EP2758948B1 - Fire detector with sensor array - Google Patents

Description

State of the art

The invention relates to a fire alarm for a fire alarm system for monitoring a monitoring area with an interface for communication with the fire alarm system, with a sensor device for spatially resolved recording of the monitoring area in an IR wavelength range and for outputting sensor signals, and with an evaluation device for determining a fire condition by evaluation the sensor signals of the sensor device, the fire status being transmitted to the fire alarm system via the interface.

Fire detectors are locally installed units that are mounted on a ceiling or wall, for example, and are used to detect a fire in their vicinity. In known embodiments, up to three different types of sensors are used in order to detect fires at an early stage. An optical sensor is often used that detects smoke by means of the scattering of light particles. Also in use is a chemical sensor that detects carbon monoxide that is formed during a fire. Furthermore, a temperature sensor is often used, which gets its information from a temperature-dependent resistor. Since the transport mechanism used for temperature detection is heat convection and the air heating between the source of the fire and the fire alarm requires a certain amount of time, this type of sensor is slow. In addition, infrared sensors are known which evaluate the thermal radiation bundled at one point, but regardless of the direction of incidence. The speed of reaction of these sensors is fast, but only one overall statement can be made for the environment with this type of sensor.

Another approach is in the document DE10110231A1 described. This document relates to a sensor system with an optical diaphragm, the optical diaphragm being arranged in front of a measuring sensor and the optical diaphragm having a material that can be varied in its optical properties for generating the optical glare effect. The optical diaphragm is designed, for example, like a grid, with the individual grid fields being selectively activated or deactivated. This makes it possible, for example, to allow light signals to pass through to the sensor only from selected solid angle areas and to block them from other solid angle areas, so that a spatially resolved observation of the environment can take place here.

The U.S. 5,248,884 A describes an infrared detector comprising a thin layer of photosensitive material on a transparent dielectric material with an array of planar antennas adjoining the surface of the thin layer.

From the WO 2007/107988 A2 a device for the detection and localization of a thermal event and for providing a reaction to the thermal event is known. The device comprises at least one infrared thermal imaging device for obtaining a specified area. Furthermore, the device comprises a processor for processing the thermal image and for detecting hot spots in the thermal image.

The EP 1 381 005 A1 describes an event reporter with a camera for the observation of a surveillance space and an evaluation stage in which the images recorded by the camera are examined for the occurrence of parameters characteristic of events to be monitored.

The EP 1 329 860 A2 discloses an apparatus for flame detection. The device determines a flame from an image obtained by photographing a monitored object with an imager.

From the US 2003/03887 A1 an imaging fire detector is known which is used to detect a fire from a recorded sequence of images. Either a video camera or an infrared camera is used as the imaging device.

The pamphlet DE 10 2008 001 383 A1 , which probably represents the closest prior art, describes a detection device for detecting fires with an imaging sensor element which is designed to output image data. Furthermore, the detection device comprises an optical device which is connected upstream of the sensor element, the sensor element and the optical device together forming a camera device for monitoring the monitored area. An evaluation device is designed to detect the fires by evaluating the image data.

Disclosure of the invention

The subject matter of the invention relates to a fire alarm with the features of claim 1. Preferred or advantageous embodiments of the invention emerge from the dependent claims, the following description and the attached figures.

According to the invention, a fire alarm, in particular a point fire alarm, is proposed. The fire alarm is suitable and / or designed for monitoring a surveillance area. The monitoring area can be, for example, a room, a room, a hall or another open or closed area.

The fire detector is used to detect a fire in the monitoring area and thereby to determine a fire condition. Optionally, the fire detector is designed to provide further fire status information on the fire status, such as a fire probability or a fire position.

The fire alarm forms part of a fire alarm system, which preferably has a plurality of such fire alarms, which are connected to one another for signaling purposes via a network and optionally in addition to a fire alarm center to monitor the state of the fire or Exchange fire status information. For the signaling coupling of the fire alarm to the fire alarm system, the fire alarm has in particular an interface for communication with the fire alarm system. In particular, the interface is designed to couple a field bus to the fire alarm.

The fire alarm comprises a sensor device for spatially resolved recording of the monitored area. The recording of the monitoring area is preferably implemented by an optical image of the monitoring area on the sensor device. This can be an undistorted image, but it is also possible that a distorted image is accepted in order to enable 360-degree detection, for example. The monitoring area is recorded in an IR wavelength range which preferably begins at a wavelength greater than 2 μm, in particular greater than 3 μm. In particular, the sensor device is only sensitive from the mentioned wavelengths. The detection is particularly preferably carried out in an FIR (FAR INFRA RED) area.

In addition, the fire alarm comprises an evaluation device for detecting a fire in the monitoring area and for determining a fire condition by evaluating the sensor signals of the sensor device. The evaluation of the sensor signals is thus implemented within the fire alarm, with in particular the fire status or fire status information being transmitted to the fire alarm system via the interface. In particular, no sensor signals are transmitted to the fire alarm system. This has the advantage that any fire alarms, including other types of fire alarm, can be used in the fire alarm system, since the interface definition only requires the transfer of the fire status or the fire status information and dispenses with fire alarm-specific sensor signals. In particular, the fire detector forms an embedded system for determining and transmitting the fire status.

According to the invention, it is proposed that the sensor device is designed as a sensor field or includes this. The sensor field is implemented as a flat area with sensor units, so that the sensor device is designed as an image sensor or includes it. In particular, the Sensor device or the sensor field designed as an image sensor of an infrared camera. The sensor device outputs a measured value field as sensor signals, which field comprises the spatially resolved measured values of the sensor units or values derived therefrom. In particular, the measured value field is a temperature image of the monitored area.

The advantage of the invention is that by using an IR-sensitive sensor field on the one hand, information from the thermal radiation is used, in which the relevant information is transmitted at the speed of light and is thus available virtually without delay. On the other hand, complex system technology such as the optical diaphragm is dispensed with, so that the structural design can be kept simple and thus not susceptible to failure and inexpensive. The sensor field is preferably designed to be uncooled, so that complex cooling measures can be dispensed with.

In a particularly preferred implementation of the invention, the fire alarm has a wall or ceiling housing, which is designed in particular for a recessed or plastered arrangement in the monitoring area. This design has the advantage that the fire alarm according to the invention can be used in a fire alarm system instead of the ceiling or wall fire alarms known to date without having to adapt the ambient conditions.

The sensor field is designed as a bolometer field. The sensor field has a multiplicity of bolometers or microbolometers as the sensor units, which are arranged, for example, in a grid-like manner or distributed in concentric circles. In particular, the bolometer field is uncooled. Radiations in an IR wavelength range greater than 5 μm are particularly preferably recorded by the bolometer field.

A bolometer is a radiation sensor that detects energy emitted in the form of electromagnetic radiation, preferably via the bolometer heating up due to the absorption of the energy. The resulting heating creates an ohmic resistance of the bolometer or changed in each sensor unit of the sensor field and can be used as a basis for a sensor signal. As an alternative to this, a temperature-dependent change in the characteristic curve of diodes in the bolometer or in the sensor unit can also form a basis for a sensor signal. In particular, the sensor field is according to the document WO2007 / 1447663A1 educated.

When the fire alarm is put into operation, calibration can be carried out with a black body radiation source, in which a minimum temperature and a maximum temperature are assigned to a corresponding voltage across the resistors or diodes. The current is kept constant during calibration. With the calibration, an exact measurement in the temperature range between the two points of the minimum temperature and the maximum temperature is possible. If the measuring object behaves like a black body, then a modeling of the radiation intensity and thus also the electrical power or the voltage with constant power supply of the sensor units of the bolometer field with the relation U ∼ T ⁴ between the two calibration points is correct or sufficiently accurate. The sensor device is thus able to output a calibrated temperature image as a measured value field.

With the aim of creating a cost-effective implementation of the fire alarm, it is proposed that the number of sensor units in the sensor field be selected to be less than 20,000, preferably less than 10,000 and in particular less than 5,000. This special adaptation of the sensor field to the needs of the fire alarm is based on the consideration that it is not necessary for the detection of a fire to record a high-resolution image of the monitored area by the sensor field. Rather, it is sufficient to use a very limited number of sensor units and, for example, to use a sensor field with 30 x 40 or 100 x 150 sensor units. The sensor units form pixels in a later image.

This configuration on the one hand lowers the costs for the sensor field and on the other hand keeps the effort for the evaluation of the sensor signals low, since in comparison to conventional infrared cameras with more than 100,000 or 200,000 sensor units, the evaluation effort is correspondingly lower. For example, it can be sufficient that the spatial resolution in a monitoring area with a size of 4 mx 4 mx 2.5 m is worse than 10 cm, preferably worse than 30 cm. This low resolution also allows the evaluation device to be equipped with low computing power and, for example, to be designed as an inexpensive microcontroller.

The evaluation device has a feature extractor module and a detection module, wherein the feature extractor module extracts a feature or a plurality of features from the sensor signal and the detection module detects a fire condition and optionally additionally a normal condition based on the feature or features.

One of the greatest challenges in fire detection is the correct distinction between fires and disturbance variables. Disturbance variables are objects and processes that have fire properties and can therefore accidentally trigger a fire alarm, although there is actually no danger. Examples of such disturbance variables are hot air, cigarettes, candles, steam, hot stove plates, etc. To increase the detection reliability of the fire alarm, the feature extractor module extracts different features from the sensor signals and the features are evaluated by the detection module. In contrast to one-dimensional sensors, such as simple infrared detectors, it is possible with a sensor field to obtain a larger number of features with regard to intensities, temporal progressions of the intensities, areas, temporal progressions of surfaces, etc. and then to evaluate them.

Possible features that are assessed by the detection module are listed below. The detection module can use some, all or any selection of the features:
One possible feature relates to a maximum temperature in a measured value field of the sensor field. To determine this size, one determines the pixel value or entry of a measured value field that has the absolute highest temperature. At The maximum temperature shows best the speed advantage that fire detection with infrared radiation offers compared to fire detection via thermal convection. In order to filter out statistical outliers, the pixel value of the 2, 3, 4 or 5 highest temperature can be used instead of the pixel value with the absolute highest temperature.

When an open fire breaks out, the measured value field is immediately followed by a rapid rise in the maximum temperature, often up to saturation of the electronics, at, for example, 500 ° C., so that the maximum temperature feature is highly informative. Due to the rapid increase in the maximum temperature, especially in the case of an open fire, the time course of the maximum temperature over several measurement value fields is significant and should be provided as a possible further feature. For example, the time derivative of the time profile of the maximum temperature can be evaluated over several meter fields.

Another feature relates to the average temperature in a measured value field, all or at least a large part of the entries or pixel values in the measured value field being evaluated. Another feature relates to the temporal course of the average temperature over several measured value fields, wherein the temporal derivation of the temporal course of the average temperature can be evaluated.

A further feature relates to the total energy in the measured value field, in a preferred embodiment the measured temperature values at each sensor unit being weighted with their fourth power and then the mean value being formed over all sensor units. The total energy is proportional to this quantity according to the Stefan-Boltzmann law. Due to the weighting with the 4th power, even small areas in the measured value field with excessive temperatures lead to a significant change in the total energy. The time course of the total energy over several measured value fields is also suitable as a possible feature. Particularly preferably, the time course of the total energy is integrated over time.

Another possible feature relates to the number of sensor units in a measured value field above a limit temperature, whereby the size or expansion a fire or another hot object can be detected relative to the entire surveillance area. This method offers a good way of differentiating disturbances with a small surface, such as cigarettes or a tea light, from larger-scale fires. Another possible feature relates to the time course of the number of sensor units over several measurement value fields above a limit temperature, whereby, for example, an expansion of a fire and the speed of the expansion can be registered.

Another feature relates to the average temperature of all entries or pixel values of a partial area (hot spot) with entries or pixel values above a threshold value. This feature uses a two-stage analysis, with sub-areas in the monitoring area with entries above a threshold value being extracted first and the average temperature in this sub-area being determined in a second step. Another possible feature relates to the time course of this average temperature. The advantage of this evaluation is that the feature is more meaningful than the average temperature of the entire room.

Another possibility, after the "hot spots" have been identified, is that they are followed over time (tracking) and these are examined independently of one another. The various hot spots can be distinguished from one another on the basis of their position.

Another possible feature relates to the ratio between a maximum height and a maximum width of a partial area with entries or pixel values above a threshold value (hot spot). With this characteristic, the typical shape of an open fire is used in comparison to disturbance variables.

Other possible features relate to the change in position of partial areas (hot spot) with entries or pixel values above a threshold value, the determination of the position of a fire source and / or the determination of flicker frequencies of partial areas, since e.g. open fires have a typical flicker frequency of approx. 3 Hz, which for example can be recognized by a Fourier transformation of the sensor signals.

All, some, in particular a selection of the features mentioned are evaluated in their entirety by the detection module. For the evaluation, for example, there are simple methods in which each characteristic whose value is above a limit value is evaluated with a 1 and otherwise with a 0. If the number of features evaluated with 1 exceeds a further limit value, a fire condition is concluded. In a further development of this method, the evaluations of the features can also be weighted in order to emphasize significant features before less significant features.

Figur 1

eine schematische Blockdarstellung eines Brandmeldesystems mit mehreren Brandmeldern;

Figur 2

ein Blockschaltbild eines Brandmelders aus der Figur 1;

Figur 3

einen Graphen zur Illustration des Merkmals "Maximaltemperatur";

Figur 4

einen Graphen zur Illustration des Merkmals "Durchschnittstemperatur";

Figur 5

einen Graphen zur Illustration des Merkmals "Gesamtenergie";

Figur 6

einen Graphen zur Illustration des Merkmals "Verhältnis Höhe/Breite";

Figur 7

einen Graphen zur Illustration des Merkmals "Durchschnittstemperatur im Hotspot";

Figur 8

einen Graphen zur Illustration des Merkmals " Anzahl von Sensoreinheiten/Pixeln/Einträgen in einem Hotspot".

Further features, advantages and effects of the invention emerge from the following description and the attached figures. Show:

Figure 1

a schematic block diagram of a fire alarm system with several fire alarms;

Figure 2

a block diagram of a fire detector from Figure 1 ;

Figure 3

a graph to illustrate the feature "maximum temperature";

Figure 4

a graph to illustrate the feature "average temperature";

Figure 5

a graph to illustrate the feature "total energy";

Figure 6

a graph to illustrate the feature "ratio height / width";

Figure 7

a graph to illustrate the feature "average temperature in the hotspot";

Figure 8

a graph to illustrate the feature "number of sensor units / pixels / entries in a hotspot".

The Figure 1 shows in a schematic block diagram a fire alarm system 1 with several fire alarms 2. The Fire detectors 2 are arranged in a room 3 as a monitoring area and are used to detect a fire 4.

The fire alarm system 1 comprises a fire alarm center 5 to which the fire alarms 2 are connected via a field bus or a network 6. The fire alarm center 5 can also be connected to other, not shown, also different types of fire alarms. Status information of the fire alarms 2 is transmitted via the network 6, in particular a fire status and further metadata on this fire status are transmitted as status information.

The fire alarms 2 are designed as structural units, which according to the Figure 1 for example, can be used for wall mounting or ceiling mounting, in particular for surface mounting or flush mounting. The fire alarms 2 have a detection area α of, for example, 40 degrees or 60 degrees; in modified embodiments, the fire alarms 2 can also be designed as 360-degree fire alarms. For this purpose, the fire alarms 2 can have a fisheye lens or a DOME lens.

The Figure 2 shows one of the fire alarms 2 in a schematic representation. The fire alarm 2 comprises a sensor device 7 which comprises a sensor field with a multiplicity of sensor units. In this exemplary embodiment, there is a sensor field made up of bolometers or microbolometers. The sensor field is designed to detect thermal radiation in an IR wavelength range, in particular this is sensitive above a wavelength of 3 μm or 5 μm. The sensor field is equipped, for example, with 30 × 40 sensor units, in particular bolometers, or with 100 × 150 sensor units or bolometers. With this resolution of the sensor field in the sensor device 7, a spatially resolved monitoring is possible, but no detailed monitoring of the monitoring area 3. This is also not necessary since the fire alarm 2 should not record any details of the surrounding area 3, but should only detect the existence of a fire 4. On the other hand, due to the small number of sensor units in the sensor device 7, significant manufacturing costs are achieved compared to conventional infrared cameras with resolutions of 480 × 640 pixels. To obtain a spatial resolution of the fire detector 2 in To implement the monitoring area 3, the sensor device 7 is preceded by an imaging device 8, for example an optical system. The imaging device 8 enables the monitoring area 3 to be imaged onto the sensor device 7 within the viewing angle α. The sensor field, in particular the bolometer field, can be calibrated in such a way that the sensor device 7 provides, as sensor signals, an image of the monitored area 3 with temperature values as pixel values as a measured value field with an areal resolution or spatial resolution.

The sensor signals of the sensor device 7, designed as measured value fields with temperature values or as temperature images, are transferred to an evaluation device 9, which has a feature extractor module 10 and a detection module 11.

The evaluation device 9 is implemented as a microprocessor which - compared to a DSP - has a lower energy consumption and is characterized by low costs. The reduction in the number of sensor units in the sensor device 7 and the evaluation algorithms described below mean that an inexpensive microcontroller can be used.

The feature extractor module 10 examines the sensor signals for a plurality of features, which are described below. The features with their parameters are transferred to the evaluation module 11, with a decision being made by evaluating the features in their entirety as to whether a fire condition or a normal condition is present in the monitoring area 3.

In a simple embodiment of the evaluation module, the characteristics of the features are compared with previously defined threshold values, a logic table being formed in which features with exceeded threshold values with 1 and with not exceeded threshold values with 0 are kept. If a previously defined number of threshold values is exceeded, a fire condition is assumed.

Another possibility is the different weighting of characteristics, with the most suitable characteristics for differentiating between fires and malfunctions being weighted the most. Another possibility is the use of fuzzy logic to calculate the individual probabilities to an overall probability for a fire condition.

It is also possible that the evaluation module 11 contains a classifier which is trained in advance with sensor signals from real sources of fire and from disturbance variables such as smoldering cigarettes, hot stove plates, candlelight, etc. Such training methods of classifiers and the evaluation of classifiers are sufficiently known, for example, from digital image processing, so that the implementations can be based on digital image processing.

The result of the evaluation module 11 - in particular the presence or absence of a fire condition - is transferred to an interface 12 to the network 6 and thus reported to the fire alarm system 1.

It should be pointed out that the evaluation of the sensor signals takes place within the fire detector 2 and only the evaluation results, in particular the information on fire status / normal status, are passed on via the network 6.

The Figure 3 shows in a schematic graph the time courses of the feature “maximum temperature” of a smoldering fire 13 and an open fire 14. The feature “maximum temperature” is determined as follows. The sensor unit or the pixel with the highest temperature value is determined from a measured value field of the sensor device 7. In the graph of the Figure 3 the maximum temperature determined in this way is plotted over time. In the case of the open fire 14, it can be clearly seen that the maximum temperature rises rapidly to up to 500 ° C., the electronics being saturated above 500 ° C.. The maximum temperature feature thus represents a high level of significance for the open fire 14. However, even with the smoldering fire 13, the maximum temperature rises to over 100 ° C., so that a medium significance is given.

Another feature is the time course of the maximum temperature, the course of the curve of both the smoldering fire 13 and the open fire 14 show a significant increase in the time domain around 200 s. The derivation of the behavior over time of the maximum temperature can thus also be used as a feature. It should also be pointed out that the drop in the curve of the smoldering fire 13 and the open fire 14 is due to the small amount of material to be fired.

The Figure 4 shows a graph in the same representation as in FIG Figure 3 , but the time course of the average temperature of all sensor units of the sensor device 7 is shown. Here, the curve of the open fire 14 again shows a high level of significance, whereas the curve of the smoldering fire 13 is not very significant. Optionally, in addition - at least for the open fire 14 - the time derivative can be used again.

The Figure 5 shows a next graph in the same representation as in the previous figures, this graph showing an integration of the total energy in arbitrary units. For the total energy, the temperature values are raised to the power of 4 in order to obtain the energy according to the Stefan-Boltzmann law, and then the mean value is formed over all pixels. The proportional factors are not included for the sake of simplicity. The course of the curve for the open fire 14 is significant both when a threshold value is exceeded and when there is a gradient, so that this feature is meaningful. Another possible significant feature is given by the exponential increase in curve 14 at the beginning of the fire.

In the Figure 6 the ratio of height to width of a sub-area of the measured value field is plotted in the same representation as in the previous figures, the sub-area having entries or pixels with temperature values above a threshold value. Here, too, a significant increase above 1 can be seen in the case of the open fire 14, whereas the smoldering fire 13 remains below the value 1 with less significance. This feature is based on the consideration that an open fire in particular usually has a ratio greater than 1 due to its actual appearance.

In the Figure 7 the average temperature of pixels or entries is plotted in a sub-area of the measured value field, the sub-area being entries or Has pixels with temperature values above a threshold value. Here both the curve progression of the open fire 14 and the smoldering fire 13 show a significant progression, so that a possible feature with regard to the exceeding of a threshold value for the average temperature as well as for its temporal progression can be obtained from this evaluation.

In the Figure 8 the number n of entries or pixels is plotted in a sub-area of the measured value field, the sub-area having entries or pixels with temperature values above a threshold value. Here, too, the curve profile of the open fire 14 shows a very significant behavior, whereas the curve profile of the smoldering fire 13 has only moderate significance, so that the number n or its time profile can also be used as possible features.

Claims (8)

1. Fire detector (2), in particular for a fire detection system (1), for monitoring a monitoring area (3),
   in particular having an interface (12) for communicating with the fire detection system (1),
   having a sensor device (7) for recording the monitoring area (3) in a spatially resolved manner in an IR wavelength range and for outputting sensor signals,
   having an evaluation device (9) for determining a fire state by evaluating the sensor signals from the sensor device (7), wherein the fire state, in particular, is transmitted to the fire detection system (1) via the interface (12),
   wherein the sensor device (7) comprises a sensor field which outputs a measured value field as sensor signals, wherein the sensor field is in the form of a bolometer field,
   characterized in that
   the sensor field has fewer than 20 000, preferably fewer than 10 000, in particular fewer than 5000 sensor units,
   wherein the evaluation device (9) has a feature extractor module (10) and a detection module (11), wherein the feature extractor module (10) extracts a feature or a plurality of features from the sensor signal and the detection module (11) detects a fire state or a normal state in the monitoring area (3) on the basis of the feature(s),
   wherein a feature relates to the average temperature in a section of the measured value field with entries or pixel values above a threshold value.
2. Fire detector (2) according to Claim 1, characterized by a wall or ceiling housing for arranging the fire detector in the monitoring area (3).
3. Fire detector (2) according to Claim 1 or 2, characterized in that a feature relates to a maximum temperature in the measured value field and/or to the temporal progression of the maximum temperature over a plurality of measured value fields.
4. Fire detector (2) according to one of the preceding claims, characterized in that a feature relates to an average temperature in a measured value field and/or to the temporal progression of the average temperature over a plurality of measured value fields.
5. Fire detector (2) according to one of the preceding claims, characterized in that a feature relates to a total energy in a measured value field and/or to the temporal progression of the total energy over a plurality of measured value fields.
6. Fire detector (2) according to one of the preceding claims, characterized in that a feature relates to a number of sensor units in a measured value field above a limit temperature and/or to the temporal progression of the number of sensor units over a plurality of measured value fields above a limit temperature.
7. Fire detector (2) according to one of the preceding claims, characterized in that a feature relates to the height/width ratio of a section in the measured value field with entries or pixel values above a threshold value.
8. Fire detector (2) according to one of the preceding claims, characterized in that the detection module (11) assesses a plurality of features in their entirety in order to detect the fire state or the normal state.

Images