JPS60244821A - Processing method of fire image - Google Patents

Abstract

PURPOSE:To automate control over dust coal by picking up an image of flames of flames of the dust coal through an image pickup camera and calculating the shape and brightness of a volatile component combustion area from the difference in brightness distribution between an oxidation area and a reduction area. CONSTITUTION:The brightness distribution 1A of the oxidation area 1 at constant distance from the center line B-B' of flames is measured along a line A-A'. There is the volatile component combustion area 2 present near a burner opening part, so the brightness distribution 1A is the sum of the brightness of the oxidation area and the brightness of the volatile component combustion area. Then, a brightness distribution 1B on a center line B-B' is measured to obtain the brightness distribution of the reduction area 3. Then, the brightness distribution of only an oxidation area near the burner opening part is estimated from the brightness distribution 1B. Then, the difference between the brightness distribution 1A and estimated brightness distribution is calculated to obtain the brightness distribution of only the volatile component combustion area 2.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a flame image processing method that captures a flame image of a pulverized coal combustion process with an imaging camera and extracts feature quantities in the flame structure necessary for diagnosing the combustion state. .

[Background of the invention]

Until now, there was no technology for diagnosing the combustion state by capturing images of the flame inside the furnace with an imaging camera. Instead, operators had to observe the combustion state with their naked eyes through a peephole, and used their experience to understand and monitor the combustion state. However, with this method, the combustion state cannot be constantly monitored, and the diagnosis of the combustion state is left to the operator's experience, so there is a problem that automation of control for high efficiency combustion or low NOx combustion cannot be realized.

[Purpose of the invention]

An object of the present invention is to provide a flame image processing method for extracting the shape, brightness, and intensity of a volatile combustion region as flame feature values necessary for combustion diagnosis from a flame image of pulverized coal combustion captured by an imaging camera. There is a particular thing.

[Summary of the invention]

In image processing of a flame in a furnace, the present invention extracts a first brightness distribution curve with respect to distance from the burner opening on a line parallel to the center line by passing through the flame near the burner opening,
By utilizing the fact that the brightness distribution in a portion separated by at least twice the diameter of the burner opening is expressed by the sum of the brightness in the oxidized region and the brightness in the reduced region, the second Estimating the brightness distribution curve, the first and second
It is characterized by estimating that the difference in brightness in the brightness distribution curve is the brightness due to volatile matter combustion.

[Embodiments of the invention]

First, the basic matters of the present invention will be described.

The present invention focuses on the flame structure of pulverized coal combustion, particularly in a pulverized coal combustion method aiming at low NOx and high efficiency combustion. That is, in pulverized coal combustion, carbonized gas (volatile matter) and coke (solid carbon) are separated and ignited and burned, and the flame structure is shown in FIG. 1(a). The pulverized coal combustion process will be explained based on FIGS. 1(a) to (C).

Figure 1(a) shows an example of the structure of a flame, where 1 indicates the oxidized region,
2 indicates the volatile matter combustion region, and 3 indicates the reduction region.

Next, the combustion process will be briefly explained. Figure 1 (a
), the pulverized coal that is mixed with air and ejected into the furnace from the burner is rapidly heated by the radiant heat from the high-temperature furnace wall and flame, causing some of the pulverized coal grains to crack and collapse. Then, the particles become even finer. At the same time, the volatile matter is carbonized and carbonized gas with a volume approximately 500 times that of the pulverized coal grains is rapidly released.

■ The carbonized gas reacts with the air that is present between the particles surrounding the coke-like wood that has released the carbonized gas, resulting in diffuse combustion (forming a volatile combustion region).

■ The dehulled coke grains become pumice-like and become more buoyant, causing solid carbon to burn on the surface (forming a reduction/oxidation region).

The flame structure consists of an oxidation region 1 in which volatile matter and solid matter are burned in a state of excess oxygen in the outer flame, and a volatile matter combustion region 2 formed near the burner where volatile matter is burned at a high combustion rate. The inner flame is configured with a reduction region 3 consisting of combustion of solids in an oxygen-deficient state.

Figure 1(b) shows the brightness distribution with respect to the distance from the burner in the reduction region and oxidation region in the flame, and the brightness distribution (B-B') in the three reduction regions shows the combustion brightness of solids. On the other hand, the brightness distribution (A-A') in one oxidation region shows a brightness distribution in which the solid content combustion brightness 4 and the volatile content combustion brightness 5 are added. The combustion area of volatile matter is the distance from the burner
It is also demonstrated from FIG. 1(C) that it is formed in the range /D=0 to 1. However, D indicates the burner opening diameter.

In Fig. 1(b), IA is the oxidized region (A-A') in Fig. 1(a).
), IB is the brightness distribution in the reduced region in Fig. 1(a), 4 is the brightness distribution occupied by the solid content, that is, the solid content in the oxidized region, and 5 is the brightness distribution due to the volatile matter. ing. The horizontal axis is the distance from the burner tip, which is normalized by the burner opening diameter and shown as X/D.

FIG. 1(C) shows the generated gas distribution in the oxidation region (A-A').

Carbon dioxide (Cow) rapidly increases and oxygen (02) rapidly decreases in the distance X/D=O~1 from the burner. This indicates that volatile matter is being burned in this section because the burning rate of volatile matter is faster than that of solid matter. X/D<1.0 indicates a volatile combustion region, and X/D) 1.0 indicates an oxidation region.

From the above, the brightness distribution of the flame is observed, and the observed brightness distribution is separated into the combustion brightness of solid content and the combustion brightness of volatile matter.
In other words, the shape and brightness of the volatile matter combustion region in the flame can be extracted using brightness conversion that converts the solid content combustion brightness into relative brightness.

The brightness distribution of solid content, which is the basis for brightness conversion, is that the brightness distribution observed in the reduction region is directly equal to the brightness distribution of solid content, but in the oxidation region and volatile matter combustion region, the brightness distribution observed does not include volatile matter The combustion brightness of the solid content is estimated using the method shown below.

The first method approximates the brightness distribution of the solid content by the brightness distribution on the flame center line. The flame area to be treated is the distance from the burner including the volatile combustion area: X/D=O~2
In this area, the combustion rate of solids is slower than that of volatiles, so the combustion of volatiles becomes dominant, and the brightness of solids combustion in the oxidation region and volatile combustion region rises above the flame center line. There is not much difference even if it is approximated by the brightness distribution of .

In the second method, in order to improve the accuracy of extraction of the volatile matter combustion region, a correction calculation is performed on the brightness distribution on the flame center line, and the result is used as the brightness distribution of the solid content. When the brightness distribution on the flame center line is Ro(x), the brightness distribution R, (x, y) of the solid content at a distance y from the center line is approximated by the following equation.

Skin (x, y)=K(y)XRo←) ・・・・・・(
1) Here, K(y): Correction coefficient. However, at the flame center, K(o)=1.0. R1(x, o
)=&1(X) There are many possible ways to set the correction coefficient K (y), examples of which are shown below.

゛An embodiment of the present invention is shown in Fig. 2. 6 is a boiler furnace for pulverized coal combustion, 7 is a burner for pulverized coal combustion, 8 is an image
9 is an imaging camera, 10 is a flame image input device,
11 is an A/D converter, 12 is a memory, 13 is a flame image processing device, 14 is a combustion state diagnosis device, and 15 is a CRT display device. The flame produced when pulverized coal is burned by a pulverized coal combustion burner 7 in a pulverized coal combustion boiler furnace 6 is captured by an image fiber 8 installed in the furnace with fire resistance, and an electrical signal is transmitted by an imaging camera 9. is converted to

The flame image data converted into electrical signals is taken into the image input device 10, subjected to A/D conversion by the A/D converter 11, and then stored in the memory 12. The flame image data stored in the memory is taken into the flame image processing device 13, which executes the processing described below, and the processing results are sent to the combustion diagnosis device 14, which diagnoses the combustion state and displays the flame state on the CRT display 15. do.

Next, the processing contents of the flame image processing apparatus according to the present invention will be described.

When the flame image during pulverized coal combustion is captured with an image fiber and converted into an electrical signal using an imaging camera and visualized, the result shown in Figure 3 (a) is
) is obtained (stored in the flame image input device memory)
. Figure (a) shows the image data of the entire screen captured by the image fiber, and from this image data, the image data near the target burner (black area in Figure (a)) is extracted and flame image processing is performed. Input into the device (cutting process),
The purpose of the extraction processing is to limit the image data to a range where the above equation (1) approximately holds true, to remove the influence of other burner parts, and to reduce the calculation time and memory capacity of the processing in the image processing device. It is to reduce the data size to a value determined by the relationship. The extracted image data is shown in the same figure (b).
Shown below. Next, the extracted image data is rotated in order to facilitate subsequent processing and to facilitate image recognition when the image data is displayed on a CRT display or the like. There are many possible ways to rotate, one example of which is shown below. In other words,
The following linear transformation (affine transformation) is applied to the image data.

The result of rotation processing performed on the image data of FIG. 3(b) is shown in FIG. 3(C). Further, the brightness distribution of the image data after rotation is shown in FIG. 2(d). The brightness increases as you move to the right and away from the burner. (d') is C-C', D-D' of (d)
It shows the brightness distribution of the cross section. Next, the flame part and the burner part/furnace wall part are separated from the image data, and the flame region is extracted. The purpose of this is that in later processing, if the brightness of the reducing flame is lower than the brightness of the furnace wall, the furnace wall will be mistakenly recognized as a flame.
This is to prevent noise from occurring in the image data.

A simple method for flame region separation is threshold processing in which image data below a certain brightness level is replaced with brightness O. Let the brightness of each pixel of image data IP(i,j) be expressed by a threshold value t. As a method for setting the threshold value, if the combustion amount in the burner is constant or the combustion pattern is fixed, it may be fixed to a constant brightness level from the image data brightness distribution or to a brightness level pattern that matches the combustion pattern. . When the combustion amount fluctuates widely or irregularly, it is difficult to fix the threshold value. In this case, the threshold value is determined by the following method. First, set the area occupied by the flame area, then calculate the frequency (corresponding to the area) of each brightness level from the image data, add the frequency of each brightness level from the maximum brightness in the image data, and the addition result is the first. The brightness level at the time when the set area (area of the flame region) is exceeded is taken as the threshold value. According to this method, since the threshold value is updated to an appropriate value every time, appropriate processing is performed depending on the burner combustion amount.

The image data resulting from the threshold processing is shown in FIG. 3(e), and its luminance distribution is shown in FIG. 3(f). (f') is C of (f)
-C', shows the brightness distribution of the DD' cross section.

Next, the shape and brightness distribution of the volatile combustion region in the oxidation flame are extracted from the threshold-processed flame image data. As for the oxidation flame extraction method, if threshold processing is performed using a certain threshold as described above, as is clear from the brightness distribution in Fig. 3 (D), both the oxidation flame region and the reduction flame region are Since the brightness increases as the coal moves away from the coal, it cannot be easily separated. Therefore, as shown in Figure 1, the combustion in the oxidation region is the combustion of solid content and volatile content in the pulverized coal. Focusing on the fact that the combustion in the reduction region at the center of the flame is the combustion of solid matter, it is sufficient to perform conversion processing into a flame image using the brightness level of the reduction flame as a reference.In other words, the brightness level of the reduction flame in the flame region By applying brightness conversion to the relative brightness level of , the volatile combustion region in the oxidizing flame can be extracted.

The specific processing method is based on the standard brightness, that is, the brightness of the reducing flame, and the brightness distribution at the center of the reducing flame (c-c in Figure 3(f)).
``Brightness distribution on the twilight line'' is used as a reference, and the brightness is converted into relative brightness from the reference brightness at the X coordinate in the Y-axis direction.

R''''(i, j) JL(i, j) Ro(i, j
) ・・・・・・(4) Here, R(IT J): Luminance before conversion R(i, j): Luminance after conversion R-(r*j)'Reference luminance++J', respectively, XIY coordinates Further, the reference luminance R-(++j) is given by the following equation.

Ro(i, j)=K(j)X also(i)...
...(5) Here, K(j): Correction coefficient Ro(i): Brightness correction coefficient on the flame center line K(j) corrects the brightness distribution of solid content in the Y-axis direction from the flame center line The brightness is calculated using the following formula based on the brightness of the area (the rightmost end farthest from the burner on the image data) that consists only of solid matter after the combustion of volatile matter has finished.

Here, 5□〇: Maximum value of X coordinate If the image data length is set to around the distance X/D from the burner = 2, the right end of the image data will be as shown in Figure 1 (b).
As shown in (6), there is no effect of volatile content, and the brightness distribution of solid content can be corrected by using the correction coefficient K(j) shown in equation (6), which is composed only of the combustion brightness of solid content. FIG. 3(g) shows the extraction result of the volatile matter combustion region extracted by brightness conversion, which can be well approximated, and FIG. 3(h) shows its brightness distribution. (h') is the brightness distribution on the DD' cross section in (h).

In this way, we focused on the fact that pulverized coal combustion consists of the combustion of solids and volatiles in pulverized coal, and that oxidation flames burn solids and volatiles, and reduction flames burn solids. By converting the flame image data to the brightness level in the region, the shape and brightness of the volatile combustion region in the oxidizing flame can be extracted.

〔Effect of the invention〕

By implementing the present invention, it is possible to extract the shape and brightness of the volatile combustion region in a pulverized coal combustion flame, which is essential for combustion state monitoring 1 diagnosis during pulverized coal combustion, low NOx combustion control, high efficiency combustion control, etc. I can do it.

[Brief explanation of the drawing]

FIG. 1 shows a structural diagram of a pulverized coal combustion flame to which the present invention is applied. FIG. 2 shows an embodiment of the invention. Third
The figure shows flame image data and its brightness distribution in the processing process of the flame image processing device according to the present invention. 6... Boiler furnace for pulverized coal combustion, 7... Burner for pulverized coal combustion, 8... Image fiber, 9... Imaging camera, 10... Flame image input device, 11... A/ D
Converter, 12... Memory, 13... Flame image processing device, 14... Combustion state diagnosis device, 15... CRT display device. Y I Figure (Kyu) Figure 2 I

Claims (1)

[Claims] 1. An image processing method for flame in a furnace that performs combustion using pulverized coal as fuel, which comprises: capturing an image of the flame in the vicinity of a burner opening in the furnace with an imaging camera; and determining the direction of fuel injection from the burner. A first brightness distribution curve with respect to the distance from the burner opening on a line parallel to the center line that is perpendicular to the center line is extracted from the imaged flame data, and the diameter of the burner opening is determined. The second brightness due to oxidation and reductive combustion at the base of the flame is calculated using estimating a distribution curve, estimating that the difference between the first brightness distribution curve and the second brightness distribution curve at the flame root is the brightness due to volatile matter combustion, and determining the brightness due to volatile matter combustion from the imaging data. A flame image processing method characterized by extracting a combustion region.