CN2516967Y - Three-wavelength full scattering laser smoke identification fire detection device - Google Patents

Abstract

The utility model relates to a smoke and fire detecting device of three-wavelength scattering laser inducting, the utility model is characterized in that an aspirator pump inhales the smoke or aerosol into a smoke detection room, three lasers of different wavelength are thrown out by the laser, then rips into the smoke detection room and is decayed by smoke, exits to a photoelectric receiving tube to be conversed to a plurality of electrical signals, by amplification, after the A / D conversion by a multi-channel data acquisition card, the decaying data of the three lasers are sent to a miniature computer to be analysed and treated, the extinction coefficient ratio of the three wavelengths is obtained, which is the input of a smoke identification neural network, the smoke identification neural network analyses and judges the input data, then controls an alarm controller to drive a fire linkage alarm device to alarm when ''fire smoke'' is confirmed, and not alarm when ''non-aerosol fire'' is confirmed. The utility model has the advantages that the fire detection device can accurately and reliably identify fire smoke and non-fire aerosol, has the intelligent identification, uniform sensitivity, high reliability, environmental anti-interference capability, and other advantages.

Description

The three wave length total scattering laser smoke sensing fire hazard sniffer

Technical field:

The utility model belongs to the fire detection technology field, particularly relates to laser smoke sensing calamity sniffer.

Background technology:

Existing fire detector thoughts cigarette formula, temp sensing type, photosensitizing type etc., smoke fire detector can be divided into photoelectric smoke detector and ion smoke detector again, and smoke fire detector is to use one of the most general detector at present in the world.According to the 3rd inferior Australia Fire Science nd Annual Meeting collection (Proceedings of the Third Asia-Oceania Symposium onFire Science and Technology, June 10-12,1998, pp.319-331) introduce, existing photoelectric smoke detector be adopt smoke particle to incident light produce scattering, principle of absorption is surveyed the concentration of smog and then is realized fire alarm, has characteristics such as "dead" pollution, adaptability be stronger, with low cost.But, there is following defective aspect detection sensitivity and the reliability owing to be subjected to the limitation of this detection mechanism:

1. because the smoke particle of different-grain diameter produces the relative sensitivity difference of scattering to incident light, when the smoke particle particle diameter less than 0.4 μ m, do not produce scattered light, relative sensitivity is relatively poor;

2. the smog that produces of fire can be divided into invisible smog, translucent white, grey smog, lighttight black smog substantially, and for transparent invisible smog fully, because of can not scattered light, optical detector of fire smoke can't be surveyed; And black smog has strong absorption, blocks effect light, can only produce faint scattered light, also almost survey less than;

3. existing optical pickup apparatus is subject to the influence of environmental interference, when having dust, water vapour, oil smoke, wind and electromagnetism etc. to disturb, causes wrong report easily, and reliability is lower;

4. there is being the special place (as temple etc.) of smog then can't use for a long time.

Summary of the invention:

The utility model provide a kind of can be to the higher laser smoke sensing calamity sniffer of the reliability that fire hazard aerosol fog and non-fire gasoloid are discerned to overcome the above-mentioned defective of prior art.

This three wave length total scattering laser smoke sensing fire hazard sniffer, the constant current Laser Driven power supply 1, Laser emission assembly 2, smoke detection chamber 4, photoelectric sensor assembly 5, the alarm controller 8 that it is characterized in that comprising multi-channel data acquisition card 6, microcomputer 7 and be assembled in the cabinet constitute, one end of smoke detection chamber 4 has smog air intake opening 3, the other end has the smog gas outlet and aspiration pump 9 joins, and the outlet of aspiration pump 9 is connected to outside the cabinet by exhausr port 10; Described Laser emission assembly 2 adopts three laser instruments to be installed on the same surface level of smoke detection chamber 4 front ends, makes three beams of laser respectively behind the scioptics collimation, forms the laser beam that three beams is parallel to each other; Described photoelectric sensor assembly 5 adopts three photoelectric tubes to be installed in the light path exit of smoke detection chamber 4 other ends, receives the light intensity signal from three laser instruments respectively; Multi-channel data acquisition card 6 is inserted on the mainboard of microcomputer 7, and the signal output part of photoelectric sensor assembly 5 links to each other with the input port of multi-channel data acquisition card 6, and the output signal of microcomputer 7 is connected to the input end of alarm controller 8.

In order to improve the detection sensitivity of device, one group of right-angle reflecting prism 11 can be set at the two ends of 4 inside, smoke detection chamber, two right-angle reflecting prism 11 are at the vertical direction certain distance that staggers, after the laser that the three beams that Laser emission assembly 2 is sent is parallel to each other incides the top of concave surface of right-angle reflecting prism 11 of right-hand member earlier, reflex to the bottom of its concave surface, reflex to the bottom of concave surface of the right-angle reflecting prism 11 of left end again, after reflex to the top of its concave surface again, three beams of laser between the right-angle reflecting prism 11 at two ends through repeatedly the reflection after, finally penetrate from the bottom of right-angle reflecting prism 11 concave surfaces of left end, the photoelectric sensor assembly 5 that the three beams parallel laser of ejaculation is positioned at 4 rear ends, smoke detection chamber receives; The direct projection light path of long distance also can be adopted in smoke detection chamber 4, for example adopts length to be not less than 1 meter direct projection light path.

During work, aspiration pump sucks smog to be measured or gasoloid in the smoke detection chamber, by constant current Laser Driven power drives Laser emission assembly, sending the three beams different wavelength of laser incides in the smoke detection chamber simultaneously, after the decay of the smog in the smoke detection chamber, shine on the photoelectric sensor assembly, be converted to electric signal by photoelectric receiving tube, after signal amplifies, stick into capable A/D conversion by multi-channel data acquisition, digital quantity after the conversion is sent into microcomputer, microcomputer carries out data analysis to the attenuation data of the three beams of laser that collects, handle, obtain the extinction coefficient ratio of three pairs of wavelength, and with this input as smog identification neural network, smog identification neural network is carried out analysis and judgement to the input data, determine aerocolloidal type in the smoke detection chamber, identify " fire hazard aerosol fog " or " non-fire gasoloid ", control alarm controller 8 drives fire interlink warning equipment and sends warning when confirming as " fire hazard aerosol fog ", does not then report to the police when confirming as " non-fire gasoloid " time.

The utility model is a kind of based on the laser smoke sensing calamity sniffer that gasoloid is carried out Classification and Identification.Owing to adopt the little LASER Light Source and the constant current Laser Driven power supply of high strength, the angle of divergence of three beams different wave length, make that extracting the extinction coefficient ratio that characterizes the gasoloid essential characteristic becomes possibility, thereby can carry out Classification and Identification to fire hazard aerosol fog and non-fire gasoloid, improve the ability of the anti-environmental interference of optical detector of fire smoke, had very high reliability; Adopting initiatively bleeds analyzes the sensitivity that can improve detection, and time of fire alarming is shifted to an earlier date; Adopt smog identification neural network to carry out Classification and Identification to fire hazard aerosol fog and non-fire gasoloid exactly, avoid the interference of environmental factor, have very high reliability; The utility model has overcome the defective that existing photoelectric smoke detector exists aspect detection sensitivity and reliability, can discern accurately and reliably the fire hazard aerosol fog and the non-fire gasoloid of different-grain diameter and color, have advantages such as Intelligent Recognition, sensitivity unanimity, reliability height, anti-environmental interference ability be strong, be specially adapted to the fire detecting and alarm in particular surroundings place.

According to the Lambert-Beer law, smoke particle is done the time spent mutually with laser, its incident intensity I ₀And the pass between the emergent light intensity I is:

I=I ₀I is for seeing through the light intensity of smog in exp (μ L) (1) formula; I ₀Be incident intensity; L is average ray length of stroke; The scale-up factor that μ is and light intensity is irrelevant is called the extinction coefficient or the attenuation coefficient of medium, and it is the important parameter that characterizes smog or non-fire aerosol extinction; The Sauter mean grain size d of extinction coefficient μ and medium ₃₂, population concentration N and medium the flatting efficiency factor Q _Ex(λ, m, d ₃₂) relevant, can be expressed as: $\mathrm{\μ}=\frac{\mathrm{\π}}{4}{d}_{32}^2{\mathrm{NQ}}_{\mathrm{ex}}(\mathrm{\λ},m,d_{32}).....\left(2\right)$ Flatting efficiency factor Q in the formula _Ex(λ, m, d ₃₂) be a characteristic.

In the expression formula (2) of extinction coefficient, extinction coefficient μ was both relevant with the mean grain size and the refractive index of particulate, and was relevant with aerocolloidal concentration again.

In three wave length total scattering laser smoke sensing fire hazard detecting method, the extinction coefficient ratio that defines any a pair of wavelength is R _{λ 1/ λ 2}, can obtain by formula (1) and formula (2) $R_{\mathrm{\λ}1/\mathrm{\λ}2}=\frac{{\mathrm{\μ}}_{{\mathrm{\λ}}_1}}{{\mathrm{\μ}}_{{\mathrm{\λ}}_2}}=\frac{Q_{\mathrm{ex}}(\mathrm{\λ}1,m,d_{32})}{Q_{\mathrm{ex}}(\mathrm{\λ}2,m,d_{32})}=\mathrm{Ln}{(I_0/I)}_{\mathrm{\λ}1}/\mathrm{Ln}{(I_0/I)}_{\mathrm{\λ}2}....\left(3\right)$ By formula (3) extinction coefficient ratio R as can be known _{λ 1/ λ 2}Only relevant with the mean grain size and the refractive index of particulate, for specific particulate, the extinction coefficient ratio R of two known wavelength _{λ 1/ λ 2}Be certain value, this value has reflected the relative extent of particulate to two wavelength lasers decay, and is with the essential characteristic (mean grain size, refractive index) of particulate relevant, and can be by the mensuration of incident intensity and output intensity, carries out simple computation and obtains.

Different fire hazard aerosol fogs and non-fire gasoloid are different to the decay of different wave length laser, as stacte smog, diesel oil naked light smog, polyurethane foam plastics naked light smog and dust etc. the extinction coefficient ratio of different wave length laser there is tangible difference, can carry out Classification and Identification to different fire hazard aerosol fogs and non-fire gasoloid in view of the above, thereby reach fire detecting and alarm accurately and reliably.

The utility model adopts Artificial Neural Network, the extinction coefficient ratio of three pairs of wavelength that data processing is calculated is as the input of smog identification BP neural network, neural network is by after carrying out learning training to sample set, to the judgement of classifying of the smog aerosol in the smoke detection chamber, identify " fire hazard aerosol fog " or " non-fire gasoloid ", thereby reach fire detecting and alarm accurately and reliably.

The utility model three wave length total scattering laser smoke sensing fire hazard detecting method and device also can change into and adopt four wavelength, five wavelength ... multi-wavelength total scattering laser smoke sensing fire hazard detecting method and device, as long as its three laser instruments and three photoelectric tubes of photoelectric sensor assembly thereof are changed into four, five respectively ... laser instrument and four, five ... photoelectric tube gets final product.

Of the present utility model applying promoting the development of China's fire detecting and alarm technology realized the leap of fire detecting and alarm from the concentration type to the smog identification type.

Description of drawings:

Accompanying drawing 1 is system's principle of compositionality synoptic diagram of the utility model three wave length total scattering laser smoke sensing fire hazard sniffer.

Accompanying drawing 2 is constant current Laser Driven power supply electrical schematic diagram.

Accompanying drawing 3 is smoke detection chamber light channel structure synoptic diagram.

Accompanying drawing 4 is the biasing and amplification electrical schematic diagram of photoelectric tube.

Accompanying drawing 5 is smog identification BP neural network structure synoptic diagram.

Accompanying drawing 6 is data processing of the present utility model, judgement and control program process flow diagram.

Embodiment:

Embodiment 1

Present embodiment three wave length total scattering laser smoke sensing fire hazard sniffer, by comprising constant current Laser Driven power supply 1, Laser emission assembly 2, smoke detection chamber 4, photoelectric sensor assembly 5, alarm controller 8, the aspiration pump 9 that is assembled in the same cabinet, and composition such as multi-channel data acquisition card 6, microcomputer 7; One end of smoke detection chamber 4 has smog air intake opening 3, and the other end has the smog gas outlet and aspiration pump 9 joins, and the outlet of aspiration pump 9 is connected to outside the cabinet by exhausr port 10; Laser emission assembly 2 is installed in the front end of smoke detection chamber 4, and photoelectric sensor assembly 5 is installed in the emergent light exit of 4 rear ends, smoke detection chamber, and the outlet of the aspiration pump 9 that joins with the outlet of the smog of smoke detection chamber 4 is connected to outside the cabinet; Described Laser emission assembly 2 adopts semiconductor laser as light source, three semiconductor laser installing are on the same surface level of smoke detection chamber 4 front ends, after making three beams of laser difference scioptics collimation, form the laser beam that three beams is parallel to each other, pass smoke detection chamber 4, the photoelectric sensor assembly 5 that is positioned at 4 rear ends, smoke detection chamber receives; Described photoelectric sensor assembly 5 selects for use photodiode as receiving device, and three photodiodes are installed in the light path exit of smoke detection chamber 4 other ends, receive the light intensity signal from three laser instruments respectively; Multi-channel data acquisition card 6 is inserted on the mainboard of microcomputer 7, and the signal of photoelectric sensor assembly 5 is linked to each other by the input port of lead with multi-channel data acquisition card 6, and the output signal of microcomputer 7 is connected to the input end of alarm controller 8.

In the present embodiment, the driving power of semiconductor laser adopts closed loop current to regulate control, constant current output, and by 2 operational amplifier U1A, U1B and R2, R3, R4, R5, R6, C3 constitutes a Current Negative Three-Point Capacitance amplifying circuit, so that the base current I of triode T1 _BConstant current, and constant current value I _BThe datum decision that provides by potentiometer R7 of size; Triode T1 works in the constant current zone, and the resistance of the load current of triode T1 and load size is irrelevant at this moment, only depends on I _BSetting value; C1 and L1 constitute holding circuit, and the fluctuation of electric current is suppressed.

For improving the sensitivity of photodetection, present embodiment adopts a pair of right-angle reflecting prism 11 to be installed in the two ends of 4 inside, smoke detection chamber respectively, the right-angle reflecting prism 11 at two ends is at the vertical direction certain distance that staggers, after the laser that the three beams that Laser emission assembly 2 is sent is parallel to each other incides the top of concave surface of right-angle reflecting prism 11 of right-hand member, reflex to the bottom of its concave surface, reflex to the bottom of concave surface of the right-angle reflecting prism 11 of left end again, after reflex to the top of its concave surface again, three beams of laser between the right-angle reflecting prism 11 at two ends through repeatedly the reflection after, finally penetrate from the bottom of right-angle reflecting prism 11 concave surfaces of left end, the photoelectric sensor assembly 5 that the three beams parallel laser of ejaculation is positioned at 4 rear ends, smoke detection chamber receives.

Photodiode is worked under applied reverse bias voltage, and its output is current signal, and the output current of photodiode is directly proportional with the incident light intensity, and photocurrent flows through pull-up resistor R _L, on pull-up resistor, obtain the output voltage signal that is directly proportional with incident intensity; Output signal amplification circuit adopts secondary low noise DC voltage amplifier, shared two-way JFET input low noise high speed operation amplifier, its first order is by ICa and R1, R2 constitutes voltage amplifier, and enlargement factor is about 10, between first order amplifier and second level amplifier, by R3, R4, R5, R6, W1 and ICb constitute a zeroing circuit, to eliminate the zero point drift of photodetector, scalable potentiometer W1 makes detector be output as zero when unglazed the photograph, the enlargement factor of second level amplifier is adjustable, its enlargement factor is 1+ (R8+W2)/R7, selects R8, R7, the value of W2, the total magnification of amplifying circuit is controlled at about 10～20, regulator potentiometer W2, the size of scalable detector output voltage.

The utility model embodiment has adopted typical three layers of backpropagation (BP) neural network to discern neural network model as fire hazard aerosol fog, and the input layer of network is 3 neurons, is respectively I ₁, I ₂, I ₃, the extinction coefficient ratio (R of corresponding three pairs of wavelength _{1.55 μ m/1.06 μ m}, R _{1.55 μ m/0.67 μ m}, R _{1.06 μ m/0.67 μ m}), input value all normalizes to [0,1]; Output layer is 2 neurons, is respectively O ₁And O ₂, corresponding " fire hazard aerosol fog " and " non-fire gasoloid ", the output valve scope also is [0,1]; Middle hidden layer is 5 neuron (M ₁, M ₂..., M ₅), 3 * 5 lines are arranged between input layer and hidden layer, its weights are w1 _Ij, 2 * 5 lines are arranged between hidden layer and output layer, its weights are w2 _Jk Weight w 1 between input layer and the hidden layer _IjAnd between hidden layer and the output layer weight w 2 _Jk, behind network training, determine by sample set.

The three beams of laser that adopts wavelength to be respectively 1.55 μ m, 1.06 μ m and 670nm in the present embodiment is surveyed the fire hazard aerosol fog in the smoke detection chamber 4 simultaneously, photodiode is selected silicon PIN photoelectric diode (400～1100nm) and Long Wavelength InGaAs PIN photodiode (0.9～1.7 μ m) for use, smog identification neural network adopts three layers of error anti-pass feedforward network (BP network), input layer is 3 neurons, hidden layer is 5 neurons, output layer is 2 neurons, the extinction coefficient ratio (R of 3 corresponding three pairs of wavelength of input _{1.55 μ m/1.05 μ m}, R _{1.55 μ m/670nm}, R _{1.06 μ m/670nm}), 2 outputs corresponding " fire hazard aerosol fog " and " non-fire gasoloid ", the activation function of hidden layer and output layer adopts logarithm Sigmoid function, the weight factor of input layer to the weight factor of hidden layer and hidden layer to output layer determined behind the learning training by sample set is carried out, after weight factor was determined, smog identification neural network was promptly determined.

During work, aspiration pump 9 sucks smog to be measured or gasoloid in the smoke detection chamber 4 by air intake opening 3; By constant current Laser Driven power supply 1 driving laser emitting module 2, launching the three beams wavelength simultaneously is 1.55 μ m, 1.06 μ m, 0.67 the laser of μ m is surveyed the smog aerosol that sucks in the smoke detection chamber 3, Laser emission assembly 2 adopts the constant-current driving power supply power supply, to guarantee that Laser emission assembly 2 sends the laser of stablizing light intensity, three beams of laser incides in the smoke detection chamber 4 simultaneously, and between a pair of right-angle reflecting prism 11 through repeatedly the reflection, after three beams of laser is inhaled into smog decay in the smoke detection chamber 4, shine on the photoelectric sensor assembly 5, be converted to electric signal by the photoelectric receiving tube in the photoelectric sensor assembly 5 again, and after signal amplifies, carry out the A/D conversion by multi-channel data acquisition card 6, digital quantity after the conversion is sent into microcomputer 7 and is carried out data analysis, handle, the attenuation data of 7 pairs of three beams of laser that collect of microcomputer carries out data processing, obtain the extinction coefficient ratio of three pairs of wavelength, and with this input as smog identification neural network, smog identification neural network is carried out analysis and judgement to the input data, determine aerocolloidal type in the smoke detection chamber 4, identify " fire hazard aerosol fog " or " non-fire gasoloid ", control alarm controller 8 drives fire interlink warning equipment and sends warning when confirming as " fire hazard aerosol fog ", does not then report to the police when confirming as " non-fire gasoloid " time.

Claims (3)

1, a kind of three wave length total scattering laser smoke sensing fire hazard sniffer, the constant current Laser Driven power supply (1), Laser emission assembly (2), smoke detection chamber (4), photoelectric sensor assembly (5), the alarm controller (8) that it is characterized in that comprising multi-channel data acquisition card (6), microcomputer (7) and be assembled in the cabinet constitute, one end of smoke detection chamber (4) has smog air intake opening (3), the other end has the smog gas outlet and aspiration pump (9) joins, and the outlet of aspiration pump (9) is connected to outside the cabinet by exhausr port (10); Described Laser emission assembly (2) adopts three laser instruments to be installed on the same surface level of smoke detection chamber (4) front end, makes three beams of laser respectively behind the scioptics collimation, forms the laser beam that three beams is parallel to each other; Described photoelectric sensor assembly (5) adopts three photoelectric tubes to be installed in the light path exit of smoke detection chamber (4) other end, receives the light intensity signal from three laser instruments respectively; Multi-channel data acquisition card (6) is inserted on the mainboard of microcomputer (7), the signal output part of photoelectric sensor assembly (5) links to each other with the input port of multi-channel data acquisition card (6), and the output signal of microcomputer (7) is connected to the input end of alarm controller (8).

2, three wave length total scattering laser smoke sensing fire hazard sniffer according to claim 1, it is characterized in that the two ends inner in described smoke detection chamber (4) are provided with one group of right-angle reflecting prism (11), two right-angle reflecting prism (11) are at the vertical direction certain distance that staggers, after the laser that the three beams that Laser emission assembly (2) is sent is parallel to each other incides the top of concave surface of right-angle reflecting prism (11) of right-hand member earlier, reflex to the bottom of its concave surface, reflex to the bottom of concave surface of the right-angle reflecting prism (11) of left end again, after reflex to the top of its concave surface again, three beams of laser between the right-angle reflecting prism (11) at two ends through repeatedly the reflection after, finally penetrate from the bottom of right-angle reflecting prism (11) concave surface of left end, the photoelectric sensor assembly (5) that is positioned at rear end, smoke detection chamber (4) receives.

3, three wave length total scattering laser smoke sensing fire hazard sniffer according to claim 1 is characterized in that described smoke detection chamber (4) adopts length to be not less than 1 meter direct projection light path.

Images