AU734148B2 - Fire alarm - Google Patents

Description

Our Ref: 691432 P/00/011 Regulation 3:2

AUSTRALIA

Patents Act 1990

ORIGINAL

COMPLETE SPECIFICATION STANDARD PATENT 4.

S

S

*SSS

S

Applicant(s): Hochiki Corporation 10-43, Kamiohsaki 2-chome Shinagawa-ku Tokyo

JAPAN

Hiromitsu Ishii 5-2-801, Takasu 3-chome Chiba-shi Chiba-Ken

JAPAN

DAVIES COLLISON CAVE Patent Trade Mark Attorneys Level 10, 10 Barrack Street SYDNEY NSW 2000 Address for Service: Invention Title: Fire alarm The following statement is a full description of this invention, including the best method of performing it known to me:- 5020 10606 ON YH/YI] TT:VT allM TO, i'0/TT FIRE ALARM Technical Field This invention relates to a fire alarm system and, more particularly, to a fire alarm system in which discriminationi of the occurrece of a fire is made by calculating pararneters or data such as heat release rate, smoke production rate or gas production rate, at a fire source, by introducing a fire siinnlating mathemnatical model.

Backajound Art In the conventional fire alarm system, discrimination of fire for alarm initiation is basically made on the basis of comparison of secondary paramieters accompanying a fire, such as temperature, smoke concention or CO gas concentration, detected by sensors, or the time rate of change thereof, with preset individual threshold values, In the simplest system the occurrence of a fie is discriminated and an alarm issued when the values detected by the sensors exceed a predetermined threshold values. There is also known a system in which a fire is assumed to have occurred when the rate of temporal change obtained by difenatg the value detected by the sensor exceeds a preset value, or when future changes anticipating a fire are predicted by first or second order function approximations from past changes in the values detected by the sensors.

Thus, the conventional fire alarm system has been based on prinociple that the heat, smoke or gas, produced primarily by the fire, are detected secondarily by the sensors, and the fire is presumed to have occurred directly from these secondary parametersgoo [n gaol, 1 Yivi.sIv a 1 NOSIT]OO M3AUC 966~ 9 6 E zVV TO.91/ 9T:CT TO, tO/TT 10606 ON XH/XL] TT:VT UIHA TO, bo/TT -2- Hiowever, the process of a fire generationis variegated and the combustionL Products of the environmental conditions take part in the fire propagation process in a complex manner. Thus the conditions of the secondarily produced temperature, smoke concentration or gas concentration, are also cbanged variably by cbanges in environmental conditions, such that many difficulties are experienced when the occurrence of a fire is to be discriminated accurately and promptly from these variegated sensor data or parameters.

Disclosure of the Invention It is a primiary object of this invention to provide a novel fire alarm system in which the occurrence of a fire can be discriminated by calculating prinmy fire source parameters, such as heat release rate, smoke production rate or gas production rate, at fire source, from the sensor signal, such as the temperature, smoke concentration or CO gas concentration secondarily produced by the fire.

It is another objection of this invention to provide the above mentioned fire alarm system in which the detection reliability is further improved on the basis of the correlation of a plurality of kinds of primiary fire source parameters.- For accomplishing the above objects, this invention in its broadest aspect resides in a fire alarm system comprising: 900 [2) VIIVHISflV d'I MOSMOD SaIAVQ 966Z Me z TQqQ LT:CT TO VO/TT ~oo L~j VI'1V&LSfJN d1 N051T103 S~IAYc[ ~eo~ g ±i:cT Ta. ~O/TT sensor means provid(ed in a fire monitor area for (detecting a physical pheynomenai accompanying i1 fire such as the variation of the temperature, smoke concentration or CO gas concentration; fire-source parameter calculating means for calculating a primary fire souce parameter such as the heat release rate, the smoke production rate or the gas production rate at a fire source on the basis of detection data by said sensor means and a preset formula for an arithmetic operation; and fire judgement means for discriminating a fire on the basis of the rate of change of said fire source parameter calculated by said fire source parameter culculating means.

With the fire alarm system according to the present invention, an arithmetic logical program for a fire simulating mathematical model analyzing the fire conditions in a room °o .is applied as the above mentioned formula in the fire source parameter calculating means, and the primary fire source parameters such as the heat relese rate, smoke production rate or the gas production rate, may be calculated from secondary parameters such as the temperature, the smoke concentration or the CO gas concentration, detected by the sensor means, by reverse calculation by the above formula, and an accurate fire discrimination may then be made from the rate of change of these primary parameters of the fire source to initiate an alarm.

These primary parameters, that is the heat release rate, smoke production rate or :he gas production rate of the fire source, are determined by nature unequivically, without being 3 affected by combustion products or environmental conditions, and help improve the calculation accuracy of' fire parameters to improve in turn the reliability of fire discrimination significantly.

According to the present invention, the heat release rate, smoke production rate or the gas production rate of the fire source itself may be calculated from the secondary phenomina accompanying the fire, such as the temperature, smoke concentration or CO gas concentration sensed by the sensors, by the reverse calculation of the fire simulating mathematical model analyzing the fire condition in the room, and the fire can be discriminated from the rate of changes of the primary fire source parameters. Thus the risk of mistaken judgiment of taking a false alarm source for a fire may be minimized to improve the reliability of fire discrimination significantly.

The above and other features and advantages of this invention will become more apparent from the following description in conjunction with the accompanying drawings.

Brief Description of the Drawings Fig. 1 shows an arrangement of a fire alarm system according to an embodiment of the present invention in block diagram.

Fig. 2 shows a two-layered zone model employed in a fire discrimination algorithm in the present embodiment.

Fig. 3 is a flow diagram of a fire detection algorithm in the present embodiment.

Figs. la, 4b and 4c show sensor response during burning a wooden chair and temporal changes in the heat release rate A 0, smoke production rate A Cs and gas production rate A G as evaliat-ed in accordance with the present embodiment.

Figs. 5a, 5b and 5c show sensor response during cooking and temporal changes in the heat release rate A Q, smoke production rate A Cs and gas production rate A G as evaliated in accordance with the present embodiment.

Figs. 6a and 6b show sensor response during burning of a wooden chair in plural rooms of diffrent sizes and temporal changes of the heat release rate A Q as evaliated in accordance with the present embodiment.

Figs. 7a and 7b show temporal changes of the correlation value R, obtained from the weighting of the correlation factor R as found from the heat release rate and the smoke production rate obtained in the burning of the wooden chair as shown in Figs..4a, 4b and 4c, and temporal changes of its derivative dR /dt.

Figs. 8a and 8b show temporal changes of the correlation value R 0 obtained from the weighting of the correlation factor R as found from the heat release rate and the smoke production rate obtained in the cooking as shown in Figs. Sa, 5b and and temporal changes of its derivative dR,/dt.

Best Mode for Carrying Out the Invention In Fig. 1, a fire alarm system according to an embodiment of the present invention includes a plurality of sensors provided on a ceiling or the like of a room to be monitored, r namely a temperature sensor 10, a smoke concentration sensor 12 and a CO gas concentrat:jon sensor 14 Cor detectin the temperature 0, smoke concentration Cs, and the CO gas concentration G in an analog fashion and outputting detection signals proportional to the detected values. The fire alarm system also includes a sampling circuit 16 for receiving said detection signals from the sensors 10, 12 and 14. The detection signals are sampled at fixed time intervals in the samping circuit 16 and converted into output digital signals by an A/D convertor within the circuit.

In the present embodiment, each one of the sensors 10, 12 and 14 is provided in each fire monitor area. However, if necessary, respective two or more sensors of the same type may be provided in each fire monitor area. As to the signal transmission system between the sensors and the sampling O* circuit 16, although a direct wiring system which directly transmits the detected analog signal over a signal line is applied in this embodiment, any suitable signal teansmission system, such as a polling system consisting in polling the sensor side from the sampling circuit 16 and retransmitting the detection signal, may also be applied in place of the direct wiring system.

The present system also includes, in downstage of the sampling circuit 16, a fire souce parameter calculating unit 18 and an initial value setting unit 20 for affording various initial values for executing the mathematical model for fire simulation to the fire source parameter calculating unit 18.

An arithmetic logical program of a mathematical model for a f i re simulation is installed in advance in the re source parameter calculating unit 18. By back or reverse calculation of the mathematical model, the heat release rate, the smoke production rate and the qas production rate are calculated from the detected data originated from the sensors, i,e, the temperature f the smoke concentration Cs and the CO gas concentration G.

Various initial values for executing the fire simulating mathematical model are afforded from the initial value setting unit 20 to the calculating unit 18, which is responsive to the initial value setting pursuant to the conditions of the fire S monitor area, in which the aforementioned sensors are provided, to calculate the primary fire source parameters from

S

the sensor detection data.

The fire alarm system also includes a fire judgement unit 22 for receiving the source parameters calculated by the calculating unit 18, that is the heat release rate, the smoke production rate and the gas production rate, and an alarm display unit:24 responsive to alarm signals from the fire judgement unit 22 to initiate an alarm in optional forms, such as acoustic and or visual form.

The fire judgement unit 22 discriminates the occurrence of fire based on the rate of change of the fire source parameters exceeding a preset initiation level, or by executing fire predictive arithmetic operations, when the initiation level is exceeded, in accordance with a first order or second order function, using the so far acquired fire source parameters. When the results of fire judgements based 7 on the Lire source paramete rs are obta i ned at the fire judgeme nt unit 22, a li re. discri-i.mi nation output:, that is an alarm signal, is issued to the alarm display unit 24, from which an alarm is initiated.

The principle of the arithmetic operation of the fire source parameters, executed in the fire source parameter calculating unit 18 of Fig.l, is explained hereunder in detail.

Several mathematical models based on physical sciences have so far been proposed for analyzing the properties of the fire that has occurred in a room. These may be classofied into a field equation model and a zone model.

With these mathematical models, the flowing state of the smoke concentration or the temperature in the room is found 4 from the solution of a differential equation based on the volume of the heat or smoke produced from the fire source.

With the field equation model, an enclosed space having all inlets and outlets, doors and windows closed is used as a reference and the space inside is divided into hundreds of small sub-spaces each being cuboid with each side being tens of centimeters long. A mass conservation equation, a momentum conservation equation, an energy conservation equation, a status equation and a boundary condition are applied to each sub-space to find the flowing of the temperature or smoke concentration in the room. As characteristic of the field model, since the concentration in each smaller sub-space is calculated in detail, the phenomena during the fire, such as temperature or smoke concentration, can be grasped accurately.

8 Since calculation is executed for each of hundreds of the sub-spaces in the field equation model, calculating time is is increased to present a problem in real time processing and an inconvenience that the values of the parameters for the arithmetic operation can be changed only with difficulties.

With a zone model, an enclosed room space is basically taken as a reference and divided vertically into two or more layers. It is characteristic of the zone model to find the mean temperature or the mean smoke concentrations in the upper layer in the room space. Since it is a simple model, calculating time may be reduced and real time processing may be enabled by a personal computer.

The zone model has furtheradvantages that parameters such as the room dimension (ceiling area and height), ambient temperature, rate of heat loss or heat release per unit time may be set or changed freely or converted easily, the height to the boundary layer or interface between the upper and the lower layers can be found, and that the status of the endangered layer in the room can be grasped roughly. However, the zone model cannot be said to be superior in accuracy to the field model because the finite defference arithmetic operation and the number of terms employing the arithmetic operation is cancelled for improving the calculation time.

Hence, with the fire source parameter calculating unit 18 of the present embodiment, the field equation model may be employed whenever a detailed and precise arithmetic operation is necessitated, while the zone model may be employed whenever real time processing is necessitated- 9 10606 ON X?1/X~l T:VT HA TO, I'0/TT The following description is made of the case in which a zone model is employed, in which a real time operation may be made to discern the initial fire, although the accuracy is slightly lowered because the dama available on detection of fire occurrence is the output of the. sensor provided in the room Various methods have been used to put the zone model to practical usage. Hlowever, there Jacks an example of practical utilisation as the independent theory. With the present embod:Jment, a two-layer model ASMT (Avalable Safe Egress Time) -BI" is applied. which is one of the progrms of the miathemnatical model analysed by L.Y- Cooper") and evolved by W.D- Walton based on the tbeory.

=References= Cooper, A Mathematical Model for Estimating Available Safe Agree Time ia Fires, Fire and Materials, vol-6. Nos. 3 and 4, pp.l1 35 -1 4 4 1982, Sept[Dec.

Walton, ASET-B, A Room Fire Program for Personal Computers. National Bureau of Standiards NBSIR 85-3144; 1985 Apil, pp.'1- 35 Thus, with the present embodimnent, a reverse arithmetic operation of the mathematical model is performed on the basis of the detection data from the sensors to find changes in the amount of the beat, smoke or thegas in the fire source and to discriminate the fire occurrence on the basis of the results of the arithmetic operation- VIIVUSflV J'I NOSYMOD SaIAV(I 966E E996 z T90 LT:CT TO, tO/TT VI'IV~tSflV d1 NOSITIOJ S3IAVcI ±r:vT Ta, ~O/TT A zone model as the fire simulation calculating model, as ca-lcul.ated at the fire source parameter calculating unit 18, is~ shown diagrammatically in Iigq.2.

The zone model shown in Fig.2 is a two-layered zone model. Since it is a simple model adapted for finding the mean temperature 0 h or the mean smoke concentration Csh of the upper layer 28, the conditions are set in the following manner.

It is assumed that all of the openings, with the ir) exception of slight leakage from the floor surface, that is the inlets and outlets, doors or windows, are closed, the atmospheric pressure in the room is constant and an increase in the atmospheric pressure in the room may be disregarded due to the leakage from the floor surface.

I It is also assumed that the fire occurs at a base of fire set on the floor surface. The heat or hot smoke released from this base of fire rises by buoyancy to reach the ceiling surface. A plume 26 generated at this time rises as it laterally entrains the ambient cold air, and the hot air current reaching the ceiling is diffused and reaches the side wall surface to form a hot layer, that is an upper layer 28.

An interface 32 thus produced between the upper layer and a lower air layer 30 gradually descends towards the floor surface as -the fire proceeds with the lapse of time.

It is further assumed that, with such two-layered zone model, the temperature and the smoke concentration are uniform in each of the hotter upper layer 28 and the lower layer which is at the ambient temperature, and that interlayer heat 1 I

S

4 S S

S.

S S. S *5S* cx(:hanqe occurs by way of the plume 26.

The simulation finds,. f rom the previouslI.y q raspe1 heat release rate per unit time of: the combustion material, the temperature i) h of the Upper layer 28 and the distance Z from the base of f ire to the interface 32.- Thus the distance Z from the base of fire to the interface 32 of the upper layer 28. the mean temperature 0 h of the upper layer 28 and the smoke concentration Csh, may be found by solving the undermentioned differential equations. The initial conditions set by the initial value setting unit are shown simultaneously with the differential equations. As regards the CO gas concentration Gh, the equations similar to that for the smoke concentration Csh are applied.

Equation -Cl a Q -C2- 0Q Z513 with O< Z< Zo dz /dt QWith Z< 0 0 (with Z= -F) Equation 0Oh [Cl- nQ h/Oo0-1).C2. A Q/3 ZS/'3)/(ZOZ) with O< Z;S Zo dO h/dt= 0i n Q/(Zo+z) (with -F Z 0 Equation dCsh/dt= with O< ZA Zo 0 h/O o- A Csh/(Zo-,) (with Z& 0 1 2 111 it iai I couidi L.~ions t o) Zo z 0j h C-01 dO h /dt dCsh/dt C) o( 11 Zo' A Qo C1/C2); ACs/ A Cso -Zo 3 *Qo 1 3 /C2 (where 8 Cs/ n Cso)=l 0o(Ci I Qo' /C2) 2 8 Qf/ A Oo+5 (C1*- L\ Qo+C2- 6QO1I/ 3 ZOlI (Cl 1 Q02 3 /C2) 5+ (Zo/ (Cl* Qo)) A Csf/,n Cso !N Of/ A_ Qo-f5 -C 2. Qo'1' 3 -1Zo 2 ]I(6.ZoR/ 3 where: 0 0 0* 00 0 *0 0 S

S

0000

S.

0

S.

bOSS S C *5 Sb 5055 *000

S

*05*

I

o Cl C2 nx Of Cs f

.ANQ

.3 C s A 00 ACso

S

H

F

CP

L R

LC

o1 9 0 (l-LC) p -CP- 0 o- S) o0 -dA Qo/dt; -d8 Cso/dt. heat release rate paer unit time; smoke production rate per unit time; heat release rate at initial time; smoke production rate at initial time; f loor area of enclosure; ceiling height; height of base of fire; specific heat of air; radiative heat loss fraction; convective heat loss fraction; ambient temperature; gravitational acceleration; density of air; 1 3~ With such two-I ayer zone model chcinges in the Lemper ature or in t lie simok e concent rat ion a c tre asped as changes in the rate of heat: release per unit time or as changes in the rate of smoke production per unit time and, as for the distance to the interface 32. the changes in the heat release rate per unit time are grasped as the changes per unit area. Further, if an advanced Euler's equation is employed, for solving the differential equations, its arithmetic operation may be performed with higher speed.

In the present embodiment, the detected temperature 0 at the temperature sensor 10 and the detected smoke concentration sensor 12 are handled as the mean temperature 0 h and the mean smoke concentration Csh, respectively, of the upper layer 28 in the two-layered model, to calculate the changes in the rate of heat release and smoke production per unit time.

~Fig.3 shows a flow diagram of the fire detection algorithm based on the estimation of the fire source parameter in the embodiment of Fig.l.

Referring to Fig.3, prior to actuation of the system, initial values are set at step S1 in the fire source parameter calculating unit 18 by the initial value setting unit During this initial value setting, all of the initial approximating values other than Cl, C2. A OL. 8 CsL, heat release rate per unit time 8 Q and smoke production rate per unit time A Cs, shown at the proviso of the initial conditions of the differential equations and for the twolayered zone model, are input or set by an internal arithmetic operation.

1 /I After setting the initial values at step SI, the system is in the active condition, and the program proceeds to step S2. where the detected data of the temperature 0 from the temperature sensor 10, the smoke concentration Cs from the smoke concentration sensor 12 and the gas concentration G from the CO gas concentration sensor 14 are sampled at fixed time intervals at the sampling curcuit 16.

Then, a change in the heat release rate per unit time A Q is found at steps S3 to S6. Thus, at step S3, the initial value of A Q is set. At step S4, the mean temperature e h in the upper layer and the distance to the smoke layer Zh at Sthis time are calculated by the arithmetic operation of the aforementioned ASET-B. Then at step S5. the absolute value of the difference between the mean temperature 0 h calculated by the ASET-B and the temperature 9 detected at the temperature sensor 10 is compared with a predetermined value c and the steps S3 to 55 are repeated until the difference becomes equal to or lesser than the value c such as 0.001, to increase the initial set value of A Q gradually. At step S6, the value of A Q when the condition of the step S5 is satisfied is set as the value of the heat release rate at this time.

The program then proceeds to step S7 where the smoke production rate A Cs and the gas concentration A G are set.

At step S8. the temperature 0 h already found at this time is used to execute the arithmetic operation of ASET-B to find the smoke concentration Csh and the gas concentration Gh.

At step S9, it is checked whether the absolute value of the difference between Cs and Csh and the absolute value of the 1 ldi1.ference Lbetween G and Gh are not more than a predetermined value such as; 0.001. 1 f t:hhis condition is not met, the steps S7 to S9 are repeat-ed to increase A G and A Cs gradually. The values of Cs and A C at the time point when the condition of step S9 is satisfied are set as the smoke production rate and the gas production rate at such time (step S 10) The program then proceeds to step S1l where it is analyzed whether the change in the heat release rate A Q i set at the step S6 and the changes in the heat release rate A Cs and in the gas production rate A G set at the step exceed the previously set fire discrimination criteria (alarm initiation level). If analysis of the results of the arithmetic operation at step S11 shows that these results exceed the alarm initiation level, the program proceeds to *ooo step S12 where a predictive arithmetic operation is carried out using the so far obtained changes in the heat release rate A Q, in the smoke production rate A Cs and in the gas i production rate A G. The Newton's regressive interpolation formula, for example, may be employedfor such predictive arithmetic operation. Besides the above mentioned predictive operation, the changes in the first order difference and/or the second order difference until the time point a predetermined number of times of sampling before the current time point when the result of the arithmetic operation exceeds the initiation level, or correlation and/or weighted correlation value between respective result; of the arithmetic operations may also be found by way of an arithmetic operation for fire I G discrimination at step S12.

At: step S13, the occurrence of a fire is judged in accordance with the result obtained at the. step S12.

Figs.4a. 4b and 4c are charts in which temporal changes in the heat release rate A Q, in the smoke production rate A Cs and in the gas production rate A G, obtained by the fire source parameter calculating unit 18 of this embodiment, are shown with the sensor detection temperature e the distance

L

to the interface of the upper layer, the detected smoke concentration Cs and the detected gas concentration G, as measured by the fire source parameter calculating unit 18 of the present embodiment, on the occasion of a fire experiment in which a chair materials: cloth, urethane foam and wood is burned at the floor center of a room having a floor area of 6.7 x 4.3 2 8 .81mz and a ceiling height of 2 Figs.5a, Sb and 5c are similar charts for an experiment for false alarm source in which, by way of an example of a cooking in a kitchen, nine dishes of fish are grilled in the same room as that of Fig.4.

It is seen from comparison of the results of Fig.4 for fire and those of Fig.5 for false alarm source that the temporal changes in the heat release rate A Q shown in Fig.4a in case of a fire shows an acute peak at the time point when the temperature 0 rises suddenly with the progress of the fire. Conversely, no such peak is observed with the change in the heat release rate A 0 in the case of the false alarm source shown in Fig.5a. Thus the occurrence of a fire may be discriminated from the correlation when both the temperature 0 andc t:lh heat release rate A Q rise linearly. In case of a tJ re, t:le change in the smoke production rate A Cs shown in Fig.4b and the change in the gas production rate AG shown in Fig.lc rise correlatively to peak values with relation to the change in the heat release rate Q shown in Fig.4a, so that more accurate fire discrimination may be achieved by checking the correlation between at least two of the three parameters, namely the heat release rate A Q, the smoke production rate A Cs and the gas production rate A G.

.Conversely, in case of a false alarm source, shown in ig.5, there is no correlation between the changes in the smoke production rate A Cs and in the gas production rate A G on one hand and the change in the heat release rate A Q' on the other, from which the false alarm source can be discerned accurately. It is also possible to discriminate the fire and the false alarm source on the basis of the fact that the smoke production rate A Cs and the gas production rate A G for the false alarm are similar in the change pattern to those for the fire as shown in Fig.4 but the extent of change is lesser in the case of the false alarm than in the case of fire.

Figs.6a and 6b show the results of the similar experiments in which temporal changes in the heat release rate A are plotted against temperature 0 for different size of room. The results show satisfactory coincidence in the calculated changes in the heat release rate A Q despite change in the room size. It is thus seen that the same change in the heat release rate A 0 may be obtained with the present embodiment for the same fire without regard to the room size.

18 'This applies for the rate of smoke production A Cs and the rate of: gas production A G as well.

A specific embodiment of a fire judgement unit 22 shown in Fig.l is hereinafter explained.

With the present embodiment, the occurrence of a fire is discriminated by the correlative arithmetic operation at the fire judgement unit 22 with the use of two of the parameters obtained at the fire source parameter calculating unit 18, namely the heat release rate A Q, the smoke production rate A Cs and the gas production rate A G.

The correlation factor R is first defined by the following formula R =Sxy/J Sx-Sy (4) where Sxy, Sx and Sy are expressed by the following formulas: m2 Sx Xi X )2 i m t

)Z

SSy Yi Y )z Sxy Y Xi-Yi n-X-Y where X, Y denote a combination of any of two of A Q, A Cs and A G; X .Y denote temporally averaged values; n denotes the number of date used m2 ml 1 The correlation factor R calculated by the equation (4) is multiplied, for weighting, by the absolute value |DI of a composite vector D determined by two values employed in the 1 9' corre.at: ion calculat ion f ind a weighted correlation value The absolute( v.,luie II) of. the composite vector used for the wei(ghting can be cdefined as following equation: I I Ui+Vi j (6) where U and V denote any of two of the fire source parameters A O, A Cs and A G which have been separately converted with optimized scaling and i and j are the unit vectors of the respective dimensions.

Since the correlation factor R and the composite vector D are changed temporally, the weighted correlation value R, is expressed as the time function by the following formula; Ro(t) R(t)+ID(t)l (7) Thus the value of the correlation factor R, calculated at a certain time point, is weighted in dependence upon the absolute value of the composite vector of the two detected values U and V, and the correlation value R, is found, which is the correlation factor R weighted more pronouncedly for the 0: arger values of U and U.

Fig.7a shows temporal changes in the weighted correlation value R, found from the equations to by using the heat release rate A Q and the smoke production rate A Cs during the fire shown in Figs.4a and 4b. In Fig.7a, the correlation value R, shows an acute change in the peak. Thus a fire may be identified when the correlation value Ro exceeds a preset threshold value R,.

Fig.7b shows differentiated data of the correlation value Ro shown in Fig.7a. Marked changes sufficient to identify a fire are similarly demonstrated in these differentiated 2 0 da t:a.

Fig.-8. shows the correlation va.e t ound in accordance with the formulas to with respect to the heat release rate Q and the smoke production rate A Cs for non-fire shown in Figs.5a and 5b. The correlation value in this case remains at a lower level than the threshold value so that the false alarm source can be identified. Fig.8b shows S temporal changes in the differentiated values of the correlation value R, shown in Fig.8a.

In the above described embodiment, the heat release rate e.

per unit time A Q, the smoke production rate per unit time A Cs and the CO gas production rate per unit time A G are calculated as the primary fire source parameters. However.

since ions are produced by the fire flame, ionsensors may be provided in the fire monitor area and the change in the rate per unit time of the ion production from the fire source may be calculated in the similar manner from the detected data of o the ion sensor as the primary fire source parameters so as to be used as additional fire discriminating data.

Claims (5)

1. A fire alarm system including: sensor means provided in a fire monitor area for detecting secondarily heat, smoke or gas produced primarily by the fire, to obtain detection data corresponding to temperature, smoke concentration or CO gas concentration in said area; calculation means for calculating a primary fire source parameter including a heat release rate, smoke production rate or CO gas production rate of a fire source on the basis of said detection data from said sensor means with the use of a preset arithmetic operation formula to produce at least two continuously varying parameters of a fire source; and judgement means for discriminating occurrence of a fire on the basis of the rate of change of each of said two continuously varying parameters.

2. The fire alarm system according to claim 1, wherein said calculation means is adapted to calculate said primary fire source parameter on the basis of said detection data from said sensor means with the use of reverse arithmetic operation of a fire simulating mathematical model for finding from the solution of a differential equation the flowing state of the temperature, the smoke concentration or CO gas concentration in a room being monitored, based on the volume of the heat, smoke or CO gas produced from the fire source, to analyse the conditions of a fire occurred in the room being monitored.

3- The fire alarm system according to claim 1, wherein said judgement means includes correlation calculating means for discriminating a fire by the correlation calculation employing at least two kinds of parameters obtained by said calculation means and selected from the group consisting of the heat release rate, smoke production rate or CO gas production rate of the fire source.

4. The fire alarm system according to claim 3, wherein said correlation calculating means is adapted to discriminate a fire by the calculation results weighted on the basis of an absolute value of a composite vector D which is determined by at least two kinds of parameters employed in said correlation calculation. goo a VIIVHSLJNV d'I NOSITIOD SaIAVa 966Z ZZ6 zo TQS SU:CT To, to/TT [0606 ON XH/X1L TT:gi aHM TO, VO/TT P.\WPDOCMAMD\SPa 6s]43ZDOCAAMW~rAMl432-1014M -23 The fire alarm system according to claim 1, wherein said sensor means includes temperanire, smoke concentration and CO gs concenftation sensors.

6. A fire alarm system, substantially as herein described with reference to the accompanying drawings- DATED This 10th day of April, 2001 HOCHEKI CORPORATION and BIROMITSU ISIMf By Its Patent Attorneys DAVIES COLUISON CAVE 600 VIIVUSflV d'I N0SI1100 SaIAVO 966Z Z9Z6 Z TO T RU:CT TO, VO/TT