CN1963878A - Intelligence inspection prewarning forecasting apparatus for fire of high-rise building - Google Patents

Abstract

The invention discloses a high-rise building fire intelligent monitoring, early warning and forecasting device, which includes a fire signal detector, a signal transmission device, a data manager, a fire alarm controller and an alarm device. One end of the data manager is connected with an analog-to-digital converter. The fire signal detector is connected, and the other end is connected with the fire alarm controller through the signal transmission device, and the fire alarm controller is connected with the alarm equipment through the digital-to-analog converter. The device of the invention is developed based on the nonlinear dynamic mechanism of high-rise building fires, and quantitatively calculates the complexity and uncertainty of fires based on signals such as on-site detection temperature, gas flow velocity, smoke concentration, and pressure, and analyzes the current fire situation and fire development. Trend, high reliability and scientific.

Description

High-rise building fire intelligent monitoring, early warning and forecasting device

technical field

The invention relates to a fire monitoring, early warning and forecasting device, in particular to a high-rise building fire intelligent monitoring, early warning and forecasting device with a self-learning function.

Background technique

With the rapid development of urban economic construction, the continuous improvement of people's living standards and the prosperity of other undertakings, the urban land is becoming more and more tense, and the buildings are being developed in the direction of high-rise and densification. The methods are also becoming more and more diversified. With the increase of electricity load and gas consumption, higher and stricter requirements are put forward for the design of automatic fire alarm system. Fires in high-rise buildings have the characteristics of rapid fire spread, difficult evacuation of personnel, difficulty in fighting and rescue, and many fire hazards, so they are extremely harmful. In order to ensure the safety of people's life and property, the design of automatic fire alarm system has become one of the most important contents in the design of high-rise buildings. Use high-quality and advanced fire monitoring, early warning and forecasting devices to accurately judge the location of the fire, and accurately predict whether the fire source will develop into a fire, and predict the level and development of the fire, so as to facilitate timely rescue, which is especially important to ensure the safety of high-rise buildings. .

Fire is a burning phenomenon that is out of control in time and space. At present, to improve the accuracy and timeliness of fire warning in high-rise buildings, what is most lacking is the mechanism research of fire, a complex object phenomenon, and the accurate and accurate prediction of fire occurrence and development. method of judging in time. For the manifestations of the combustion of ordinary combustible substances, firstly, combustion gas is generated, and then smoke is released. Only under the condition of sufficient oxygen supply can combustion be achieved, flame is generated, and a large amount of heat is emitted, which raises the ambient temperature. Therefore, in most cases, the fire development always takes a long time in the initial stage and the smoldering stage. At this time, the destructiveness of the fire has not reached the maximum. occurrence of serious disasters.

In the current high-rise building fire monitoring, early warning and forecasting devices, a single smoke sensor, temperature-sensitive optical cable or CCD (computer-controlled display camera) is usually used. The means are single and the reliability is poor. For example, the smoke sensor cannot detect the flame produced by the burning of alcohol, the temperature sensor is difficult to detect the smoldering fire, and the CCD camera cannot distinguish the difference between the moving high-temperature object and the fire, which may cause a missed alarm; at the same time, the current Some light-sensitive, smoke-sensitive, and temperature-sensitive detection technologies can only detect flames or fires in certain detection areas, but cannot determine the exact location of the fire. In addition, due to environmental interference, missed detections and false detections often occur. reported situation.

Contents of the invention

The purpose of the present invention is to overcome the defects and deficiencies existing in the existing high-rise building fire monitoring, early warning and forecasting devices, and provide an accurate and efficient high-rise building fire intelligent monitoring, early warning and forecasting device based on artificial intelligence and fuzzy control technology, with self-learning function .

A high-rise building fire intelligent monitoring, early warning and forecasting device includes a fire signal detector, a signal transmission device, a data manager, a fire alarm controller and an alarm device, and one end of the data manager communicates with the fire signal detector through an analog-to-digital converter. The other end is connected to the fire alarm controller through a signal transmission device, and the fire alarm controller is connected to the alarm equipment through a digital-to-analog converter.

The data manager is a computer, including two parts of a database and a data processor; the data manager is connected to a high-rise building system ventilation heat transfer measurement device; the fire signal detector is an ion smoke sensor, a gas sensor and/or temperature sensor; the fire alarm controller also includes a manual alarm button; the alarm equipment includes fire linkage, fire broadcast, fire alarm telephone, emergency lighting and alarm recording equipment.

Compared with the early warning technology of existing building fires, the present invention has the following advantages and effects:

(1) Data transmission adopts multi-priority network communication technology. The new fire alarm signal under any node machine always has the highest priority to ensure that it is transmitted to the fire control system before other event signals, combined with computer parallel processing technology to ensure short response time and fast running speed of the system;

(2) Use fuzzy neural network technology to predict the fire alarm level of high-rise building fires, return the prediction results to the system, use the self-learning ability of the neural network, and continuously modify the sample set and discrimination rules to achieve high fault tolerance and intelligence of the system;

(3) Introduce nonlinear dynamics to carry out basic theory on the fire mechanism of high-rise buildings, establish fire models of high-rise buildings with different structural characteristics, and analyze the complexity and variability of fires according to the signals such as on-site detection temperature, gas flow rate, smoke concentration, and pressure. Quantitative calculation with determinism, analysis of the current fire situation and fire development trend, provide a strong theoretical basis for rescue and fire fighting work, and greatly enhance the reliability and scientificity of the early warning system.

Description of drawings

Fig. 1 is a structural representation of the present invention;

Fig. 2 is the schematic diagram of judging the fire disaster situation by the data processor shown in Fig. 1;

Fig. 3 present invention judges the principle diagram of whether fire takes place and locates;

Fig. 4 is a schematic diagram of the non-linear processing process for predicting the development of disaster situation in the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings.

As shown in Figure 1, the high-rise building fire intelligent monitoring, early warning and forecasting device of the present invention includes a fire signal detector 1, a fire signal detector 2, an analog-to-digital converter 3, a data manager 4, a monitor 5, a signal transmission device 6, Fire alarm controller 7, digital-to-analog converter 8, and alarm equipment 9.

The data manager 4 is realized by a computer, and it is composed of a data processor 4-1 and a database 4-2 connected to each other; the fire signal detectors 1 and 2 can be ion smoke detectors, gas sensors, temperature sensors or flame light detectors It can also be a combination of ion smoke sensor, gas sensor or temperature sensor and flame light detector; the data processor 4-1 is connected with the fire signal detectors 1 and 2 through the analog-to-digital converter 3, and the data processor 4-1 It is respectively connected with the monitor 5 and the ventilation and heat exchange measuring device 10 of the high-rise building system; the data processor 4-1 is also connected with the fire alarm controller 7 through the signal transmission device 6, and the fire alarm controller 7 communicates with the alarm through the digital-to-analog converter 8 The equipment 9 is connected, the fire alarm controller 7 is provided with a manual alarm button 11, and the alarm equipment 9 includes the equipped fire-fighting linkage 9-1, fire-fighting broadcast 9-2, fire alarm telephone 9-3, and emergency lighting 9-4 in high-rise buildings. , Alarm record 9-5 and other equipment. The fire signal detector can use HST8110 intelligent photoelectric smoke detector, HST8120 intelligent temperature sensor, HST8130 intelligent smoke and temperature composite controller, HST8140 intelligent smoke and temperature CO composite controller.

The present invention adopts the fuzzy neural network, uses the fire signals obtained by detectors, investigations, experiments and theoretical calculations to form a training sample set, and uses fuzzy neural network technology for training to form an intelligent early warning of fire with high fault tolerance, complex pattern classification and identification device. As shown in Figure 2, the fire detection neural network designed by the present invention comprises three input neurons, five hidden neurons and three output neurons. The three items IN1, IN2, and IN3 on the left are the input layer. In practical applications, the three signals of smoke, temperature and gas emitted by the fire detector are converted and normalized to [0, 1], and then passed to the input layer; the three items on the right OT1, OT2, and OT3 are the output layers, which respectively represent the fire probability, fire danger probability, and smoldering fire probability, and the output value range is also [0, 1]; IM1~IM5 between the input and output layers is the hidden layer, and the input signal passes through the hidden layer and then sent to the output layer. There are 15 connecting arcs between IN(I) and IM(J) and OT(K), and their weights are Wij and Vjk respectively. The sum from the input layer to the middle layer is defined as NET1(j):

$\mathrm{<span\; class="notranslate">\; NET</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\stackrel{\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; INi</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>{\mathrm{<span\; class="notranslate">\; W</span>}}_{\mathrm{<span\; class="notranslate">\; ij</span>}}<span\; class="notranslate">\; )</span>$

The time value of NET1(j), that is, the output of the hidden layer, is converted to [0, 1] with the Sigmoid function;

$\mathrm{IM}=\frac{1}{1+\exp [-\mathrm{NET}1\left(j\right)\&Center\; Dot;r1]}$ Similarly, the sum from the intermediate layer to the output layer is defined as NET2(k):

$\mathrm{<span\; class="notranslate">\; NET</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; IM</span>}<span\; class="notranslate">\; \&CenterDot;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; jk</span>}}<span\; class="notranslate">\; )</span>$

Likewise, NET2(k) is also converted to [0, 1]:

$\mathrm{<span\; class="notranslate">\; OTK</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; [</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; NRT</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ]</span>}$

The role of r1 and r2 is to correct the coefficient of the slope of the Sigmoid function curve, usually taken as 1.0 and 1.2 respectively. In this fire detection system, a 12-mode learning definition table is used, as shown in the table below.

Table self-learning definition

serial number input output   Smoke detector  Heat detector   Gas Detector Fire Probability Fire probability Smoldering fire probability D R D R D R 1 0.1 0 1 0.7 0.661 0.6 0.702 0.9 0.802 2 0.3 0.5 1 0.9 0.885 0.9 0.889 0.1 0.037 3 0.1 0 0.2 0.3 0.254 0.2 0.187 0.4 0.289 4 0.5 0.1 0.8 0.8 0.829 0.8 0.786 0.7 0.722 5 0 0.3 0.1 0.1 0.094 0.1 0.098 0.1 0 6 0 0 1 0.4 0.453 0.7 0.588 0.3 0.376 7 0 1 0 0.2 0.190 0.3 0.307 0.05 0 8 0.3 0.2 0.5 0.7 0.781 0.6 0.701 0.3 0.247 9 0.6 0.8 0.8 0.95 0.902 0.95 0.904 0.05 0.073 10 0.2 0 0.3 0.6 0.542 0.4 0.431 0.75 0.756 11 0.1 0 0.1 0.1 0.189 0.05 0.119 0.1 0.205 12 0.4 0.2 0 0.7 0.714 0.65 0.529 0.2 0.260

In this way, the square error E _m of the mth input mode and the sum E of the square errors of the 12 modes can be expressed as:

${\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 3</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; o</span>}{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

and $\mathrm{E.}=\underset{m=1}{\overset{12}{\mathrm{\&Sigma;}}}\left({\mathrm{E.}}_m\right)$

Adjust the weight Wij, Vjk to make E reach the minimum, that is, the neural network learning process is completed. After determining the weight value, the neural network input layer starts to receive the potential value of the detector, calculates the output value according to the above method, compares the calculated value with the fire probability, the fire danger probability, that is, the smoldering fire probability, and finally decides whether or not it occurs Judgment of fire. As shown in Figure 3, the detected values of fire signals (such as smoke, temperature, etc.) are converted by the digital-to-analog converter 3, and enter the fuzzy neural network calculation module 4-1-1 of the data processor 4-1 to perform fuzzy calculation of the size of the fire. Recognition, the recognition result outputs a judgment signal B (0 or 1) whether a fire will occur.

Changes in environmental factors such as climate, heat flow, and natural light in high-rise buildings will interfere with the discrimination results of monitoring instruments. Wherein, due to the difference between the discrimination output signals A and B caused by neural network calculation errors, the present invention uses neural networks to analyze the discrimination results of manual monitoring. The learning is improved to realize self-learning and self-adaptive functions.

On the one hand, high-rise building fire is a complex nonlinear feature affected by many factors. On the other hand, although it is affected by many thermal disaster factors, it shows complexity. However, in similar environments and Under the same conditions, the occurrence and development of fire can reflect similar laws. High-rise buildings can usually be divided into spaces such as atriums, rooms, stairwells, and elevator shafts. The building materials, structures, space features, and heat exchange and ventilation systems in various spaces are designed according to their standard requirements, so that similar spaces have similar characteristics. burning environment. Based on the understanding of the characteristics of high-rise buildings and the combustion process, the invention uses a method of combining nonlinear models and pattern recognition to predict the development of fires in high-rise buildings. According to the characteristics of different types of space buildings in high-rise buildings, the volatile pyrolysis and combustion model, the combustibles flashover model, the flame propagation model, the fire spread model, the chaos model of the smoke plume, and the coupled image lattice model of the smoke spread are established. Thus, a fire prediction model considering various related factors is constructed. According to different types of spaces, basic research, laboratory experiments, field investigations and numerical simulations are used to accumulate relevant characteristics to form a fire development model set, which is stored in the basic fire data, classified, and stored in the fire database 4-2 .

As shown in Figure 4, the analog signal obtained by the fire signal detector is transmitted to the fire development prediction module 4-1-3 of the data processor 4-1 through the analog-to-digital converter 3, and the high-rise building system ventilation heat transfer measurement device 10 measures the The ventilation and heat transfer parameters are also transmitted to the fire development prediction module 4-1-3 through the data interface. When the data processor 4-1 detects a fire, it obtains the specific location of the fire by addressing the monitoring signal, and transmits the positioning signal to the fire development prediction module 4-1-3. The fire development prediction module 4-1-3 uses the positioning signal to represent the detection signal of the current fire development degree: temperature, pressure, smoke concentration, gas composition and ventilation conditions, and queries the space fire of this type of high-rise building in the database 4-2. Referring to the model set, carry out model comparison, and analyze whether the current fire belongs to smoldering, initial fire or belongs to the category of large fire. Then these detection signals are sent to the corresponding nonlinear prediction calculation model, through the volatile pyrolysis and combustion model, the combustibles flashover model, the flame propagation model, the fire spread model, the chaos model of the smoke plume, and the smoke spread model. Coupled mapping lattice model for real-time prediction of fire development. The model comparison and prediction results of fire development are displayed on the monitor 5, as well as the corresponding processing decisions stored in advance by the database 4-2. At the same time, the current fire characteristic signal and the development degree of the fire are fed back to the database 4-2 to supplement the reference model set.

Due to the variability of the fire phenomenon, some physical quantities are affected by other factors in the environment, and the instantaneous value shows a certain degree of randomness, so that the actual detected signal is difficult to completely match the given model. Therefore, the model comparison in the present invention uses The method of fuzzy recognition is judged by the closeness between the object to be monitored and the known pattern.

In summary, the present invention is designed to use advanced fire detection algorithms and fire pattern recognition methods to judge fire information. It is an intelligent fire detection and alarm system with "higher intelligence". Its working principle is as follows:

The temperature, pressure, smoke, gas composition and other data obtained by the on-site detection and investigation of the detector are input into the high-rise building fire data manager 4 through the analog-to-digital converter, and then the physical quantity signal change process is transmitted to the fire department through the signal transmission device 6. The alarm controller 7 compares the change process of the physical quantity signal with the high-rise building fire reference model established by this system. A fire alarm signal is sent out, and after judging that there is a fire alarm signal, the fire alarm signal is transmitted to the alarm device 9 through a digital-to-analog converter, and the fire-fighting agency carries out rescue work after receiving the fire alarm. The manual fire alarm button 11 artificially responds to the fire alarm controller 7 to effectively implement the alarm, and the fire disaster judgment and fire alarm accuracy rate made by the device each time are fed back to the fire data management of high-rise buildings through the analog-to-digital converter. Device 4, the artificial neural network technology self-learns the feedback data, continuously revises the sample set and discrimination rules, and then improves the forest fire reference model; the finally fed back data is stored in the fire database 4-2, which will be used as a high-rise building fire discrimination in the future basic data. This design uses the fire nonlinear mechanism, fuzzy theory and artificial neural network technology to establish a high-rise building fire database. It has the ability of self-learning and the process of knowledge discovery, and obtains reliable training sample sets and discrimination rules with high fault tolerance. It can continuously supplement and improve itself, which is helpful to the mechanism research of high-rise building fire and the degree of intelligent fire alarm, and greatly improves the accuracy, timeliness and reliability of high-rise building fire alarm.

Claims (6)

1, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building, it is characterized in that comprising fire signal detector, apparatus for transmitting signal, data management system, fire alarm control unit and panalarm, one end of described data management system is connected with fire signal detector by analog to digital converter, the other end is connected with fire alarm control unit by apparatus for transmitting signal, and described fire alarm control unit is connected with panalarm by digital to analog converter.

2, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building according to claim 1, it is characterized in that: described data management system is a computing machine, comprises database, data processor two parts.

3, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building according to claim 1 and 2 is characterized in that: described data management system is connected with skyscraper system ventilation heat exchange measurement mechanism.

4, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building according to claim 3 is characterized in that: described fire signal detector is ionic smoke sensor, gas sensor and/or temperature sensor.

5, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building according to claim 4, it is characterized in that: described fire alarm control unit also comprises manual pull station.

6, a kind of intelligence inspection prewarning forecasting apparatus for fire of high-rise building according to claim 5 is characterized in that: described panalarm comprises fire-fighting link, fire broadcast, Fire telephone, emergency lighting and alarm logging equipment.

Images