JPH0240414A - Method and device for diagnosing combustion condition - Google Patents

Abstract

PURPOSE:To grasp a combustion state with accuracy by allowing a model factor which estimates a combustion state based on a characteristic value of a flame image to store an influence affected by the change in said combustion state as knowledge, and revising said model factor one by one based on said change. CONSTITUTION:Flames at the root of a burner 1 are measured as an image with an image fiber 3 so that they may be taken into an ITV camera 4. The flame image thus taken in is stored in a flame memory 6 by way of an A/D converter 5. A processor 7 extracts a characteristic value of flames. The ingredient of the exhaust gas is estimated by an estimation model which uses the characteristic value from the relation-ship between the aforesaid characteristic value and the exhaust gas. Since the exhaust gas ingredient changes markedly with combustion conditions, an AI processor 8 is adapted to revise and change the estimation model so that a highly accurate estimation model may be obtained constantly.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method and apparatus for diagnosing combustion conditions, and particularly to a knowledge-separated combustion condition diagnosis method and apparatus suitable for large-scale plants.

[Conventional technology]

Regulation values have been set for substances contained in the combustion gas during boiler operation, especially the harmful substances nitrogen oxides (hereinafter referred to as NOx), and the amount produced must be kept below the regulation values. drive.

On the other hand, it is desirable that the combustion efficiency of the boiler is always maintained at its maximum during operation. The unburned portion of the exhaust gas serves as a guideline for calculating combustion efficiency. As the amount of unburned matter in the exhaust gas increases, the combustion efficiency decreases, and the amount of combustion slaked increases to obtain the same output. However, conventionally, it has taken a long time to detect unburned components, so understanding combustion efficiency during operation has had to rely on experience and intuition.

Recently, the use of coal has been increasing in place of gas and oil as fuel, and the use of pulverized coal and CWM (coal/
water slurry), COM (coal/oil slurry)5, etc.

In particular, when coal is used as fuel, all of the nitrogen contained in coal is converted into NOx through combustion, so the amount produced is 2oOO~3000pP! It will become a huge thing. Furthermore, since the combustion speed is extremely slow compared to gas and oil, the furnace temperature decreases and the unburned content in the ash tends to increase, which has been a problem.

On the other hand, as a method of knowing the combustion state during boiler operation, (1
) Flame detector attached to the burner nozzle to detect the flame, (2) Detector attached to the exhaust gas outlet or flue to detect components contained in the exhaust gas, (3) Information inside the furnace. There was an ITV camera installed with an explosion-proof mechanism on the top of the furnace to obtain information. Such a detection end shown in Fig. 2 is for detecting the ignition or extinguishment of a banana for (1), and for (2)
) is installed to detect whether or not the limit value set by environmental regulations is exceeded. (3
) ITV camera is used to monitor the combustion status of the entire furnace.

[Problem to be solved by the invention]

However, such conventionally installed devices have the following drawbacks.

(1) A flame detector is a device that detects ``ignition'' or ``extinguishment'' of a flame at the exit of a burner.It detects the flame of another burner even though the flame is extinguished and mistakenly detects it as ``ignition''. This decision is left to the operator. This is because the flame detector can basically only output ON (ignition) and OFF (extinguishing) signals at the burner outlet.

(2) Exhaust gas component detectors require tens of seconds to several minutes for analysis, and there is a problem with the real-time performance of the analysis values.
Because the detection was performed at one point, the effective value of the component was not known, so it was only a clue to the state of combustion in the furnace.

(3) The ITV camera attached to the top of the furnace.

Flames were photographed from the opposing burners, but the images showed the flames swirling, so operators had no choice but to rely on their experience and intuition to judge the combustion state. Furthermore, when installing an ITV camera, an explosion-proof mechanism is essential as a safety measure, and its maintenance is difficult work.

An object of the present invention is to memorize the influence of changes in combustion conditions on model coefficients for estimating the combustion state using feature quantities of flame images as knowledge, and to sequentially modify the model coefficients using this knowledge to improve combustion efficiency. An object of the present invention is to provide a method and device for estimating a state.

[Means to solve the problem]

The above problem can be solved by measuring the flame as an image, processing the image to calculate the feature quantity of the flame, estimating the exhaust gas state quantity of the flame from the feature quantity and pre-stored estimation calculation conditions, and then measure the exhaust gas state quantity of the flame, calculate the difference between the estimated exhaust gas state fi amount and the measured exhaust gas amount, and obtain knowledge about the relationship between the difference and the pre-stored calculation conditions of the exhaust gas state quantity. correcting the estimation calculation conditions based on the estimation calculation conditions and diagnosing the flame exhaust gas state quantity from the modified estimation calculation conditions and the feature amount, or capturing the flame as an image and guiding the image to a predetermined position; a camera for photographing the image captured by the camera; an analog-to-digital converter for converting the image captured by the camera from analog to digital; and a frame memory for storing the image converted by the converter. a processor that performs image processing on the image stored in the frame memory to calculate a feature quantity and estimates the exhaust gas state quantity of the flame from pre-stored estimation calculation conditions; and a processor that measures the exhaust gas state quantity of the flame. A relationship between an exhaust gas measuring device and a calculation condition for an exhaust gas state amount in which a difference between an estimated exhaust gas state amount estimated by the processor and a measured exhaust gas state amount measured by the exhaust gas measuring device is stored in advance. a knowledge processor for modifying the estimation calculation conditions based on knowledge about the estimation calculation conditions; and means for transmitting the modified estimation calculation conditions to the processor.

The present invention is solved by providing a display device for displaying the output of the knowledge processor.

[Effect]

A feature quantity is calculated from an image of the flame, and a difference is calculated between the estimated exhaust gas state quantity of the flame calculated from the feature quantity and pre-stored estimation calculation conditions and the measured exhaust gas state quantity obtained by measuring the exhaust gas of the flame. Then, the estimation calculation conditions are modified based on the knowledge of the relationship between the difference and the pre-stored calculation conditions for the exhaust gas state quantity, and the flame exhaust gas state quantity is determined from the modified estimation calculation conditions and the feature quantity. Estimate.

〔Example〕

An embodiment of the present invention will be described with reference to FIGS. 1 and 3 to 6.

In a power generation boiler, as shown in Figure 1, each stage burner 1
Air is introduced into the furnace 12 along with fuel from the furnace 12, and the heat generated by combustion in the furnace 12 is supplied to the heat transfer tubes 13, and exhaust gas is released into the atmosphere from a chimney (not shown). In such a boiler, it is empirically known that the state of the flame at the base of the burner 1 greatly influences the combustion state, including the state downstream thereof.

FIG. 1 is a block diagram showing the entire embodiment. The device of this embodiment measures the flame at the base of the burner 1 as an image using an image fiber 3 protected by a protective tube 2, and is installed outside the furnace. captured on ITV camera 4. The captured flame image is A/D converted by the A/D converter 5, stored in the frame memory 6, and then processed by the processor 7 (7i (i=1, 2,
3. ...))) to perform image processing and extract the flame features. Focusing on the correlation between this feature amount and exhaust gas (NOx, unburned gas, etc.), exhaust gas components are estimated using an estimation model using the feature amount.

However, the exhaust gas components vary greatly depending on the load, coal type, and fuel type (pulverized coal, CWM (coal/water slurry)5, etc.). Therefore, by correcting and changing the estimation model using the AI processor 8, highly accurate estimation results are always obtained.

The flame image captured in the frame memory 6 is subjected to the processing shown in FIG. 3 by the processor 7.

The image input processing 7a attempts to check and average the stability of the flame (for example, equation (1)) by averaging one image for determining the presence or absence of flame.

Here, X: average of input images The flame image is preprocessed by performing correction (correction), noise removal (removal of noise components included in the flame image), and the like.

Next, in the image processing 7c, half-threshold processing (processing using the following equation (2) on a grayscale image), edge processing 1, etc. are performed.

Here, the concentration Xth at the position of X(i, j); coordinates b*j);
The semi-valued limit value feature amount extraction process 7d extracts feature amounts from the image using area calculation, center of gravity calculation, perimeter calculation, distance calculation, and the like.

The exhaust gas estimation process 7e uses the extracted feature quantities to perform (3)
The exhaust gas state quantity is estimated using the formula. As an example, NOx
(Nitrogen oxides) The estimated amount pNOx is shown below.

Here K1. K2. k,, k2. ”'-=k n
; coefficient flame characteristic quantity lNOX; NOx estimation index A; flame spread B; flame ignitability C; flame temperature (or flame brightness) D7 flame length It is transmitted to the AI processor 8 through a line.

The process flow of the processor 7 described above is shown in the fourth section.
It is a diagram.

The transmitted estimated amount of exhaust gas is processed by the AI processor 8 as follows. As an example of processing, NOx estimated amount p
FIG. 5 shows NOx.

(1) Incorporating exhaust gas analysis results C○ (-carbon oxide), 02 (oxygen), in steady state
Exhaust gas state quantities such as NOx (this value is referred to as aNOx) are detected by the detection end 10 and taken into the AI processor 8 via the PIlo (process person/output 11if) 11.

(2) Taking in the estimation result of the processor 71 The result (denoted as p No x) estimated by the processor 71 (1"It 2y 3+'; the number of processors 7) is taken into the AI processor 8 through a line.

(3) Determine the difference between aNOx and pNOx Use equation (4) to find the difference dN○X between aNOx and pNOx,
It is checked whether the difference is within a preset limit range.

dNOx=aNOx−pNOx−(4)LL≦
If dNOx≦UL, where LL; lower limit value UL; lower limit value dNOx; deviation LL>dNOx, pNO* exceeds the limit range and is estimated to be larger than the actual NOx. Also, d
In the case of NOx>UL, since PN○X is estimated to be below the limit range, the inference section corrects/changes the estimation model.

(4) Inference (correction/change of estimation model) In the inference section, the relationship between the flame image during combustion shown in Fig. 6, its characteristic quantities, constants, etc., and NOx (= a No x), unburned matter, etc. is stored in the knowledge database.

For example, if d N Ox becomes larger than UL,
Based on the knowledge in Figure 6, it can be inferred that the coal type was changed and the fuel ratio of the changed coal type was higher than that of the previous coal type. (K engineering +
The result is that it is only necessary to correct it to Δ) and (K2+Δ).

Furthermore, it was found that the burner pattern, load, 02 fine particle size, etc. are greatly affected by changes in combustion conditions, and the relationship between these and the flame during normal operation is as shown in Table 1 below.

That is, when the ratio of 1, for example, 0□ changes, it becomes as shown in equation (5).

When the ratio of Q2 is large, Table 1 (k □ ~ to 4: Coordination) By storing such relationships as a knowledge database as in Figure 6, it is easy to add new knowledge. At the same time, more detailed inference becomes possible.

(5) Transmission to processor 7i The conclusion of the inference section is transmitted to processor 71, and the inference model is corrected/changed.

(6) Display Output The conclusion of the inference section, the difference between a N Ox and p N Ox, is displayed. At this time, if the inference process is also displayed, it will be easier to understand.

(7) Termination It is determined whether all processors 71 have been terminated.

According to this embodiment, the combustion state during boiler operation can be determined satisfactorily by sequentially correcting or changing the estimation model using knowledge.

Additionally, although there is basically no need to separate processors for estimation and inference as in this embodiment, the same knowledge can be used by multiple estimation processors, reducing the load and providing knowledge to each processor. It is easy to expand and can be configured at low cost since there is no need to install it.

〔Effect of the invention〕

According to the present invention, the exhaust gas condition is estimated from the flame image,
Compare this estimation with the measured results of the exhaust gas condition, judge the reason for the difference based on pre-memorized knowledge, modify the exhaust gas condition estimation conditions, and accurately estimate the exhaust gas condition to determine the combustion condition. Since it is possible to accurately grasp the situation, even operators with little experience can operate combustion equipment such as boilers efficiently.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a conceptual diagram showing a conventional combustion state measurement and monitoring device, Fig. 3 is a block diagram showing internal processing of the processor, and Fig. 4 is a block diagram showing the internal processing of the processor. FIG. 5 is a flowchart showing internal processing of the AI processor, and FIG. 6 is a diagram showing an example of knowledge. 1...Burner, 3...Image fiber, 4...I
TV camera, 5... A/D converter, 6... Frame memory, 7... Processor, 8... AI processor. 2nd =

Claims (2)

[Claims] (1) Measure the fire as an image, process the image to calculate the feature quantity of the flame, estimate the exhaust gas state quantity of the flame from the feature quantity and pre-stored estimation calculation conditions, and Measure the exhaust gas state quantity of the flame, calculate the difference between the estimated exhaust gas state quantity and the measured exhaust gas state quantity, and based on knowledge of the relationship between the difference and the pre-stored calculation conditions of the exhaust gas state quantity. A method for diagnosing a combustion state, comprising: modifying the calculation conditions for estimation, and diagnosing the state quantity of flame exhaust gas from the modified calculation conditions for estimation and the feature quantity. (2) means for capturing the flame as an image and guiding the image to a predetermined position; a camera for photographing the image captured by the means; and an analog camera for converting the image photographed by the camera from analog to digital. a digital converter; a frame memory that stores the image converted by the converter; and a frame memory that performs image processing on the image stored in the frame memory, calculates a feature amount, and calculates a feature value based on pre-stored estimation calculation conditions. a processor that estimates the exhaust gas state quantity of the fire, an exhaust gas measuring device that measures the exhaust gas state quantity of the fire, and calculates a difference between the estimated exhaust gas state quantity estimated by the processor and the measured exhaust gas state quantity measured by the exhaust gas measuring device. a knowledge processor that corrects the estimation calculation condition based on knowledge of the relationship between the difference and a pre-stored calculation condition of the exhaust gas state quantity; and transmits the modified estimation calculation condition to the processor. A combustion state diagnosing device comprising: a means for diagnosing a combustion state; and a display device for displaying an output of the knowledge processor.