JP2656494B2 - Flame detector - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a flame detection device, and more particularly to a flame detection device having a control unit capable of detecting a flame without any adjustment.

[Conventional technology]

In a large-sized boiler for a thermal power plant, a large-sized boiler for other business establishments, and other combustion devices, it is necessary to accurately grasp the combustion state as a precondition for appropriately controlling the combustion devices.

Taking a large boiler for a thermal power plant as an example, the increase in the number of burners installed due to the increase in the capacity of the boiler and the DSS (Daily
Start Stop) Burner flame detection by increasing the number of burner ignitions / digestions associated with operation, adopting a combustion method to suppress the generation of nitrogen oxides (NOx), and changing the combustion behavior due to diversification of fuels used. Further improvement in reliability is desired.

Here, flame detection methods are generally broadly classified into an ion flame detection method and an optical flame detection method. Of these, the ion flame detection method uses a flame contact electrode as a sensor, and this electrode burns out over time and cannot be used continuously for a long time. Therefore, it is usually used as a flame detector for ignition burners. It is only used, and the range of use is limited. On the other hand, the optical flame detection method is a method of judging the state of ignition / extinguishing based on the magnitude of the light emission intensity from the flame.In the past, a single light receiving unit was used, but this device detects the flame. The accuracy is low, and the inventors have separately proposed an apparatus having the following configuration.

FIG. 29 shows the structure of the light receiving portion of the multi-view type flame detecting device. At the tip of the device body 102 made of metal or other equivalent material, for example, grooves 100a, 10
0b, 100c are formed and these grooves 100a, 100b, 100c
The light receiving optical fibers 101a, 101b, and 101c are housed therein. As a result, the flame detector obtains a visual field in three directions, and the flame detection accuracy is greatly improved.
Note that these three optical fibers are connected to a subsequent photoelectric conversion unit (not shown) via the groove 103 and are photoelectrically converted to reach a determination unit, which determines the operating state of the burner to be monitored.

FIG. 30 shows a state in which the above-described flame detection device is installed in a burner section of a large-sized boiler.

As shown, the flame detector main body 102 is housed in a sleeve 106 to which cooling air A4 is supplied so that the main burner 1
It is arranged almost in parallel with 05, and its tip reaches the vicinity of the flame F formed by the main burner. As described above, the flame detector has three fields of view L1, L2, and L3, thereby performing accurate flame detection. A1 is primary air, A2 is secondary air, A3 is tertiary air, and G is recirculated exhaust gas. Further, in the above-described configuration, the flame detector is disposed above the main burner. However, the same effect can be obtained by disposing the flame detector below the main burner.

Next, a conventional flame judgment method in the flame detector having the above-described configuration will be described.

FIG. 16 shows a flame detection determination unit. This determination unit applies a DC (direct current) offset to the flame signal (a) converted by the photoelectric converter, and amplifies the signal level of the biased flame signal (b). , A setter adjusting mechanism 3 for transmitting a setting signal (d), and a comparing mechanism 4 for comparing the presence or absence of a flame. (E) is an output signal for flame determination.

FIGS. 17 (a) to 17 (e) are diagrams showing the waveforms of the above-mentioned signals, and each symbol corresponds to the symbol of the signal. In each of the figures, the horizontal axis represents time, and the vertical axis represents signal level. These signals (analog signals) are finally converted into digital signals as shown in (e), and are determined as binary values of ignition and fire extinguishing. More specifically, at time 0 of (i) signal (a) with respect to at level L _1, when the in level at time 0 of the signal (b) and L _2, the bias L is as in equation Become.

L = L ₂ −L ₁ This L is adjusted by the bias adjustment mechanism 1.

(Ii) The pp value (peak-to-peak value) of the signal (b), that is, the maximum signal-minimum signal is G ₁ , and the p-value of the signal (c) is
Assuming that the −p value is G ₂ , the gain G is expressed by the following equation.

G = G ₂ / G ₁ This G is adjusted by the gain adjusting mechanism 2 in FIG.

(Iii) The signal (d) is a DC signal set to a certain level. This signal (d) is set by the set value adjusting mechanism 3.

(Iv) A method for determining the flame determination signal (e).

Signal (c) ≧ (d): ignition Signal (c) <(d): extinguishing The above determination is made by the comparison mechanism 4 and output.

18 and 19 show a specific setting method for the flame detector.

 In the device having this configuration, the setting is performed in the following order.

(A) Bias adjustment The bias setting volume 9 is adjusted so that the output voltage display lamp indicated by reference numeral 5 in FIG. This is the bias voltage shown in FIG.

(B) Gain adjustment The flame during ignition of the burner is adjusted by the span adjustment volume (oil) 7 or the span adjustment volume (gas) 8 so that the output voltage display lamp in FIG. The voltage is 0 / 10V).

(C) Adjust with the set voltage knob so that the set voltage display lamp becomes about 2V. Reference numeral 6 denotes a lamp for displaying the set voltage volume.

(D) In FIG. 19, it is determined that the flame is present when the output voltage is equal to or higher than the set voltage, and that no flame is present when the output voltage is lower than the set voltage.

(E) In order to ensure the stability with the flame (prevent accidental misfire), the gain should be increased by the initial descent point gain, and the level of the amount of light in the presence of the flame should be saturated at the upper limit. Is common. However, in this case, there is also a possibility of a miscatch (a mistakenly checking another flame at the time of extinguishing a fire to determine that there is a flame). For this reason, the operator adjusts the gain based on his own experience and judgment according to the individual situation of the boiler and the burner installed in this boiler, and the point of flame judgment is determined. In this case, the set value voltage is generally kept constant.

(F) Gain adjustment is performed by an external switch contact signal.
The type can be set. Specifically, when the fuel used in the combustion device is oil, the amount of ultraviolet rays of the flame is small, and therefore the output voltage is low, so the gain is set high,
Conversely, when gas is used as the fuel, the amount of ultraviolet rays is large and the output voltage is high. Therefore, the constant of the gain adjustment circuit is switched so as to lower the set value of the gain.

[Problems to be solved by the invention]

The following problems have been pointed out in the above-described conventional configuration, and it is desired to solve them.

(I) The interference level is relatively stable, but the amount of flame varies greatly depending on operating conditions as described below. Therefore, in the conventional method that relies on the gain adjustment, a deviation always occurs between the adjustment of the gain and the actual operating state, and it is difficult to always make an appropriate determination.

(Ii) Even if a judgment mechanism having such a large number of set values as to be able to absorb all changes due to operating conditions can be installed, the problem remains. That is, it is practically impossible to first assume all possible situations, and more specifically, the burner must be ignited for gain adjustment and the burner must be extinguished for bias adjustment. However, the order of ignition and extinction of the burners is generally automatically determined by the load of the combustion device. Thus, it is virtually impossible to force the combustion device to perform a special operation for the flame detector, ignoring its ignition / extinguishing sequence.

(Iii) The amount of flame generally changes greatly as shown in the following example. Conventional devices cannot adequately cope with this change in light quantity and often make erroneous decisions.

 First, the reason for the change in the amount of flame light will be considered.

(A) In order to perform low NOx combustion, measures are taken to reduce the partial pressure of air (adjustably, the partial pressure of oxygen) in the wind box by introducing exhaust gas. In this case, as shown in FIG. As described above, when the air partial pressure is reduced, the light amount is reduced, and as a result, the output voltage of the flame detector is reduced, and the determination of the presence or absence of the flame is confused.

(B) Similarly, due to low NOx combustion, in a primary gas mixing burner (commonly called a PG dual burner) that mixes part of exhaust gas after the burner gun, if the exhaust gas partial pressure with respect to the primary air amount is increased, FIG. As shown in (1), the amount of flame decreases, and the same problem as in the case (a) occurs.

(C) Even the burners installed in the same burner stage and having the same injection direction have a large difference in the amount of flame light as shown in FIG. 22, which is a major obstacle in determining the presence or absence of a flame.

(D) In the PG dual burner, as shown in FIG. 23, the combustion injection hole 13 at the burner tip 11 is formed in two directions. Therefore, the burner flame is divided into two according to the state of formation of the injection holes. Therefore, the light quantity of the flame of the burner greatly differs depending on the detection position. FIG. 25 (A) shows a case where the detected light amount is the lowest because the flame detector 14 is located farthest from the flame F unevenly distributed with respect to the burner chip 12, and FIG. The case where the detected light amount becomes maximum due to the flame detector 14 being located in the flame F is shown. In the burner having this configuration, the combustion injection direction may be changed every time the burner is cleaned, so that stable flame detection can hardly be expected.

(E) In addition, a change in the amount of air supplied from the over-air port for supplying air for combustion of unburned components generated by low-oxygen combustion and a change in the load of the combustion device itself also change the amount of flame.

In view of the above, the following problems have been pointed out in addition to the problems described above as the conventional flame detector.

(I) Instability of flame detection, misfiring In the gain adjustment shown in FIG. 19, the output characteristic as shown in FIG. 26 is obtained in the flame detector due to the change in the oxygen partial pressure in the combustion air supplied to the burner. Is generated. In this case, A is gain 40%, B is gain 50%, C is gain
Assuming a characteristic of 60%, at a gain of 40%, the output voltage drops below the point d where the amount of oxygen is relatively large, and the detection accuracy decreases. The output voltage decreases when the voltage becomes below. That is, as described above, stable detection accuracy can be ensured when the output voltage is always 10 V and is in a saturated state, and the detection accuracy decreases with an output voltage lower than this. Further, when the gain is 40% or less, at the point d or less, a misfire may be determined to be extinguishing even though ignition is being performed.

(Ii) Decrease in reliability of miscatch and misfire detection Due to the circumstances described above, if the gain is increased to improve the detection accuracy, the following problem occurs.

In other words, if the gain is set to a large value, it reacts to even a small amount of light, and may detect a flame of a burner other than the detection target, for example, a flame of an opposing burner, and may generate a miscatch that emits a signal during ignition of the target burner. The nature becomes high. For example, the 27th
As shown in the figure, when an adjacent flame of several hundred mV arrives, miscatch does not occur at a gain of 40% because it is smaller than the flame presence / absence determination voltage of 2V, but becomes close to 2V at a gain of 50% or more and a gain of 60% Then it goes over 2V, leading to miscatch.
For this reason, the conventional flame detector employs a system that unconditionally determines that there is no flame when the main valve that supplies fuel to the burner is fully closed. However, in this method, if the target main valve is damaged, or if the fuel is partially supplied and combustion continues even though the valve is fully closed due to a failure of the limit switch for fully closing the main valve, etc. Helpless in Further, if the gain is increased too much, there is a possibility that the misfire detection is delayed. For example, if the burner, which had a flame with the main valve open, caused a misfire or blowout for some reason,
Since only the fuel is injected, if left unchecked, the fuel scattered in the furnace may explode and burn, resulting in a catastrophic event. For this reason, it is not permissible for the flame detector to delay the misfire detection. In particular, when the fuel is gas, it is required that the misfire state be detected within 2 to 3 seconds after the misfire and the main valve be fully closed.

(Iii) Misfire / extinguishing failure or delay occurrence Another important function of the flame detector is to shift the burner from the extinguished state to the ignition state (burner ignition) or vice versa. In the transition state of “burner extinguishing”, there is a function of accurately determining the presence or absence of a flame. In the above (i) and (ii), there was a problem related to flame detection in the absence of an ignition torch flame.
The flame detection in the transition state is performed in a state where the torch flame exists. Since the bias setting is generally performed in the absence of the torch flame, setting the gain high results in the determination of the presence of the flame using only the interference light of the torch flame. For this reason, in the automatic control system, although the burner main valve is fully closed, a misfire failure or a fire extinguishing delay occurs due to a signal indicating that there is a flame.

Conversely, if the gain is too low, the main burner flame will react with the presence of a torch flame, and it will not be possible to make an ignition determination within a specified time, resulting in ignition failure or ignition delay. If there is a failure in ignition or extinguishing, the ignition or extinguishing operation will be performed again.However, for safety reasons of the system, the program is designed so that this flame detector is excluded from the automatic control system if it fails twice in a row. As such, such ignition failures will have a significant impact on the overall system.

(Iv) Re-adjustment frequency is high and the adjustment time is long Even if the gain adjustment is appropriate, since the adjustment is only optimal at the time of adjustment, it must be re-adjusted over time. A lot of work must be devoted to the adjustment work. In addition, in many cases, a defective adjustment is found in the reconfirmation stage after the adjustment. In this case, the readjustment is unavoidable, and the adjustment time tends to be long.

[Means for solving the problem]

As described in detail above, the present invention has a number of problems with the conventional apparatus, and is configured in view of the fact that it also hinders automatic control of combustion differences. It is characterized by being a detector.

That is, a signal from a sensor that detects light is converted into an electric signal, and a value obtained by averaging interference flames with a predetermined time constant is used in a flame detection device that determines the presence or absence of a flame from a plurality of elements that affect the amount of light. A mechanism for setting as a flame presence / absence determination voltage, a mechanism for setting a set value for flame presence / absence determination for each operation state determined by a set of signal variation elements such as a load variation of the combustion device, and a mechanism for setting an unset value to the unset value. A determination mechanism having an arithmetic function for estimating an optimum set value from a set value of a known element close to a condition of an element having a set value.

[Action]

A mechanism capable of setting a value obtained by averaging interference flames with a certain time constant as a flame presence / absence determination signal minimizes the influence of flames other than the determination target.This flame presence / absence determination signal is used for flame determination by a mechanism capable of being set for each burner condition. The influence of the variation of the burner conditions in the burner condition is minimized, and an appropriate flame presence / absence judgment is made by a mechanism capable of estimating an optimum set value based on the set values of other burner conditions already set.

[Background of the present invention]

Next, before describing the embodiments of the present invention, the circumstances leading to the configuration of the present invention will be described more specifically in order to facilitate understanding of the present invention.

In the conventional flame detector, after adjusting the bias by the interference flame, it is necessary to determine what percentage of the gain is to be practically determined based on the fluctuation factors such as the operating conditions of the combustion device and the flame during ignition with a large fluctuation range. He pointed out that there was a fundamental problem in this point, in that it was necessary for the member to adjust the amount by eye while checking it with the naked eye. Compared with such a fluctuation factor, the fluctuation level of the interference flame level is considerably smaller. That is, as shown in FIG. 28, the change in the level of the interference flame is very small with respect to the change in the flame level due to the boiler load.
Therefore, with the conventional bias adjustment and gain adjustment constant,
If a value obtained by averaging the light amount level of the interference flame with a certain time constant (for example, 60 seconds) is used as the flame presence / absence determination signal, it is possible to obtain an extremely stable set value.

The light level of the interference flame depends on the oxygen partial pressure of the combustion air supplied to the wind box and the burner, the exhaust gas recirculation ratio, the combustion injection direction from the burner, the supply amount of unburned combustion air from the over air port, and the burner. Is generally relatively stable. On the other hand, the fluctuation range due to the fluctuation factors such as the presence or absence of the torch flame and the boiler load is relatively large. For this reason, if the mechanism is such that the flame presence / absence setting signal can be set by a variable element having a large fluctuation range, highly reliable setting is possible. In the setting by the conventional gain adjustment method, it is theoretically possible to have a set value for each variable element because the fluctuation range of the variable element is too large, but in practice, this setting is almost impossible.

In a large boiler for a thermal power plant, the number of burners to be installed is very large, and the order of ignition and extinction of each burner is very complicated. Specific logic is required for flame detection as well as the performance of the flame detector located at
FIG. 12 shows the relationship between the burner ignition state of a boiler having 48 burners installed in a furnace and the boiler load. The order of firing and extinguishing the burner in the boiler should be such that the heat load on each part of the boiler is as uniform as possible.
Care is taken to minimize the thermal stress on the boiler during fire extinguishing. For this reason, the order of ignition and extinction of each burner is very complicated as shown in FIG. In addition, the number described at the tip in each graph indicates the firing order of the burners.

Specifically, for example, when learning is started while the boiler is stopped and learning is performed with the load pattern of image transmission, the fire is extinguished with a load of 150 MW to 300 MW, for example, cells C-1 and C-4. The burner is set in three load bands up to 300 MW, but cannot be set in a load band of 300 MW or more. However, in normal boiler operation, you will only experience ignition and fire extinguishing in the pattern shown in the figure.Thus, even though it is not necessary to actually set all the burners, it is necessary to operate at full load. Or operate a specific burner in preparation for applying excessive stress to the boiler. For this reason, in the above-described example, if other burners can be set using the inference mechanism based on the conditions already set up to 300 MW, the presence or absence of a flame can be determined in all conditions. You can do it. The present invention is configured based on such an idea.

〔Example〕

Next, examples of the present invention will be described. In order to facilitate understanding of the contents, the description will be made separately for the overall configuration and signal processing.

(1) Overall Configuration (i) In FIG. 1, a burner flame is detected by a flame detector 15. The detected flame is fiber optic cable 16
To the detector panel 20 via the The optical fiber cable 16 is directly connected to a photoelectric converter 17 provided in the detector panel 20, and the detected light is converted into an electric signal here. The converted electric signal is input to the determination board 18 to determine the presence or absence of a flame. The judgment board 18 can know the flame signal output from another flame judgment board via the normal board 19. Flame judgment board
The operation state of the burner base valve is input to 18 as a signal, and the operation signal of the burner base valve is output to the automatic burner board 30 via the relay board 21 based on the result determined by the determination board. Operating conditions such as boiler load information and fuel pressure are input to the communication board 19, and the data is output to all the decision boards. (Ii) As shown in FIG. 29, the tip of the flame detector has a core diameter of, for example, 800 μm.
The optical fibers 100a, 100b, and 100c are set such that the tips of the fibers are oriented toward the viewing directions of 0, 15, and 30 degrees with respect to the axial direction of the tip of each of the optical fibers.

(Iii) Three optical fiber cables introduced into the panel 1
00a, 100b, and 100c are independently converted to electric signals by two light receiving elements, a visible light receiving element and an infrared light receiving element, for each fiber by the composite photoelectric conversion element in the photoelectric converter 17, and then amplified. Is done. Therefore, the output of the photoelectric converter, that is, the output to the determination board is 3 (the number of visual fields) × 2 (the number of channels) = 6.

(2) Signal Processing (i) Current Light Quantity Calculation Referring to FIG. 2, the signals of channels 1 to 6 are frequency-decomposed into eight frequency bands for each channel. The lower three frequency bands are called region 0, and the upper five frequency bands are called region 1. In this way, the frequency range is set in advance, and in the case of a gas flame, the power spectra for the region 0 and the region 1 are calculated and used as the current light amount. Also, at the time of oil combustion, the power spectrum is calculated only for the region 1 and is set as the current light amount.

(Ii) Interference flame learning In Fig. 3, the set value for making the ignition / extinguishing judgment is an instruction from the observer when the main valve is closed (when extinguishing), that is, the set switch is set to ON (SET SW ON). Is set by The settings are recorded independently for each channel and for each presence or absence of a torch flame.

The set value is based on the current light amount, the maximum value of the instantaneous value of about 20 seconds time constant (referred to as “instantaneous maximum value”) and the maximum value of the average value of about 60 seconds time constant (referred to as “average maximum value”). And a minimum value (referred to as “average minimum value”).

The same calculation is performed for the remaining five channels 2 to 6, and the result is recorded as a set value.

The setting value of the flame presence / absence determination is set as shown in Table 1 below.

(Iii) Calculation of hit ratio In FIG. 4, when the burner main valve is open, the hit ratio for each channel is calculated.

If the value of h is (a) when the current light quantity ≧ the instantaneous maximum value, h = 1.0. (b) When the current light quantity <the instantaneous maximum value, h = −1.0, the hit rate H is H _NEW = (H _OLD · n + h ) / (N + h) is updated at each calculation cycle. n is reset to n = 1 when the main valve changes from closed to open or when the burner state changes. In this case, the change of the value of H is -1 ≦ H ≦ 1, and the meaning is as shown in the following table.

(Iii) Channel Weight Calculation In FIGS. 5 and 6, the value obtained by converting the hit ratio of each channel into an integer value is referred to as a channel weight (W).

Where W is the maximum integer not exceeding W = 10 ^H.

More specifically, it is an integer value of 0 ≦ W ≦ 10, and an area (H> 0.0) where the difference in light amount is small depending on the presence or absence of a flame can be evaluated in nine levels from 1 to 10.

In the drawing, the occurrence of a Ch error indicates a channel whose range is large or whose light amount is small in the self-diagnosis.

(V) Single Channel Determination In FIG. 7, a single channel determination value D is determined for each channel by comparing the current light amount with a set value. At this time, if a set value is set, an optimum value is estimated from the set values already set and used as a set value (hereinafter, this function is referred to as an “estimate function”). It is determined as shown in Table 2 below.

(A) When current light quantity <average minimum value D = 0 (absolute fire extinguishing area) (b) Average maximum value ≦ current light quantity <average minimum value D = 1 (fire extinguishing area) (c) Average maximum value ≦ When current light quantity <momentary maximum value D = 2 (fire extinguishable area) (d) Instantaneous maximum value ≦ current light quantity D = 3 (ignition area) (vi) Judgment for all channels In FIG. By integrating the data for the six channels, an all-channel determination value TD is obtained. According to the determination method shown in FIG. 7 based on the D value for 6 channels, the number of channels, and the channel weight, TD is determined for the case where the main valve is open and the case where the main valve is closed.

Here, D is an integer value of −100 ≦ + D <100, and is larger as the possibility of ignition is higher and smaller as the possibility of fire extinguishing is higher.

(Vii) Comprehensive judgment In FIG. 9, all channel judgment values TD are summed with the previous value, and the maximum and minimum are set to 100 and −100, and the value is comprehensively judged TS.
And

The numbers in parentheses in the figure indicate the lower limit of the channel weight used for the determination. That is, the comparison is performed except for channels having channel weights smaller than the numerical value in parentheses.

(Vii) Flame presence / absence output In FIG. 10, according to the comprehensive judgment TS, (a) If TS ≦ −100, fire extinguishing is set.

(B) When −100 <TS <100 Contact keep.

(C) In the case of 100 ≦ TS Ignition contact.

(Ix) Judgment suspension / impossible judgment In all channel judgments, when TD = 0 (when it is not possible to judge the presence or absence of flame), the calculation cycle is held as judgment judgment, and if it continues for 5 minutes or more, Dirt on the light receiving portion of the flame detector issues a warning or the like to the monitoring person by informing the burner combustion state is unstable.

(X) Communication system In FIG. 11, one burner stage (for example, burner 8
1) communication board (CCB) 19 and internal bus 2
Connected at 8. Communication boards are also provided at other stages, and each communication boat is connected by each GP-IB bus 29.

(A) The CCB 19 periodically issues a data transmission request to the decision board (SBC) 18 and receives the flame information of the decision board. At this time, the communication board 19 receives only information from the determination board connected to its own internal bus.

(B) The other-stage SBC flame information held by the other-stage CCB is transferred to CP-
Receive through IB bus 29.

(C) When the flame information of the other SBC is complete, edit the information including the information of the SBC of the own stage and the input information such as the load of the CCB.
Send to BC all at once.

(D) By connecting a host computer to the GP-IB bus 29, necessary information is displayed by a method such as graphic display.

Next, an inference (inference) function that plays a central function in the present invention will be described.

(1) Influence on light quantity and burner condition In the table, N indicates the number of patterns.

Assume that 60 burner conditions are prepared and set for each condition. At this time, for example, as shown in FIG. 28, changes in different conditions for the same item are often continuous smooth curves, and the effect on the change in light amount differs for each item. This effect is in the order indicated by the number of patterns in Table 4 above, and it is known that the amount of light varies greatly with different types of fuel, and the effect of the GM ratio is small.

(2) Setting value recording method As shown in FIG. 13, setting values are recorded in a so-called tree structure. This tree is created with the levels 0 to 3 set in the descending order of the influence on the light quantity, that is, in the order of the torch flame, the load, and the GM. This tree has only a root when all the patterns are not set, and when setting of one pattern is completed, a path from the root of the tree to a lower set value is sequentially made one by one.

(3) Estimation Method In FIG. 14, for example, it is assumed that patterns have been set for and. Here, for example, a burner is ignited at a boiler load of 300 MW, and an interference value cannot be set unless the burner is extinguished. Therefore, when the load increases to 450 MW (pattern), the interference value cannot be set.
For this reason, the average value of the set values of the adjacent and already set patterns is used as the set value of the pattern. Since the level of influence on the light quantity is in descending order from level 0, the pattern is most appropriate, that is, the condition distance is the shortest, and the pattern with torch flame is not used as an estimated value. That is, a set value included in a tree of a portion having a node at a set lighting level as a root is selected, and an average value thereof is used as an estimated value.
Note that the portion connected by the dotted line in the figure indicates that no setting has been made.

(4) Estimation example Next, it is confirmed that the above estimation method is effective.
The case of the -1 upper burner will be described as an example.

(Set values) In FIGS. 15 (A), (B) and (C), the values indicated by the solid lines indicate the instantaneous maximum values in each load band. Indicates an ignition range.

(Estimated value) Since the C-1 upper burner is ignited in a load range of 150 to 300 MW, it cannot be set in a load range of 300 to 450 MW or 450 to 600 MW in normal load operation. Therefore, values estimated from 0 to 60 MW, 60 to 150 MW, and 150 to 300 MW based on the method described above are indicated by dotted lines in the figure. Therefore, when the load is set in normal load operation, the area below the dotted line is the fire extinguishing area, and the area above the dotted line is the ignition area.

(Fire extinguishing standard) The maximum value of the average light amount at the time of burner ignition is represented by ○-○ in the figure, and the minimum value is represented by △-△. Burner flames are ○-○ and △-△
Since it fluctuates in the range surrounded by, the flame determination can be performed if the range is stable and the ignition range.

(Evaluation) In this embodiment, the flame determination is possible if at least one of the six channels is valid. In gas combustion, the visible band is almost zero, so the evaluation is made in three infrared channels.
In the figure, even if the estimation channel is used, the above criterion is sufficiently satisfied in the 0-degree infrared region, and the conditions are satisfied in the estimation portion even in the 15-degree and 30-degree infrared regions.

〔effect〕

The present invention provides a mechanism for setting a value obtained by averaging interference flames with a predetermined time constant as a flame presence / absence determination signal as described in detail above, and using this flame presence / absence determination signal as a load fluctuation of a combustion device. A flame detection device including a mechanism for setting for each burner condition, and a determination mechanism having a function of estimating an optimal set value based on set values of other burner conditions that have also been set for unset burner conditions. (I) The interference flame to be set has a small fluctuation range due to the burner state with respect to the light quantity at the time of ignition, has high reproducibility, and has a high reliability of the set value because the judgment signal is set for each burner state. There is no fire extinguishing failure.

(Ii) It has an estimating mechanism that can set the optimal set value based on other set values, so by implementing the actual load operation of the combustion device such as a boiler,
It can be used immediately without any separate adjustment, and the practicality can be greatly improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing an overall configuration of a flame detection device according to an embodiment of the present invention, FIG. 2 is an operation system diagram showing a concept of current light amount calculation, FIG. 3 is a system diagram showing a concept of interference flame learning, FIG. 4 is an operation system diagram showing the concept of hit ratio calculation, FIG. 5 is an operation system diagram showing the concept of channel weight calculation, FIG. 6 is a diagram showing a change state of the hit ratio, and FIG. FIG. 8 is a system diagram showing the concept of the all channel judgment method, FIG. 9 is a system diagram showing the concept of the comprehensive judgment method, FIG. 10 is a conceptual diagram showing the flame judgment logic,
FIG. 11 is a system diagram showing a communication function of the entire apparatus, FIG. 12 is a diagram showing a boiler load and an ignition state of a burner, FIG. 13 and FIG.
FIG. 14 is a theoretical tree showing a method of estimating set values, FIGS. 15 (A) to 15 (C) are diagrams showing a method of estimating set values, and FIG. 16 is a system diagram of a flame detection mechanism of a conventional flame detector. FIG. 17 is a diagram showing a signal output state in each circuit of the conventional device.
FIG. 18 is a front view of a control panel of the conventional device, FIG. 19 is a diagram showing a flame determination method in the conventional device, and FIG. 20 is a graph showing the relationship between the oxygen partial pressure in the air supplied to the burner and the amount of flame. FIG. 21 is a diagram showing the relationship between the PG ratio and the amount of flame light, and FIG.
FIG. 22 is a diagram showing the relationship between the burner installation location and the light amount of the burner flame, FIG. 23 (A) is a front view of the burner chip of the PG dual burner, FIG. 23 (B) is a side view of the burner chip, and FIG. 25A is a diagram showing the relationship between the ejection direction and the light amount,
(B) is a conceptual diagram showing an arrangement state of a burner flame and a flame detector. FIG. 26 is a diagram showing a relationship between an oxygen partial pressure in air supplied to the burner portion and an output voltage of the flame detector. Is a graph showing the relationship between the gain and the output voltage, and FIG. 28 is a diagram showing the relationship between the light quantity of the ignition flame and the interference flame and the boiler load.
Fig. 29 is a side view of the light receiving part showing details of the light receiving part of the flame detector,
FIG. 30 is a side view of a wind box showing an arrangement state of a flame detector. 15 Flame detector, 16 Fiber optic cable, 17
... photoelectric converter, 20 ... operation panel

Claims (1)

(57) [Claims] 1. A flame detecting device for converting a signal of a sensor for detecting light into an electric signal and determining presence / absence of a flame from a plurality of factors affecting a light amount, wherein an interference flame is averaged by a predetermined time constant. A mechanism for setting the determined value as the flame presence / absence determination voltage, a mechanism for setting the set value of the flame presence / absence determination for each operation state determined by a set of signal variation elements such as load variation of the combustion device, and a mechanism for setting an unset value. A determination mechanism having an arithmetic function for estimating an optimal set value from a set value of a known element that is close to the condition of the element of the unset set value.