one combustion flame image corresponds to one temperature;
in specific implementation, a VideoCapture module in a cv2 packet can be used for acquiring images, combustion flame images at different time intervals are acquired by changing delay time, T can be selected to be 10 seconds, when the delay time is set to be 10s, the output images are images at ten-second time intervals, and of course T can also be selected to be other times;
s2, in order to reduce the interference of Gaussian noise and impulse noise in the combustion images of the hearth, denoising each combustion flame image by using a self-adaptive weighted mean-median filtering method, wherein the specific operation method comprises the following steps:
s21, converting the combustion flame image into an RGB three-channel image to obtain a R, G, B matrix of the RGB three-channel image, and then converting the R, G, B matrix into a gray matrix, so that one pixel point on the combustion flame image can only correspond to one gray coefficient;
the formula for converting the R, G, B matrix into a Gray scale matrix is Gray ═ R0.299 + G0.587 + B0.114, where: gray represents a converted Gray coefficient matrix, and R, G, B represents coefficient matrixes of three channels respectively;
s22, denoising pulse noise and Gaussian noise in the combustion flame image noise by using a self-adaptive weighted mean-median filtering method based on a gray matrix corresponding to the combustion flame image, restoring original information of the image, wherein denoising calculation formulas are shown as formulas 1, 2 and 3:
the method comprises the following steps that zxy represents data to be denoised, namely data in the center of a window, mean represents a window data mean value, med represents a window data median value, and zxy' represents corrected data; obtaining weights according to the distances between the median mean value and the data to be denoised respectively, so that the weights have the filtering characteristics of median filtering and mean filtering;
the image processing is not processed by a whole image together, but the data is divided into a plurality of small blocks for processing, wherein the window refers to a window with translation algorithm, the gray matrix is assumed to be a 25 × 25 matrix, the window refers to a window of data processed each time, for example, when the window size defaults to 3 × 3, the central data is the data processed by the filtering at this time, the current data is processed, the data is translated to the next 3 × 3 window, the processing is continued, and the next line is processed after one line is processed;
assume the original gray matrix as follows:
the window size is assumed to be 3 x 3, so the first step in image processing is:
the middle thickened 24 is the data to be processed, a new result is obtained by using a filtering algorithm to replace the original 24, the result obtained after filtering is assumed to be 23, the next step is rightward translation, and the steps are as follows:
the thickened part 23 is the result obtained in the first step, and the thickened part 34 is the data to be processed;
s3, extracting characteristic quantities of the denoised combustion flame images of the combustion flame images, and then corresponding the temperatures of the combustion flame images to the corresponding extracted characteristic quantities one by one; the specific operation method for extracting the characteristic quantity is as follows:
s31, cutting and dividing the combustion flame image along two or more transverse cutting lines and two or more longitudinal cutting lines, where a gray scale coefficient of a gray scale matrix corresponding to each intersection of the two or more transverse cutting lines and the two or more longitudinal cutting lines is a feature quantity of the combustion flame image, as shown in fig. 4;
s32, extracting characteristic quantities of all the combustion flame images according to the steps to obtain characteristic quantities corresponding to different temperatures, wherein the coefficient matrix corresponding to the characteristic quantity of each combustion flame image and the coefficient matrix formed by the corresponding temperatures of all the combustion flame images are as shown in the formula 4:
wherein x is11、x12…xndThe gray scale coefficients corresponding to the respective intersections are shown, the subscript nd represents the number of rows and columns of the cutting lines, the input represents a coefficient matrix corresponding to the characteristic amount of the combustion flame image, wendu1、 wendu2…wendunThe temperature corresponding to all the combustion flame images is represented, and the output represents a coefficient matrix formed by the corresponding temperatures of all the combustion flame images;
s4, performing data fitting by using the established corresponding relation between the characteristic quantity and the temperature, and then inputting the combustion flame image to be identified, namely directly extracting the temperature corresponding to the combustion flame image to be identified, wherein the specific operation method comprises the following steps:
s41, taking the characteristic quantity corresponding to the combustion flame image as the input of the KNN algorithm, taking the temperature corresponding to the combustion flame image as the output of the KNN algorithm, and fitting the data;
referring to fig. 5, the algorithm is implemented as follows:
s411, arbitrarily selecting a point x, and expressing the coordinate of the point x as (x)1,x2);
S412, find the euclidean distance between the remaining points and x, as shown in equation 5 below:
s413, finding out k points nearest to X according to the k parameters of the KNN algorithm, and recording the k points as y1、y2…ykThe coordinates of the k points are (y)11,y12)、(y21,y22)…(yk1,yk2);
And S414, averaging the gray coefficients corresponding to the coordinates of the k points to obtain the prediction result of the x.
S42, optimizing k parameters of the KNN algorithm by using a grid search algorithm and cross validation, setting parameters of cross validation, dividing data into a training set and a validation set, and finding a k value with the highest classification accuracy by using the data of the training set in a range of 1-11;
s43, verifying the accuracy of the algorithm by using the data of the residual verification set, namely the error between the predicted image temperature and the temperature obtained by the actual temperature sensing detector;
the grid search algorithm is an exhaustive search algorithm, which means that all possible values of the parameter to be optimized are listed, the prediction effect is tried one by one when the value of the parameter to be optimized is the values, and the value with the minimum error rate is selected as the value of the parameter to be predicted;
cross validation refers to dividing the original data set into two parts, namely a training set and a test set, wherein the ratio of the training set to the test set is 9: 1. as shown in fig. 7, from the column viewpoint, in the first column, the training set is divided into five equal parts, the training sets 2, 3, 4 and 5 are used to fit the data relationship, the unused training set 1 is used to verify the accuracy of the algorithm, then the result of the Test set is predicted to obtain Test1, and the steps are repeated for 5 times. Combining the five training sets for verification to obtain a new training set, averaging the obtained tests 1 to 5 to obtain a new Test, namely a Test set, and then predicting the result by using the newly obtained training set and the Test set. The cross validation can fully utilize the data and furthest mine the internal relation of the data;
and S44, inputting the combustion flame image to be recognized, and calling a fitted algorithm to directly obtain the temperature corresponding to the combustion flame image to be recognized after filtering.
The invention provides a control method of a low-concentration gas combustion system, which comprises the following steps:
w1, acquiring a plurality of continuous combustion flame images (from a furnace starting temperature rising stage to stable combustion of the furnace) in the furnace, and acquiring the temperatures corresponding to the plurality of combustion flame images by using the temperature identification method based on the flame images;
w2, judging the combustion state of the hearth according to the temperature information corresponding to the multiple combustion flame images, obtaining a reasonable range of air inflow when the temperature is a certain amount according to the corresponding relation between the temperature and the air inflow when the hearth normally combusts, increasing the air inflow in the reasonable range when the temperature of the hearth is in an increasing state, reducing the air inflow in the reasonable range when the temperature of the hearth is in a decreasing state, and maintaining the air inflow unchanged when the temperature of the hearth is not changed.
And (3) observing the trend of temperature change of the continuous images through temperature identification of the continuous images, determining whether the continuous images are in stable combustion or unstable combustion, and adjusting the air intake quantity according to the combustion state and the level of the air intake quantity during normal combustion. The levels of air inflow (oxygen, carbon monoxide and methane) corresponding to different combustion states and temperatures are different, and the proper air inflow is adjusted according to the combustion states and the temperatures, so that the phenomena of overhigh or overlow temperature and the like are prevented, and the stable operation of equipment is prevented from being influenced and even serious consequences are avoided. The stable combustion refers to a state in which the temperature is maintained at a certain level of rise, fall, or balance, and these three states correspond to a range of intake air amount within which the intake air amount is changed so that the combustion is in a stable state. The unstable combustion means that the intake air amount is not within the normal range, a sudden change occurs, or the intake air amount is not properly input to cause a change in the combustion state, which may cause damage to the equipment, or the like. When the temperature of the hearth is in a rising state according to the temperature identified by the image, properly increasing the air inflow and enabling the air inflow to be in a proper air inflow range, and enabling the temperature to be in a stable rising state; if the temperature of the hearth is in a descending state, properly reducing the air inflow and enabling the air inflow to be in a proper air inflow range, and enabling the temperature of the hearth to be in a stable descending state; when the displayed temperature is not changed, the intake air amount needs to be maintained in the intake air amount range corresponding to the temperature.
In a specific practice, according to the temperature identification method of the present invention, a furnace combustion video is first converted into combustion flame images with a granularity of 10 seconds, for a total of 248 sheets. The experimental data are 2488 temperature data from 9 hours 23 to 10 hours 4, and 248 temperature data corresponding to the pictures are taken as shown in fig. 2. And then each image is converted into an R, G, B three-channel data form, a R, G, B matrix is converted into a gray matrix, one image corresponds to a group of gray data matrixes, the image is denoised by using a self-adaptive weighted mean-median filtering method in order to reduce the influence of noise, and the denoising result is shown in fig. 3.
Extracting features from the new gray matrix data according to the feature extraction method of fig. 4 to form a set of feature quantities, wherein the size of the picture is 720 × 480, taking the gray coefficients of 15 points in the picture to construct the feature quantities, and converting the original image into 15 feature quantities corresponding to one temperature.
And fitting by using the KNN algorithm and taking the characteristic quantity as an input and the temperature as an output. And optimizing the KNN parameter by using a grid optimization method and combining cross validation, wherein the fold number of the cross validation is 5, random nine-tenth data is used for algorithm fitting, the remaining one-tenth data is used for validating the error rate of the algorithm, MAPE is used as an evaluation index, the calculation formula of the MAPE is shown as a formula 6, and the result is shown in FIG. 6.
In the formula: pre is the temperature predicted from the image, y refers to the temperature detected by the sensor, and the smaller the MAPE, the lower the prediction error of the algorithm.
Next, 6 consecutive images are input, the temperature of each image is recognized by the above-described temperature recognition method, and the combustion state of the furnace is determined based on the temperature change of the image. The successive image temperatures are shown in fig. 8.
Fig. 8 is the temperature of the input 6 consecutive pictures, which can be seen from the figure as the temperature becomes a descending state, but the variation range is about 1.2 ℃ from the temperature data, so that the combustion state corresponding to the furnace at this moment is judged as stable combustion.
According to the collected data of the combustion of the hearth, the air intake amount range of the hearth with the normal combustion temperature of 1032.4-1033.8 ℃ is known as follows: the input concentration of carbon monoxide is 1% to 4%, and the input concentration of oxygen is 11.8% to 12.0%. Therefore, in order to ensure that the furnace continues to stabilize within this temperature range, the input concentration of carbon monoxide into the concentration meter should not exceed the above-mentioned range.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the above embodiment method can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., a rom/ram, a magnetic disk, an optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the present specification and drawings, or used directly or indirectly in other related fields, are included in the scope of the present invention.
Claims (9)
1. A temperature identification method based on flame images is characterized in that: the method comprises the following steps:
s1, acquiring a combustion video in the hearth, detecting the temperature in the hearth in real time through a temperature sensing detector, capturing the acquired combustion video in the hearth by taking the T time as the granularity to obtain a plurality of combustion flame images in the hearth, and corresponding the combustion flame images to the temperature at the corresponding moment detected by the temperature sensing detector one by one;
s2, denoising each combustion flame image by using a self-adaptive weighted mean-median filtering method;
s3, extracting characteristic quantities of the denoised combustion flame images of the combustion flame images, and then enabling the temperatures corresponding to the combustion flame images of the combustion flame images to correspond to the extracted characteristic quantities one by one;
and S4, performing data fitting by using the established corresponding relation between the characteristic quantity and the temperature, and then inputting the combustion flame image to be identified, namely directly extracting the temperature corresponding to the combustion flame image to be identified.
2. The flame image-based temperature identification method according to claim 1, wherein: in step S1, the specific operation method of the screenshot is as follows:
the image acquisition is carried out by using a VideoCapture module in a cv2 package, and combustion flame images at different time intervals are obtained by changing delay time.
3. The flame image-based temperature identification method according to claim 1 or 2, wherein: the specific operation method of step S2 is as follows:
s21, converting the combustion flame image into an RGB three-channel image to obtain a R, G, B matrix of the RGB three-channel image, and then converting the R, G, B matrix into a gray matrix;
s22, denoising pulse noise and Gaussian noise in the combustion flame image noise by using a self-adaptive weighted mean-median filtering method based on the gray matrix corresponding to the combustion flame image, wherein the denoising calculation formula is shown as formula 1, formula 2 and formula 3:
the data to be denoised is represented by zxy, namely the data at the center of the window, mean represents the mean value of the window data, med represents the median value of the window data, and zxy' represents the corrected data.
4. The flame image-based temperature identification method according to claim 1 or 2, wherein: in step S3, the specific operation method of extracting the feature amount is as follows:
s31, cutting and dividing the combustion flame image along more than two transverse cutting lines and more than two longitudinal cutting lines, wherein the gray coefficient of the gray matrix corresponding to each intersection point of the more than two transverse cutting lines and the more than two longitudinal cutting lines is the characteristic quantity of the combustion flame image;
s32, extracting characteristic quantities of all the combustion flame images according to the steps to obtain characteristic quantities corresponding to different temperatures, wherein the coefficient matrix corresponding to the characteristic quantity of each combustion flame image and the coefficient matrix formed by the corresponding temperatures of all the combustion flame images are as shown in the formula 4:
wherein x is11、x12…xndThe gray scale coefficient corresponding to each intersection point is shown, the subscript nd represents the number of rows and columns of the cutting lines, and the input represents the characteristic amount of the combustion flame imageCorresponding coefficient matrix, wendu1、wendu2…wendunThe temperatures corresponding to all the combustion flame images are indicated, and the output indicates a coefficient matrix formed by the corresponding temperatures of all the combustion flame images.
5. The flame image-based temperature identification method according to claim 1 or 2, wherein: the specific operation method of step S4 is as follows:
s41, taking the characteristic quantity corresponding to the combustion flame image as the input of the KNN algorithm, taking the temperature corresponding to the combustion flame image as the output of the KNN algorithm, and fitting the data;
and S42, inputting the combustion flame image to be recognized, and calling the fitted algorithm to directly obtain the temperature corresponding to the combustion flame image to be recognized.
6. The flame image-based temperature identification method according to claim 5, wherein: in step S4, the fitting of the data includes the following steps:
optimizing the k parameter of the KNN algorithm by using a grid search algorithm and cross validation, setting the parameters of the cross validation, dividing the data into a training set and a validation set, and finding out a k value which enables the highest classification accuracy by using the data of the training set in the range of 1-11; the accuracy of the algorithm is verified using the data of the remaining verification set.
7. The flame image-based temperature identification method according to claim 5, wherein: the algorithm in step S41 is implemented as follows:
s411, arbitrarily selecting a point x, and expressing the coordinate of the point x as (x)1,x2);
S412, find the euclidean distance between the remaining points and x, as shown in equation 5 below:
s413, k according to KNN algorithmThe parameters find out the k points nearest to X, and the k points are marked as y1、y2…ykThe coordinates of the k points are (y)11,y12)、(y21,y22)…(yk1,yk2);
And S414, averaging the gray coefficients corresponding to the coordinates of the k points to obtain the prediction result of the x.
8. A control method of a low-concentration gas combustion system is characterized in that: the method comprises the following steps:
w1, acquiring a plurality of continuous combustion flame images in a hearth, and obtaining the corresponding temperatures of the plurality of combustion flame images by using the flame image-based temperature identification method according to any one of claims 1 to 7;
w2, judging the combustion state of the hearth according to the temperature change trend corresponding to the obtained multiple combustion flame images, and finally adjusting the air inflow according to the combustion state of the hearth.
9. The control method of a low-concentration gas combustion system as set forth in claim 8, wherein: and judging the combustion state of the hearth according to the temperature information corresponding to the multiple combustion flame images, obtaining a reasonable range of air inflow when the temperature is a certain amount according to the corresponding relation between the temperature and the air inflow when the hearth normally combusts, increasing the air inflow in the reasonable range when the temperature of the hearth is in an increasing state, reducing the air inflow in the reasonable range when the temperature of the hearth is in a decreasing state, and maintaining the air inflow unchanged when the temperature of the hearth is unchanged.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010984130.9A CN112200735A (en) | 2020-09-18 | 2020-09-18 | Temperature identification method based on flame image and control method of low-concentration gas combustion system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010984130.9A CN112200735A (en) | 2020-09-18 | 2020-09-18 | Temperature identification method based on flame image and control method of low-concentration gas combustion system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112200735A true CN112200735A (en) | 2021-01-08 |
Family
ID=74015459
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010984130.9A Pending CN112200735A (en) | 2020-09-18 | 2020-09-18 | Temperature identification method based on flame image and control method of low-concentration gas combustion system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112200735A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113177364A (en) * | 2021-05-21 | 2021-07-27 | 东北大学 | Soft measurement modeling method for temperature of blast furnace tuyere raceway |
CN118168348A (en) * | 2024-05-14 | 2024-06-11 | 常州陆铭光电科技有限公司 | Hearth double-spectrum temperature monitoring system and monitoring method |
Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090082937A1 (en) * | 2005-03-29 | 2009-03-26 | Toyota Jidosha Kabushiki Kaisha | Fuel injection control device for engine |
CN101976433A (en) * | 2010-09-09 | 2011-02-16 | 西安电子科技大学 | Method for inhibiting speckle noise of polarized SAR (Search and Rescue) data |
CN102338471A (en) * | 2011-09-02 | 2012-02-01 | 中国石油天然气股份有限公司 | Optimized operation monitoring method for heating furnace |
CN105303536A (en) * | 2015-11-26 | 2016-02-03 | 南京工程学院 | Median filtering algorithm based on weighted mean filtering |
CN105865217A (en) * | 2016-04-21 | 2016-08-17 | 中冶南方工程技术有限公司 | Control method of gas flow of burner nozzle for baking furnace |
CN106355812A (en) * | 2016-08-10 | 2017-01-25 | 安徽理工大学 | Fire hazard prediction method based on temperature fields |
CN107085833A (en) * | 2017-04-13 | 2017-08-22 | 长安大学 | Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal |
CN107274365A (en) * | 2017-06-15 | 2017-10-20 | 中国矿业大学(北京) | A kind of mine image intensification method based on unsharp masking and NSCT algorithms |
CN107525092A (en) * | 2017-10-18 | 2017-12-29 | 西安昱昌环境科技有限公司 | Rotary RTO temperature control system and method |
CN108194942A (en) * | 2017-12-15 | 2018-06-22 | 中国大唐集团科学技术研究院有限公司火力发电技术研究所 | A kind of method for controlling flame temperature |
CN108278657A (en) * | 2018-01-27 | 2018-07-13 | 卫勇锋 | Constant pressure constant-temperature control method for wall-hung boiler heating system |
CN108921864A (en) * | 2018-06-22 | 2018-11-30 | 广东工业大学 | A kind of Light stripes center extraction method and device |
CN109284756A (en) * | 2018-08-01 | 2019-01-29 | 河海大学 | A kind of terminal censorship method based on OCR technique |
CN109514043A (en) * | 2018-12-13 | 2019-03-26 | 刘堂斌 | A kind of effective welding robot welding control system |
CN110094882A (en) * | 2018-01-31 | 2019-08-06 | 芜湖美的厨卫电器制造有限公司 | Gas heater and control method, computer readable storage medium, heating system |
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 |
CN111223069A (en) * | 2020-01-14 | 2020-06-02 | 天津工业大学 | Image fusion method and system |
CN111664976A (en) * | 2020-06-19 | 2020-09-15 | 北京化工大学 | Visual halo-based array pressure measurement method and device and preparation method |
-
2020
- 2020-09-18 CN CN202010984130.9A patent/CN112200735A/en active Pending
Patent Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090082937A1 (en) * | 2005-03-29 | 2009-03-26 | Toyota Jidosha Kabushiki Kaisha | Fuel injection control device for engine |
CN101976433A (en) * | 2010-09-09 | 2011-02-16 | 西安电子科技大学 | Method for inhibiting speckle noise of polarized SAR (Search and Rescue) data |
CN102338471A (en) * | 2011-09-02 | 2012-02-01 | 中国石油天然气股份有限公司 | Optimized operation monitoring method for heating furnace |
CN105303536A (en) * | 2015-11-26 | 2016-02-03 | 南京工程学院 | Median filtering algorithm based on weighted mean filtering |
CN105865217A (en) * | 2016-04-21 | 2016-08-17 | 中冶南方工程技术有限公司 | Control method of gas flow of burner nozzle for baking furnace |
CN106355812A (en) * | 2016-08-10 | 2017-01-25 | 安徽理工大学 | Fire hazard prediction method based on temperature fields |
CN107085833A (en) * | 2017-04-13 | 2017-08-22 | 长安大学 | Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal |
CN107274365A (en) * | 2017-06-15 | 2017-10-20 | 中国矿业大学(北京) | A kind of mine image intensification method based on unsharp masking and NSCT algorithms |
CN107525092A (en) * | 2017-10-18 | 2017-12-29 | 西安昱昌环境科技有限公司 | Rotary RTO temperature control system and method |
CN108194942A (en) * | 2017-12-15 | 2018-06-22 | 中国大唐集团科学技术研究院有限公司火力发电技术研究所 | A kind of method for controlling flame temperature |
CN108278657A (en) * | 2018-01-27 | 2018-07-13 | 卫勇锋 | Constant pressure constant-temperature control method for wall-hung boiler heating system |
CN110094882A (en) * | 2018-01-31 | 2019-08-06 | 芜湖美的厨卫电器制造有限公司 | Gas heater and control method, computer readable storage medium, heating system |
CN108921864A (en) * | 2018-06-22 | 2018-11-30 | 广东工业大学 | A kind of Light stripes center extraction method and device |
CN109284756A (en) * | 2018-08-01 | 2019-01-29 | 河海大学 | A kind of terminal censorship method based on OCR technique |
CN109514043A (en) * | 2018-12-13 | 2019-03-26 | 刘堂斌 | A kind of effective welding robot welding control system |
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 |
CN111223069A (en) * | 2020-01-14 | 2020-06-02 | 天津工业大学 | Image fusion method and system |
CN111664976A (en) * | 2020-06-19 | 2020-09-15 | 北京化工大学 | Visual halo-based array pressure measurement method and device and preparation method |
Non-Patent Citations (9)
Title |
---|
H. CHEN, X ET AL: "Recognition of the Temperature Condition of a Rotary Kiln Using Dynamic Features of a Series of Blurry Flame Images", 《TRANSACTIONS ON INDUSTRIAL INFORMATICS》 * |
H. CHEN, X ET AL: "Recognition of the Temperature Condition of a Rotary Kiln Using Dynamic Features of a Series of Blurry Flame Images", 《TRANSACTIONS ON INDUSTRIAL INFORMATICS》, vol. 12, no. 1, 29 February 2016 (2016-02-29), pages 148 - 157, XP011597923, DOI: 10.1109/TII.2015.2500891 * |
唐永鹤等: "基于自适应滤波的单像素宽形态学边缘检测", 《信号处理》 * |
唐永鹤等: "基于自适应滤波的单像素宽形态学边缘检测", 《信号处理》, vol. 27, no. 8, 31 August 2011 (2011-08-31), pages 2 * |
朱永红等: "基于改进BP神经网络陶瓷梭式窑火焰图像识别方法", 《中国陶瓷工业》 * |
朱永红等: "基于改进BP神经网络陶瓷梭式窑火焰图像识别方法", 《中国陶瓷工业》, vol. 27, no. 1, 29 February 2020 (2020-02-29), pages 38 - 1 * |
王一丁: "《工业图像处理》", pages: 34 * |
赵君爱: "基于图像处理的工件表面缺陷检测理论与方法研究", 《中国博士学位论文全文数据库工程科技Ⅱ辑》 * |
赵君爱: "基于图像处理的工件表面缺陷检测理论与方法研究", 《中国博士学位论文全文数据库工程科技Ⅱ辑》, vol. 2017, no. 11, 15 November 2017 (2017-11-15), pages 029 - 23 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113177364A (en) * | 2021-05-21 | 2021-07-27 | 东北大学 | Soft measurement modeling method for temperature of blast furnace tuyere raceway |
WO2022242109A1 (en) * | 2021-05-21 | 2022-11-24 | 东北大学 | Soft measurement modeling method for temperature of tuyere raceway of blast furnace |
CN113177364B (en) * | 2021-05-21 | 2023-07-14 | 东北大学 | Soft measurement modeling method for temperature of blast furnace tuyere convolution zone |
CN118168348A (en) * | 2024-05-14 | 2024-06-11 | 常州陆铭光电科技有限公司 | Hearth double-spectrum temperature monitoring system and monitoring method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111583187B (en) | PCB defect detection method based on CNN visualization | |
CN112200735A (en) | Temperature identification method based on flame image and control method of low-concentration gas combustion system | |
CN112819093B (en) | Man-machine asynchronous identification method based on small data set and convolutional neural network | |
JP6216024B1 (en) | Trained model generation method and signal data discrimination device | |
CN111582071B (en) | SF6 instrument image reading method based on HRNet network model | |
CN109614908B (en) | Flame combustion state detection system and detection method | |
CN112560919A (en) | Man-machine asynchronous recognition method based on one-dimensional interpretable convolutional neural network | |
CN114782694A (en) | Unsupervised anomaly detection method, system, device and storage medium | |
CN109376686A (en) | A kind of various dimensions human face data acquisition scheme, acquisition system and acquisition method | |
CN115798022A (en) | Artificial intelligence identification method based on feature extraction | |
CN111124896B (en) | Metamorphic test system for primary and secondary peak ratio calculation algorithm | |
CN113920673A (en) | Intelligent indoor fire monitoring method and system | |
CN105913429A (en) | Calculating method for visual perception response time delay index of intelligent terminal user | |
CN114926682A (en) | Local outlier factor-based industrial image anomaly detection and positioning method and system | |
CN111862036B (en) | Tool abnormal placement detection method and system based on double-index measurement learning | |
CN116127604B (en) | Method and system for processing anti-collision data of automobile | |
CN116561539A (en) | Model robustness optimization method based on interpretability algorithm | |
CN110334807A (en) | Training method, device, equipment and the storage medium of deep learning network | |
CN111815604B (en) | Blast furnace tuyere monitoring method and device, electronic equipment and storage medium | |
CN113449631A (en) | Image classification method and system | |
JP6786364B2 (en) | Learning device and learning method | |
CN109934305A (en) | Image-recognizing method and device based on image recognition model | |
CN109101793A (en) | A kind of personal identification method and system based on static text keystroke characteristic | |
CN113124414B (en) | Data processing system and method | |
CN117558307B (en) | Threshold voltage adjusting method, device and equipment of memory chip 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 |