JP2003217047A - Flame detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flame detector capable of highly accurately detecting a flame by eliminating a source of misinformation and filtering out noise. <P>SOLUTION: The flame detector detecting a generated flame in the event of fire by detecting a light of a wavelength specific to the flame is provided with a plurality of light detecting means detecting a plurality of lights of different wavelengths one another specific to the flame and outputting a detection signal according to each wavelength, a calculating means calculating a normalized correlation value between each of waveforms for each detection signal detected by the plurality of the light detecting means and a ratio of a moving average between each of waveforms for each detection signal, and a flame generation deciding means deciding as the flame is generated when the normalized correlation value and the ratio value of the moving average calculated by the calculating means respectively fall within prescribed ranges. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a flame detecting device,
More specifically, light with a wavelength peculiar to a flame (for example, infrared rays,
The present invention relates to a flame detection device that detects a flame generated during a fire by detecting (e.g., ultraviolet rays).

[0002]

2. Description of the Related Art Two conventional inventions of this kind are known, for example, the invention of Japanese Patent Publication No. 6-611119 (Invention A) and the invention of Japanese Patent Application Laid-Open No. 2000-57456 (Invention B). . The cross-correlation fire detector according to Invention A includes a first detector and a second detector, and cross-correlation function signals (correlation value) of outputs from these two detectors.
And the output ratio are used for fire judgment. In the cross-correlation circuit of Invention A, the sum of the multiplication values of the data of the two signal waveforms at the same time is used as the cross-correlation function signal. Further, in the flame detecting device of the above-mentioned invention B, the output signal of the detecting element is sampled at the sampling frequency of 32 Hz by the microprocessor via the prefilter and the amplifying section.

[0003]

The above-mentioned invention A utilizes the output signals of the two detectors and simplifies the circuit structure, but the cross-correlation function signal is not normalized and the maximum possible value is Since it is unknown, even if it is known that the two waveforms are similar if the cross-correlation function signal is large, it is not possible to know how similar they are. Therefore, a differentiating circuit and a second-order differentiating circuit for differentiating the waveform are further provided to judge the fire by observing the phase of the waveform from the output. On the other hand, the flame detection device of the invention B uses 32 Hz as the sampling frequency of the microprocessor as described above.

Therefore, in Invention B, due to the sampling theorem, aliasing noise occurs when a frequency component of 16 Hz or more is input into the microprocessor. In addition, the pre-filter and the amplifier are usually integrated,
When noise is mixed in here, it is always amplified. Therefore, the commercial power frequency (50
(60 Hz, 60 Hz) noise becomes aliasing noise in the microprocessor, and commercial energy frequency noise that becomes aliasing noise is superimposed near the fluctuation frequency of the flame used for flame determination. As described above, the devices of the conventional inventions A and B have the above-mentioned problems.

The present invention has been made in order to solve the above-mentioned problems of the conventional device, and it is a "normalized correlation value" indicating the degree of similarity between the waveforms of the detection signals detected by the plurality of photodetection means. And the ratio value of the moving average that indicates the amplitude ratio between the waveforms, a flame that can detect flames with high accuracy by eliminating false sources of light captured by the detection unit. It is intended to realize a detection device.
Another object of the present invention is to realize a flame detection device that can detect flames with high accuracy by eliminating commercial power supply frequency noise.

[0006]

SUMMARY OF THE INVENTION The present invention is a flame detecting device for detecting light having a wavelength peculiar to a flame to detect a flame generated during a fire by detecting a plurality of lights having different wavelengths peculiar to the flame. A plurality of light detecting means for outputting a detection signal corresponding to each wavelength, a normalized correlation value between the waveforms of the respective detection signals detected by the plurality of light detecting means, and a moving average of the waveforms of the respective detection signals. And a flame generation determination means for determining that a flame has occurred when the normalized correlation value and the moving average ratio value calculated by the calculation means are within a predetermined range. The flame detecting device is configured. Further, in the above, the flame occurrence determination means is a flame detection device that determines that it is a flame when the ratio value of the normalized correlation value and the moving average is within a predetermined range in a predetermined time or more. It is composed. Further, in the above description, the number of the plurality of light detecting means is three or more, forming the plane coordinates around the normalized correlation value and the ratio value of the moving average, and the normalized correlation values between the waveforms of the respective detection signals. And the ratio value of the moving average, and each coordinate point is plotted on a plane coordinate, the calculation means calculates the distance between the coordinate points, and the flame occurrence determination means determines the flame when the distance is within a certain range. The flame detection device is configured to determine that the occurrence of the above has occurred. Further, in the above, the flame occurrence determination means is a flame when the normalized correlation value and the ratio value of the moving average and the distance are within a predetermined range and within a certain range within a certain period of time. This is a flame detection device configured to judge that.
Further, in the above description, the flame occurrence determination means constitutes a flame detection device that performs flame determination using the center of gravity of the normalized correlation value and the ratio value of the moving average and the standard deviation. Further, in the above description, the plurality of light detecting means constitutes a flame detecting device including a first light detecting means for detecting infrared rays having a wavelength in the 4.4 μm band, which is peculiar to a flame. Further, the present invention is a flame detection device for detecting light having a wavelength peculiar to a flame to detect a flame generated at the time of a fire, and a light detecting means for detecting light having a wavelength peculiar to the flame and outputting a detection signal. And a sampling means for sampling the detection signal detected by the light detection means at a sampling frequency higher than 120 Hz. Further, in the above, the flame detection device is configured so that the sampling frequency is a power of 2.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION Generally, a flame generated during a fire has an inflection point (peak value) near a 4.4 μm band due to resonance radiation of CO ₂ gas as shown by a solid curve Wf in FIG. It is known that infrared rays with a wavelength peculiar to a flame are emitted. The three-wavelength flame detection device, as shown by the three dotted lines in the figure, has a 3.9 μm band centered on the 4.4 μm band and a 5.
The detection target is the 0 μm band, and a light detection unit for detecting light (infrared rays) in each wavelength band is provided. In addition, a detection method for detecting infrared rays in the 4.4 μm band and the 3.9 μm band or the 5.0 μm band is adopted in the two-wavelength system. The horizontal axis and the vertical axis in FIG. 4 are wavelength and relative intensity, the curve Ws shows the sunlight, and the curves Wmh and Wml show the emission spectrum distributions of the high temperature object and the low temperature object in the black body radiation, respectively, and their peak values. Is determined by the temperature of the blackbody.

On the other hand, in the case of infrared rays radiated from the same radiation source like a flame at the time of fire, the detection signal of the light detecting means for detecting the light in the 4.4 μm band and the light other than the 4.4 μm band are detected. The similarity between the two signal waveforms and the detection signal of the detection means is high. Further, as shown by the curve Wf, in the case of infrared rays emitted from the flame, the relative intensity of infrared rays in the 4.4 μm band is large, and the relative intensity of infrared rays other than the 4.4 μm band is small. That is, the output of the detection signal detected by the photodetector in the 4.4 μm band is large, and the output of the detection signal detected in the region other than the 4.4 μm band is small. Therefore,
Ratio of the output amplitude of the waveform other than the 4.4 μm band to the waveform of 4.4 μm (the output amplitude of the waveform detected from other than the 4.4 μm band / the output amplitude of the waveform detected from the 4.4 μm band) Is small and limited to a predetermined range.

On the other hand, the curves Ws, Wmh, Wml
As shown by, the range of the ratio value of the infrared rays radiated from a heat source other than the flame or a noise source is completely different from that of the flame. Therefore, the degree of similarity (referred to as a normalized correlation value γ) between the waveforms of the respective detection signals detected by the photodetection means other than the 4.4 μm band and the 4.4 μm band, and the waveforms of the respective detection signals. By using the amplitude ratio (referred to as a moving average ratio value β) for flame determination, it is possible to perform highly accurate flame detection by eliminating false sources of heat and noise. Since the normalized correlation value γ is normalized to a value in the range of 1 to -1, it is easy to judge the similarity between the waveforms, and the correlation between the waveforms increases as the value approaches 1. The present invention detects a plurality of lights having different wavelengths peculiar to flames, and shifts the normalized correlation value γ indicating the degree of similarity between the waveforms of the detection signals and the amplitude ratio between the waveforms of the detection signals. It is an object of the present invention to realize a flame detection device capable of highly accurately discriminating a flame generated during a fire by using the average ratio value β. In the case of flame, the normalized correlation value γ and the ratio value β of the moving average are within the respective predetermined ranges, so that the flame generated during a fire can be discriminated with high accuracy.

Then, as will be described later, photodetection means for the 4.4 μm band and photodetection means for selecting and detecting wavelengths different from each other other than the two or more 4.4 μm bands (for example, 3.9).
.mu.m band and 5.0 .mu.m band), the normalized correlation value .gamma. and the moving average ratio value .beta. by the 4.4 .mu.m band photodetecting means and the 3.9 .mu.m band photodetecting means are calculated, and .4 μm
The normalized correlation value γ and the moving average ratio value β are calculated by the band light detecting means and the 5.0 μm band light detecting means, and the normalized correlation value γ and the moving average ratio value β are used as axes. After forming the plane coordinates, plotting each coordinate point by the respective normalized correlation value γ and the ratio value β of the moving average on the plane coordinates,
When the distance between coordinate points is calculated, it is within a certain range in the case of a flame, so this characteristic was applied to flame determination to realize a flame detection device capable of highly accurate flame detection.

It should be noted that the flame occurrence discriminating means is configured such that the normalized correlation value γ, the moving average ratio value β, and the distance are within a predetermined range and within a predetermined range within a predetermined time. If it is determined that the flame is present in some cases, more highly accurate flame detection becomes possible. Further, when the flame occurrence determination means performs flame determination using the center of gravity of the normalized correlation value γ and the ratio value β of the moving average and the standard deviation, it is possible to detect the flame with higher accuracy.

Embodiment 1 FIG. 1 is a block diagram showing the configuration of Embodiment 1 of the present invention,
FIG. 2 is a flowchart showing the operation of the first embodiment. In the first embodiment, the present invention is described as a three-wavelength flame detection device. In FIG. 1, reference numeral 1 is a flame detection device. 21 to 23 are A, B, and C detection units, and 31 to 33 are amplifiers 41 to 4 that amplify the outputs of the detection units 21 to 23.
3 is an anti-aliasing filter (anti-a) that removes a specific frequency component, which will be described later, from the outputs of the amplifiers 31 to 33.
lias-filter) and 5 are A / D converters. Further, 6 is a judgment unit, 7 is a fire signal generation circuit, and 8 is a general-purpose multiplier. The general-purpose multiplier 8 is composed of a circuit specialized for multiplication and operating at high speed. The general-purpose multiplier 8 may be a product-sum calculator such as a digital signal processor (DSP), or may be a numerical calculator. The detection units 21 to 23 are examples of a plurality of light detection units that detect a plurality of lights having different wavelengths peculiar to a flame and output a detection signal corresponding to each wavelength. The general-purpose multiplier 8 calculates a normalized correlation value indicating the degree of similarity between the waveforms of the detection signals detected by the plurality of photodetecting means and a moving average ratio value indicating the amplitude ratio between the waveforms of the detection signals. It is an example of calculation means for calculating. The determination unit 6 is an example of a flame generation determination unit that determines that a flame has occurred when the normalized correlation value and the moving average ratio value calculated by the calculation unit are within predetermined ranges.

Here, although the explanation is repeated prior to the explanation of the flow chart of FIG. 2, a part of the explanation of the operation is added and the moving averages α _An , α _Bn , α _Cn and the normalized correlation values used in the present invention are added. γ
_{The formulas} for calculating _ABn and γ _ACn are explained below. A ~
Each of the detection units 21 to 23 of C has sensitivity to infrared rays in a specific wavelength band, for example, 4.4 μm, 3.9 μm, and 5.0 μm bands. The amplifiers 31 to 33 add a bias voltage to the detection signals of the detection units 21 to 23 to make it a positive output, and further select and amplify a signal component of 4 Hz or more, which is an electric flame flicker frequency band, among the detection signals. . As anti-aliasing filters 41 to 43, as pre-processing for sampling, the analog signals from which the frequency components of 1/2 or more of the sampling frequency (128 Hz in this case) of the amplified signals from amplifiers 31 to 33 are removed are output. However, since the anti-aliasing filter is an analog circuit, it is difficult to make a high-order filter with high accuracy and at low cost. Since a low-pass filter of about 3rd to 4th order is used for the antialiasing filter, commercial power supply frequency noise (50 Hz, 60 Hz) is slightly mixed.

The A / D converter 5 converts the analog signal passed through the anti-aliasing filters 41 to 43 into a digital signal by sampling it at a sampling frequency of 120 Hz which is a power of 2 and exceeds 120 Hz. The judgment unit 6 stores a program (not shown), a digital filter, a RAM for temporarily storing digital signals from the A / D converter 5 and the like, and an R for storing a flame judgment value described later and the like.
It has an OM and an MPU that controls them. By the program, the digital signal is taken into the judging unit 6 at the sampling frequency exceeding 120 Hz and temporarily stored in the RAM. In this case, the sampling frequency is 2n (n = 7) 128 so that the determination unit 6 can easily perform the calculation.
It is set to Hz. In addition, the determination unit 6 uses a digital filter described later to electrically generate a flame flicker frequency band of 4 to 30 Hz in the digital signal stored in the RAM.
Is extracted, and the extracted signal is temporarily stored in the RAM. Here, since the sampling frequency is set to be twice as high as 60 Hz or more, the commercial power supply frequency noise (50 Hz, 60 Hz) mixed in the analog signal from the anti-aliasing filter does not become aliasing noise and can be removed by the digital filter. . Since the extracted signal includes the bias voltage, the calculation signal obtained by dividing the extracted signal by the bias voltage is calculated and temporarily stored in the RAM of the determination unit 6. For example, the bias voltage is calculated by dividing the average value of the extracted signals for the past N samplings by the number N of extracted signals, and is divided from this extracted signal to calculate the calculation signal.

The general-purpose multiplier 8 is used for the calculation of the digital filter, and the normalized correlation value γ and the moving average ratio value β are calculated by using a predetermined number of calculation signals temporarily stored in the RAM of the judging section 6. The distance L1 between two points, which will be described later, is calculated from the normalized correlation value γ and the moving average ratio value β, and is temporarily stored in the RAM of the determination unit 6. Then, the judgment unit 6
The flame judgment is performed by comparing the normalized correlation value γ, the moving average ratio value β, the distance L1 between the two points, and the like stored in the AM with the flame judgment value and the like stored in the ROM. The fire signal generation circuit 7 outputs a fire signal based on the flame judgment result by the judgment unit 6.

Further, the functions of the judging section 6 and the general-purpose multiplier 8 will be described in detail below. The digital filter described above is a bandpass filter for removing noise such as commercial power supply frequency and detecting flame fluctuations, and is realized by a trans-versal-filter or the like. Although other configurations may be used, the transversal filter has a linear phase characteristic and requires less waveform distortion. The above transversal filter is given by: y (t) = w ₀ x (t) + w ₁ x (t−Δt) + w ₂ x (t−2
Δt) ... w _N-1 x (t- (N-1) Δt) where wi: weighting coefficient i = 0 to N-1 x (t): digital signal at time t Δt: sampling interval y (t) : Output result (extracted signal)

Based on the calculation signal calculated using the extracted signal obtained by the above method, the detection units 21 to 2 for each wavelength band
As the output amplitude of 3, the moving averages α _An , α _Bn , α _Cn of the absolute values of the calculation signals within a constant interval (predetermined number), and the normalized correlation values γ _ABn , γ _ACn of different wavelengths are _obtained , respectively. Also,
Absolute moving average α _{Bn with} respect to the moving average α _An ,
Ratio values of α _Cn β _ABn (= α _Bn , / α _An ), β _ACn (= α _Cn
/ Α _An ) is also calculated. Since the general-purpose multiplier 8 is used for all the above calculations, efficient calculations can be achieved.

For flames, γ _ABn , γ _ACn , β _ABn (= α
_Bn / α _An ) and β _ACn (= α _Cn / α _An ) are statistically concentrated within a predetermined range. This predetermined range can be represented as a circular range as shown in FIG. 3 by a predetermined number of normalized correlation values γ _ABn , γ _ACn , and the center of gravity and standard deviation σ of the ratio values β _ABn , β _ACn . Furthermore, the above four variables are (β _ABn , γ _A
Bn ), (β _ACn , γ _ACn ), the distance L1 between the two points
When asked for, there is a tendency to concentrate within a certain range (for flames). Therefore, for example, the conditions (flame judgment value) judged as flame are normalized correlation values γ _ABn and γ _ACn of 0.8 to 0.9, and ratio values β _ABn and β _ACn of 0.1 to 0.5 and 2. Distance between points L1
Is set to 0.2 to 0.3, and the determination unit 6 determines that the flame is detected when the above condition is satisfied. In addition, the rate at which the above conditions are satisfied within a certain period is calculated, and if the determination section 6 determines that a flame has been detected when the rate is equal to or higher than a predetermined rate, the accuracy becomes higher. The above method can be efficiently configured by repeatedly using the general-purpose multiplier 8 many times, and moreover, highly accurate fire detection is possible.

The above-described normalized correlation value, moving average ratio value, etc. are calculated by the following equations. 3 at time t _n
An, Bn, and Cn are used as the calculation signals of the two detectors 21 to 23.
Then, moving averages α _An , α _Bn , and α _Cn are given by the following equations (1), (2), and (3), respectively.

[0020]

[Equation 1]

If the standard deviations corresponding to the calculation signals An, Bn, Cn are σ _An , σ _Bn , σ _Cn : Normalized correlation values γ _ABn and γ corresponding to the calculation signals An, Bn, Cn.
_ACn is _expressed by the following equations (6) and (7).

[0022]

[Equation 2]

Using the ratio values β _ABn and β _ACn and the normalized correlation values γ _ABn and γ _ACn sequentially stored in the RAM of the judging unit 6 in this way, respective standard deviations are _expressed by the following equations (8) to (11). And the distance L2 between the centers of gravity, the distance L2 between the centers of gravity, and the distance L1 between the two points are expressed by the following equations (12) to (14).

[0024]

[Equation 3]

Subsequently, the operation of the first embodiment will be itemized and described with reference to the flowchart of FIG. 2 as follows. Thus, using the ratio values β _ABn and β _ACn and the normalized correlation values γ _ABn and γ _ACn sequentially stored in the RAM of the determination unit 6, the respective standard deviations are shown by the following equations (8) to (11), The respective centers of gravity, the distance L2 between the centers of gravity, and the distance L1 between the points are shown by the following equations (12) to (14).

(S1) The amount of infrared rays in each of the wavelength bands A to C is converted into an electric signal from the detectors 21 to 23 and taken in. Each wavelength band of A to C to be detected is, for example, 4.4 μm.
m and 3.9 μm and 5.0 μm. This detection signal is output as a positive output by applying a bias voltage by the amplifiers 31 to 33, and further, a signal component of 4 Hz or higher, which is the flicker frequency band of the flame, is electrically selected and amplified. The amplified signal is converted into an analog signal by removing the frequency components of 1/2 or more of the sampling frequency (128 Hz in this case) by the anti-aliasing filters 41 to 43. At this stage, the anti-aliasing filter is an analog circuit, and some commercial power supply frequency noise (50 Hz, 60 Hz) is mixed in the analog signal. This analog signal is converted into a digital signal by the A / D converter 5. At this time, when the sampling frequency exceeds 120 Hz and is a power of 2, for example, 128 Hz, A /
Perform D conversion. The sampled digital signals are sequentially stored in the RAM of the determination unit 6 and stored for a predetermined period. For example, only digital signals for the past 64 samplings are stored, and those exceeding a predetermined period are sequentially discarded.

(S2) The digital signal stored in the RAM of the judging unit 6 is a predetermined frequency component, for example, an electric flame flicker frequency band, by the digital filter composed of the judging unit 6 and the general-purpose multiplier 8. Only the 4 to 30 Hz signal component is extracted. This extracted signal is the R of the judgment unit 6.
The data is sequentially stored in the AM and stored for a predetermined period. For example, only the extracted signals for the past 64 samplings are stored,
Those that exceed the predetermined period are sequentially discarded. Here, since the sampling frequency is set to be twice as high as 60 Hz or more, the commercial power frequency noise (50 Hz, 60 H) mixed in the analog signal from the anti-aliasing filter is mixed.
z) can be removed by a digital filter without becoming aliasing noise. Since the extracted signal includes the bias voltage, the calculation signal obtained by dividing the extracted signal by the bias voltage is calculated and stored in the RAM of the determination unit 6. For example, the bias voltage is calculated by dividing the average value of the extracted signals for the past N times of sampling by the number N of extracted signals,
A signal for calculation is calculated by dividing the extracted signal. The calculation signal is sequentially stored in the RAM of the determination unit 6 and stored for a predetermined period. For example, only the calculation signals for the past 64 samplings are stored, and those exceeding a predetermined period are sequentially discarded.

(S3) The general-purpose multiplier 8 uses the R
By the calculation signals An, Bn, Cn stored in the AM
Using equations (1) to (3) of [Equation 1], the moving average α_An，
α_Bn, Α _CnTo calculate. This moving average value is R of the judgment unit 6.
The data is sequentially stored in the AM and stored for a predetermined period. example
For example, only the moving average value of the past 64 samplings is stored.
However, those exceeding the predetermined period are sequentially discarded. In addition,
Moving average α_An, Α_Bn, Α_CnAre calculation signals An, Bn,
The absolute value of the signal output is moved on the time axis as the amplitude of Cn.
It is an averaged value.

(S4) The general-purpose multiplier 8 uses the R
Normalized correlation values [gamma] _ABn , [gamma] _ACn between different wavelengths are calculated using the calculation signals An, Bn, Cn stored in AM using the equations (6) to (7) of [Equation 2]. This normalized correlation value is sequentially stored in the RAM of the determination unit 6 and stored for a predetermined period. For example, only the normalized correlation values for the past 64 samplings are stored, and those exceeding a predetermined period are sequentially discarded.

(S5) The general-purpose multiplier 8 uses the R
Using the moving averages α _An , α _Bn , and α _Cn stored in AM, using equations (4) to (5) of [Equation 1], the ratio value β _ABn of moving averages α _Bn and α _Cn to moving average α _An , Β _ACn is calculated. The moving average ratio value is sequentially stored in the RAM of the determination unit 6 and stored for a predetermined period. For example, the past 64
Only the ratio value for the sampling is stored, and those exceeding a predetermined period are sequentially discarded.

(S6) The center of gravity (average) of the moving average αAn within a fixed time (predetermined number) stored in the RAM of the judging unit 6
Is calculated, and the barycenter value is stored in the RAM of the determination unit 6,
If the flame judgment value (above a certain level) stored in the ROM of the judgment unit 6 is reached, the processing after (S7) is performed. If the level is lower than a certain level, the process after (S1) is repeated.

(S7) Ratio values β _ABn , β _ACn of moving averages within a fixed time (predetermined number) stored in the RAM of the judging unit 6
The center of gravity (average) is calculated, the center of gravity value is stored in the RAM of the determination unit 6, and this center of gravity value is stored in the ROM of the determination unit 6 and the flame determination value (within a certain range, for example, 0.1 to 0. If it is within the range of 5, the process proceeds to (S8). If it is out of the certain range, the processes after (S1) are repeated.

(S8) The general-purpose multiplier 8 uses the R
[Equation 3] according to the ratio values β _ABn and β A _{C n} stored in AM
Using equations (8) to (9), the standard deviation σ of the ratio values β _ABn and β _ACn within a fixed time (predetermined number) is calculated, and the standard deviation value is stored in the RAM of the judgment unit 6, ROM of judgment unit 6
Flame judgment value stored in (within a certain range, for example, 0.1 to
If it is within the range of 0.2), the process proceeds to (S9). If it is out of the certain range, the processes after (S1) are repeated.

(S9) The center of gravity (average) of the normalized correlation values γ _ABn , γ _ACn within a fixed time (predetermined number) stored in the RAM of the judgment unit 6 is calculated, and this center of gravity value is stored in the RAM of the judgment unit 6. Is stored and the flame determination value stored in the ROM of the determination unit 6 is within a certain range (for example, within a range of 0.8 to 0.9), the process proceeds to (S10). If it is outside a certain range (S1)
The subsequent processing is repeated.

(S10) The general-purpose multiplier 8 uses the normalized correlation values γ _{AB n} and γ _ACn stored in the RAM of the determination unit 6 to obtain a constant value using the equations (10) to (11) of [ _Equation 3]. The standard deviation σ of the normalized correlation values γ _ABn and γ _ACn within the time (predetermined number) is calculated, the standard deviation value is stored in the RAM of the judgment unit 6, and the flame judgment value stored in the ROM of the judgment unit 6 (S within a certain range, for example, within a range of 0.1 to 0.2) (S
Proceed to 11). If it is out of the certain range, the processes after (S1) are repeated.

(S11) The general-purpose multiplier 8 calculates the ratio values β _ABn , β _ACn and the normalized correlation values γ _ABn , γ _ACn calculated in (S7) and (S9) and stored in the RAM of the judging unit 6. Based on the center of gravity (average), the “distance L2 between the centers of gravity” is calculated using the equation (13) of [Equation 3], and the “distance L2 between the centers of gravity” is stored in the RAM of the determination unit 6, and the determination unit 6 Flame judgment value stored in ROM of (within a certain range, for example 0.2 to 0.3
If it is within the range), the process proceeds to (S12). If it is outside the certain range, the processing after (S1) is repeated. As shown in FIG. 3, the “distance L2 between the centers of gravity” is calculated by assuming that the ratio value β and the normalized correlation value γ are orthogonal coordinates on the two-dimensional plane, respectively, and the 4.4 μm band detection unit A and the 3.9 μm band are used. Coordinate points based on the center of gravity of the ratio value β _ABn with the detection unit B and the center of gravity of the normalized correlation value γ _ABn are plotted on the above-mentioned plane coordinates, and also 4.4 μm
Coordinate points based on the center of gravity of the ratio value β _ACn between the band detection unit A and the 5.0 μm band detection unit C and the center of gravity of the normalized correlation value γ _ACn are plotted on the plane coordinates, and the distance between these coordinate points is calculated. It was done.

(S12) It is determined that a fire is detected by satisfying all the flame judgment values of (S6) to (S11), and the fire signal generation circuit 7 transmits a fire signal to a fire receiver (not shown). In the flow chart shown in FIG. 2, the moving average α _An , the normalized correlation values γ _ABn and γ _ACn , and the ratio values β _ABn and β _ACn are used to detect the fire with high accuracy.
The flame judgment is performed using the center of gravity of and the standard deviation σ. In addition, the flame determination is performed using the “distance L2 between the centers of gravity”. However, in order to detect fire more easily,
Except for (S8) and (S10), the moving average α is obtained in (S6).
Moving average alpha _An place the center of gravity of the An, (S7) the ratio value beta _ABn, ratio value in place of the center of gravity of the _{β ACn β ABn, β ACn,} (S
In 9), the normalized correlation values γ _ABn , γ _ACn are replaced by the normalized correlation values γ _ABn , γ _ACn , and in (S11), the “distance L2 between the centers of gravity” is replaced by the “distance L2 between two points”. Use (S
6), (S7), (S9), it may be determined that a fire is detected by satisfying each flame judgment value in (S11). Further, in that case, a predetermined ratio or more within a certain time, for example, 8 or more of the past 10 data stored in the RAM, are (S6), (S7),
A fire may be detected by satisfying each flame judgment value in (S9) and (S11). Although the present embodiment has been described as a three-wavelength type flame detection device, it can also be applied to a two-wavelength type flame detection device. For example, when the detection wavelength bands of the detection unit are set to 4.4 μm and 3.9 μm and the corresponding flame judgment values in (S6) to (S10) excluding (S11) in the flowchart of FIG. Can be detected. Further, in the present embodiment, the detection unit has been described as capturing infrared light as light having a wavelength peculiar to the flame, but in addition to this, it is also possible to configure a flame detection device by detecting ultraviolet light or visible light. .

Second Embodiment Commercial power frequency noise may be mixed in the detection signal detected by the light detecting means. Generally, when the sampling frequency is AHz when the detection signal is sampled, aliasing noise occurs when a frequency component of A / 2 Hz or more is input according to the sampling theorem. Therefore, in order to correctly sample the commercial power supply frequency of 60 Hz without making it a folding noise, the sampling theorem
It must be sampled at a sampling frequency above 0 Hz. In the case of the present invention, as shown in the first embodiment, commercial power supply frequency noise (50 Hz, 60 H)
The sampling frequency is set to a frequency exceeding 120 Hz (twice of 60 Hz) so that z) does not become aliasing noise, and the sampling frequency is set to a power of 2 so that the determination unit 6 can easily perform the calculation. For example, 128H
Set to the sampling frequency of z. As a result, the commercial power supply frequency noise does not become aliasing noise in the vicinity of the flame fluctuation frequency used for flame determination, so this noise can be removed by a digital filter, and flame detection with high accuracy is possible. The A / D converter 5 is an example of sampling means for sampling the detection signal detected by the light detecting means at a sampling frequency exceeding 120 Hz.

[0039]

Industrial Applicability The present invention provides a flame detecting device for detecting a flame generated during a fire by detecting light having a wavelength peculiar to the flame.
The degree of similarity between the waveforms of the detection signals detected by the plurality of light detection means and the plurality of light detection means for detecting a plurality of lights having different wavelengths peculiar to the flame and outputting the detection signals corresponding to the respective wavelengths is shown. The calculating means for calculating the normalized correlation value and the moving average ratio value indicating the amplitude ratio between the waveforms of the respective detection signals, and the normalized correlation value and the moving average ratio value calculated by the calculating means are respectively predetermined values. The flame detection device is provided with a flame generation determination means for determining that a flame has occurred when within the range.

Further, in the above, the flame occurrence determining means is
A flame detection device is configured to determine that the flame is present when the normalized correlation value and the moving average ratio value are within a predetermined range at a predetermined rate or more within a predetermined time period.

Further, in the above description, the plurality of photodetection means are three or more, form a plane coordinate centered on the normalized correlation value and the ratio value of the moving average, and normalize between the waveforms of the respective detection signals. The correlation value and the ratio value of the moving average, and each coordinate point according to, are plotted on the plane coordinates, the calculating means calculates the distance between the coordinate points, and the flame occurrence determining means determines the distance within a certain range. A flame detection device is sometimes configured to judge that a flame has occurred. Further, in the above, the flame occurrence determining means is a flame when the ratio value of the normalized correlation value and the moving average and the distance are within a predetermined range and within a predetermined range at a predetermined rate or more within a predetermined time period. A flame detector for judging was constructed.
Further, in the above, the flame occurrence determination means constitutes a flame detection device that performs flame determination using the center of gravity of the normalized correlation value and the ratio value of the moving average and the standard deviation.

Further, in the above, the plurality of light detecting means constitutes the flame detecting device including the first light detecting means for detecting the infrared ray having the wavelength of 4.4 μm band peculiar to the flame. Also,
The present invention is a flame detection device for detecting light having a wavelength peculiar to a flame to detect a flame generated at the time of a fire, and light detection means for detecting light having a wavelength peculiar to the flame and outputting a detection signal, The flame detection device was provided with a sampling means for sampling the detection signal detected by the detection means at a sampling frequency exceeding 120 Hz. Further, in the above, the flame detection device is configured so that the sampling frequency is a power of 2.

Therefore, according to the present invention, it is possible to provide a flame detecting device capable of eliminating a false alarm source and removing noise to detect a flame with high accuracy.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the first embodiment.

FIG. 3 is an explanatory diagram showing a plane coordinate system of a moving average ratio value and a normalized correlation value.

FIG. 4 is a waveform diagram showing an emission spectrum distribution of infrared rays emitted from various objects.

[Explanation of symbols]

1 Flame detector 5 A / D converter 6 Judgment part (flame occurrence judgment means) 7 Fire signal generation circuit 8 General-purpose multiplier (calculation means) 21 Detector A (light detecting means) 22 Detection unit B (light detection means) 23 Detector C (light detecting means) 31, 32, 33 amplifier 41, 42, 43 Antialiasing filter

Claims (8)

[Claims] 1. A flame detection device for detecting a flame generated during a fire by detecting light having a wavelength peculiar to a flame, wherein a plurality of detection signals having different wavelengths peculiar to the flame are detected and detection signals corresponding to the respective wavelengths are detected. And a normalized correlation value between the waveforms of the detection signals detected by the plurality of photodetectors and a ratio value of a moving average between the waveforms of the detection signals. And a flame occurrence determining means for determining that a flame has occurred when the normalized correlation value and the moving average ratio value calculated by the calculating means are within a predetermined range. Flame detector. 2. The flame occurrence determination means determines that the flame is present when the normalized correlation value and the moving average ratio value are within a predetermined range for a predetermined ratio or more within a predetermined time period. The flame detection device according to claim 1, wherein 3. The plurality of photodetection means are three or more, and form plane coordinates with the normalized correlation value and the ratio value of the moving average as axes, and between the waveforms of the respective detection signals. The normalized correlation value and the ratio value of the moving average, and each coordinate point is plotted on the plane coordinates, the calculating means calculates the distance between the coordinate points, the flame occurrence determining means, The flame detection device according to claim 1, wherein it is determined that a flame has occurred when the distance is within a certain range. 4. The flame occurrence determination means is configured such that the normalized correlation value and the ratio value of the moving average and the distance are within a predetermined range and within the predetermined range within a predetermined time. The flame detection device according to claim 3, wherein the flame detection device determines that the flame is present. 5. The flame occurrence determination means makes a flame determination using the center of gravity and the standard deviation of the normalized correlation value, the ratio value of the moving average, and the standard deviation. The flame detection device described in 1. 6. The plurality of light detecting means includes a first infrared detecting means for detecting infrared rays having a wavelength of 4.4 μm band peculiar to the flame. The flame detection device described in 1. 7. A flame detection device for detecting light having a wavelength peculiar to a flame to detect a flame generated at the time of a fire, and light detection means for detecting light having a wavelength peculiar to the flame and outputting a detection signal. A flame detecting device comprising: a sampling unit configured to sample a detection signal detected by the light detecting unit at a sampling frequency exceeding 120 Hz. 8. The flame detection device according to claim 7, wherein the sampling frequency is a power of two.