Electrical Fire Alarm Detector

Abstract

The present invention relates to an electric fire detection alarm and comprises an electrochemical CO detection sensor, a temperature and humidity sensor, an arc sensor in the ultraviolet wavelength band, a control unit, and an RS485 communication unit. The control unit stores the number of arcs generated per unit time in a circular ring buffer memory and calculates a moving sum based on an index pointer, and calculates temporal correlation sensitivity using the time delay and slope relationship between the leading increase pattern of the number of arcs generated and the lagging increase pattern of the CO concentration or the CO concentration increase rate. In addition, the control unit performs corrections according to temperature and humidity conditions and generates a step-by-step fire alarm while excluding false alarms caused by line noise or external gas inflow based on the moving sum and temporal correlation sensitivity. When using the present invention, Arc (Ultra Violet), CO Gas, and ambient temperature and humidity detection sensors are installed to detect major causes of electrical fires, such as short circuits, overloads, equipment overheating, leakage circuits, and poor contact, in advance, thereby having a significant effect in preventing fires.

Description

Technology Field

[0001] The present invention relates to an electric fire monitoring alarm, and more specifically, to an electric fire monitoring alarm that prevents fire by detecting in advance major causes of electric fires such as short circuits, overloads, equipment overheating, leakage battery faults, and poor contact, by installing Arc (Ultra Violet), CO Gas, and ambient temperature and humidity detection sensors. Background Technology

[0002] Generally, according to fire cause reports from fire and disaster prevention agencies, electrical fires account for 25–30% of the total number of fire incidents. The major causes of electrical fires include short circuits, overloads, equipment overheating, leakage current, ground faults, and poor contact. Accordingly, the types of electrical fires in distribution panels can be broadly classified as follows.

[0003] First, it is ignition caused by a short circuit.

[0004] Second, it is ignition caused by overheating due to poor connection.

[0005] Third, it is ignition caused by the accumulation of heat.

[0006] Fourth, ignition caused by a ground fault.

[0007] Arc detection sensors detect only continuous spark discharges, which are the direct cause of electrical fires, and cut off the load. As an example of conventional patent technology, Publication No. 10-2023-0165071 discloses a monitoring device capable of fire prevention and fault diagnosis using voltage / arc sensors in a solar junction box, along with technology for data processing and analysis for arc detection and fault diagnosis, control technology for protective devices including circuit breakers inside the junction box based on the analysis results, and procedures for an alarm system and data transmission to convey the relevant information. Additionally, Publication No. 10-2023-0086923 discloses an arc detection system in which an arc detection sensor is located in the indoor part of an ESS to detect arcs generated in the ESS. However, the aforementioned conventional technologies had the disadvantage of failing to prevent fires because arc, CO gas, and ambient temperature and humidity detection sensors were not installed in combination, or because there was no correlation between the sensors, making it difficult to accurately detect major causes of fire such as short circuits, overloads, equipment overheating, leakage battery faults, and poor contact in advance. Prior art literature

[0008] Korean Published Patent Application No. 10-2023-0165071 (Published December 5, 2023) The problem to be solved

[0009] Therefore, the present invention was devised to solve the above-mentioned problems and aims to provide an electrical fire monitoring alarm that prevents fire by detecting in advance the main causes of electrical fires, such as short circuits, overloads, equipment overheating, leakage circuits, and poor contact, by installing Arc (Ultra Violet), CO Gas, and ambient temperature and humidity sensing sensors.

[0010] The arc detection sensor of the present invention is configured to detect only continuously occurring spark discharges by monitoring the ultraviolet (UV) region, specifically the wavelength band of 185 to 260 nm, and does not react to sunlight or light such as fluorescent lamps or LEDs. means of solving the problem

[0011] The present invention relates to an electric fire detection alarm and comprises an electrochemical CO detection sensor, a temperature and humidity sensor, an arc sensor in the ultraviolet wavelength band, a control unit, and an RS485 communication unit. The control unit stores the number of arcs generated per unit time in a circular ring buffer memory and calculates a moving sum based on an index pointer, and calculates temporal correlation sensitivity using the time delay and slope relationship between the leading increase pattern of the number of arcs generated and the lagging increase pattern of the CO concentration or the CO concentration increase rate. In addition, the control unit performs corrections according to temperature and humidity conditions and generates a step-by-step fire alarm while excluding false alarms caused by line noise or external gas inflow based on the moving sum and temporal correlation sensitivity.

[0012] The control unit is characterized by updating the number of arc occurrences detected by the arc sensor at regular intervals to calculate the accumulated quantity during the previous period using a moving sum method, and generating a fire alarm when the accumulated quantity exceeds a predetermined set value.

[0013] In addition, the control unit may be configured to align the number of arcs generated per unit time of the arc sensor, the CO concentration of the electrochemical CO detection sensor, and the temperature and humidity values ​​of the temperature and humidity sensors along the same time axis, store data from a recent period in a circular ring buffer memory, and calculate the moving sum, rate of change, and lag correlation using an index pointer indicating the current storage position of the circular ring buffer memory.

[0014] In addition, the control unit may be configured to suppress false alarms caused by simple switching noise, welding light, temporary external gas inflow, or humidity changes by calculating the time delay and slope relationship between the leading increase pattern of the arc generation quantity and the lagging increase pattern of the CO concentration, and by applying a correction coefficient according to temperature and humidity conditions to calculate the fire risk.

[0015] In addition, the arc sensor is configured to detect ultraviolet wavelengths in the range of 185 nm to 260 nm and may include an optical filter or a light-transmitting window that suppresses a response to visible light and infrared bands.

[0016] Additionally, if the above movement total exceeds a first set value, a warning (Light Alarm) is generated to flash the status indicator lamp, and if the above movement total exceeds a second set value which is higher than the first set value or if the above temporal correlation sensitivity exceeds a reference value, a severe (Heavy Alarm) is generated to keep the status indicator lamp constantly lit, and the control unit may be configured to operate in either an automatic return (Auto) mode in which the alarm is automatically released when the dangerous situation is resolved, or a manual hold (Latch) mode in which the alarm is released only upon receiving a reset command from the RS485 communication unit even after the dangerous situation is resolved.

[0017] In addition, the control unit may be configured to correct the CO concentration value or the CO concentration increase rate using the temperature and humidity values, and to reflect the corrected CO concentration value or the corrected CO concentration increase rate in the temporal correlation sensitivity or fire risk calculation.

[0018] In addition, the control unit may calculate a sensor abnormality state when the output value of the electrochemical CO detection sensor, the temperature and humidity sensor, or the arc sensor deviates from the normal range or does not change for a certain period of time, and transmit the sensor abnormality state to the external master system, distinguishing it from the fire alarm. Effects of the invention

[0019] The present invention has a significant effect of preventing fires by installing Arc (Ultra Violet), CO Gas, and ambient temperature and humidity sensing sensors to detect in advance the main causes of electrical fires, such as short circuits, overloads, equipment overheating, leakage circuits, and poor contact.

[0020] In addition, unlike arc-only or CO-only detection methods, the present invention identifies precursor phenomena characteristic of electrical coating degradation and carbonization by utilizing the temporal causal relationship between arc-preceding events and CO-following events, thereby having the effect of reducing the false alarm rate while maintaining the performance of early detection of actual fire hazards.

[0021] In addition, since the present invention uses a circular ring buffer and an index pointer-based data management structure, it can stably maintain the sequence of time-series data in the last 60 seconds even in a low-speed RS485 / Modbus communication environment, and has the effect of enabling an external master system to read fire judgment basis data without omission or duplication. Brief explanation of the drawing

[0022] FIG. 1 is an external photograph of the electric fire detection alarm of the present invention. FIG. 2a is a front view of the electric fire detection alarm of the present invention, FIG. 2b is a side view of the electric fire detection alarm of the present invention. Figure 3 is a light transmission graph of the electric fire detection alarm PD sensor of the present invention. Figure 4 is a sensitivity graph of the CO gas sensor of the electric fire detection alarm PD sensor of the present invention. FIG. 5 is a correction relative temperature graph of the electric fire detection alarm of the present invention. FIG. 6 is a correction relative humidity graph of the electric fire detection alarm of the present invention. FIG. 7 is a diagram showing the communication address setting of a DIP switch for an electric fire detection alarm of the present invention. FIG. 8 is a diagram showing the communication speed setting of a DIP switch for an electric fire detection alarm of the present invention. FIG. 9 is a correction relative humidity graph of the electric fire detection alarm of the present invention. FIG. 10 is an installation diagram of the electric fire detection alarm of the present invention. FIG. 11 is an installation photograph of the electric fire detection alarm of the present invention. Specific details for implementing the invention

[0023] The present invention relates to an electric fire detection alarm and comprises a composite sensor module for detecting signs of an electric fire inside a distribution panel, the module including: an electrochemical CO detection sensor for detecting carbon monoxide (CO) gas generated by the carbonization of electrical insulation inside the distribution panel; a temperature and humidity sensor; an arc sensor for detecting ultraviolet (UV) wavelengths resulting from the occurrence of an arc inside the distribution panel; a control unit for receiving measurement values ​​from each of the sensors, calculating a fire risk level, and controlling the generation of an alarm; and an RS485 communication unit for transmitting the calculated data of the control unit to an external master system. The control unit is characterized by updating the number of arc occurrences detected by the arc sensor at regular time intervals to calculate the accumulated quantity during the previous period using a moving sum method, and generating a fire alarm when the accumulated quantity exceeds a predetermined set value.

[0024] In addition, the control unit is characterized by generating a warning (Light Alarm) that flashes a status indicator lamp when the accumulated arc quantity exceeds a first set value, and generating a severe (Heavy Alarm) that keeps the status indicator lamp constantly lit when it exceeds a second set value higher than the first set value, and is configured to operate in either an automatic return (Auto) mode in which the alarm is automatically released when the dangerous situation is resolved, or a manual hold (Latch) mode in which the alarm is released only upon receiving a reset command from the communication unit even after the dangerous situation is resolved.

[0025] In addition, it further includes a status output terminal for connecting to an external control power supply (DC 24V) and an external alarm device; and a plurality of dip switches for controlling hardware settings, wherein among the plurality of dip switches, the first to fifth dip switches set the communication address of the RS485 communication unit, and the sixth dip switch sets the communication speed (baud rate) to either 9,600 bps or 19,200 bps.

[0026] In addition, the control unit converts the arc occurrence quantity data from the arc detection sensor in 1-second intervals and the carbon monoxide concentration data from the electrochemical CO detection sensor into respective time-series input data vectors and synchronizes them in real time, and includes a fire judgment artificial intelligence model based on LSTM (Long Short-Term Memory) that is installed within the control unit to preserve information on past states and learn the long-short-term dependencies of the time-series data, and the LSTM-based fire judgment artificial intelligence model calculates the temporal correlation sensitivity between the arc leading pattern, which is the trend of change in the arc occurrence quantity over the last 60 seconds accumulated in the circular ring buffer memory, and the gas lagging pattern of carbon monoxide concentration that continuously rises with a delay in occurrence time lag corresponding thereto, and controls the generation of a final fire alarm by inferring in real time whether the input time-series data matches the transient fire characteristic of an actual electrical coating carbonization fire, rather than simple line noise or temporary gas inflow.

[0028] The present invention will be described in detail with reference to the attached drawings as follows.

[0030] The Arc (Ultra Violet), CO Gas, and ambient temperature and humidity sensing sensors, which are the main components of the present invention, will be described.

[0031] If insulation damage or carbonization occurs in the electrical sheathing inside a switchboard, it is accompanied by the generation of carbon monoxide (CO) prior to ignition. The TJ-ACTH sensor of the present invention measures the gas concentration inside the switchboard using an electrochemical carbon monoxide detection sensor.

[0033] I will explain the ambient temperature and humidity sensor.

[0034] Conventional contact-type temperature sensors detect temperature by making direct contact with the distribution panel terminal; therefore, over time, the temperature sensor loses its function due to the deterioration of the sensor contact area.

[0035] On the other hand, the TJ-ACTH sensor of the present invention is a digital output sensor and is a third-generation sensor that integrates a temperature (humidity) sensor, a 14-bit A / D converter, an EPROM storing a corrected memory value, and a digital 2-wire interface on a single CMOS chip.

[0036] The specifications and technical specifications of the alarm of the present invention are as follows.

[0037] General Specifications

[0038]

[0039] The technical specifications of the present invention are as shown in the following table.

[0040]

[0041]

[0042] The operation settings of the present invention are as follows.

[0043] Connect the control power (DC 24V) and RS485 communication to the input / output terminals. By configuring a separate circuit using the Status Output, a buzzer or lamp for the alarm can be used. To set the communication address, configure the 1st through 5th DIP switches on the right side in the following manner. The 6th DIP switch sets the communication speed (9,600 bps when Off / 19,200 bps when On). When the control power is supplied, the OP lamp lights up, and the RX / TX lamps blink during communication. The Arc lamp blinks during a Light alarm and lights up during a Heavy alarm. The CO and TEMP lamps light up when the CO gas sensor operates and the upper temperature limit sensor operates, respectively. If the reset method of the operation lamp is set to Auto, it automatically returns to its original state upon alarm release; however, if the reset method is set to Latch, it returns only when an Alarm Reset is executed via upper-level communication.

[0045] And when resetting the accumulated number of arcs, you must also execute the arc alarm reset. Depending on the accumulated number of arcs at address 40102, the PD LED blinks when the Light Alarm quantity is exceeded, and the PD LED lights up when the Heavy Alarm quantity is exceeded. Conversely, when the number is less than or equal to the Light Alarm quantity, the PD LED is automatically turned off.

[0046] In addition, the accumulated arc quantity is updated and displayed at 1-second intervals based on the index value. For example, if the current index position value of address 40108 is 11, the accumulated arc quantity of address 40102 is the total quantity for one minute from the most recent value, index 11, to 10, 9, 8, … 1, 60, 59, … 12.

[0047] The table below shows RS485 address mapping data.

[0048]

[0050] The following shows the Modbus communication data for the Arc & CO gas and temperature & humidity sensors.

[0051]

[0052]

[0054] The specific details and embodiments of the present invention will be described below.

[0056] The electric fire detection alarm according to the present invention includes a control unit that performs a time-series artificial intelligence algorithm to fundamentally block false alarms caused by external interference factors, such as line noise and welding operations occurring inside the distribution panel, and to precisely determine only fire signs resulting from the actual deterioration and carbonization of the electrical insulation. Below, the time-series data processing of the control unit, which is a core component of the present invention, and the driving mechanism of the LSTM (Long Short-Term Memory)-based fire determination artificial intelligence model will be described in detail.

[0058] (1) Time series input data vector conversion and real-time synchronization fury

[0059] The control unit of the present invention collects in real time 'data on the number of arc occurrences per second' measured by an arc detection sensor and 'carbon monoxide (CO) concentration data' measured by an electrochemical CO detection sensor. At this time, the two data, having different physical quantities and scales, undergo normalization processing within the control unit and are then converted into time-series input data vectors so that they can be matched on the same time axis. The control unit synchronizes the two converted data vectors in real time based on timestamps and transmits them to the input layer of an artificial intelligence model.

[0061] (2) Structure and characteristics of an LSTM-based fire judgment AI model

[0062] Electrical fires within a distribution panel are not temporary, one-off events, but rather possess transitional characteristics in which physical quantities change organically over time. To analyze this, the control unit of the present invention is equipped with a fire judgment artificial intelligence model based on Long Short-Term Memory (LSTM) of the Recurrent Neural Network (RNN) family, capable of flexibly learning and remembering long-term and short-term dependencies of time-series data. The LSTM model continuously calculates the correlation with currently input sensor data while preserving feature points of prior data accumulated over the past 60 seconds without loss through a cell structure composed of an input gate, a forget gate, and an output gate.

[0064] (3) Calculation of temporal correlation sensitivity between arc leading pattern and gas lagging pattern

[0065] When damage or carbonization of the insulation sheath occurs in actual electrical wiring, a unique time-lag causal relationship arises in which a physical arc occurs discontinuously first (leading event), followed by the intensification of deterioration, causing the sheath to burn and carbon monoxide (CO) gas to continuously accumulate and be emitted (lagging event). The LSTM model of the present invention extracts an "arc leading pattern," which is a trend of changes in the number of arc occurrences over the past 60 seconds accumulated and updated in real-time in a circular ring buffer memory, and a "gas lagging pattern," which is a carbon monoxide concentration that rises continuously with a certain time lag corresponding to the arc leading pattern. Subsequently, it calculates "temporal correlation sensitivity" by fusing the time gap between the two patterns and the slope of the data fluctuation. Unlike conventional methods that simply check whether two sensor values ​​are simultaneously high (AND condition), this is a step that precisely verifies whether the sequence of occurrence of the two sensor signals and the trend of the waveforms correspond to the physical electrical fire mechanism.

[0067] (4) Inference of transitional precursor characteristics and alarm control

[0068] The control unit infers in real time whether the calculated temporal correlation sensitivity and waveform analysis results correspond to simple line switching noise (only an arc occurs momentarily and no gas is generated) or exhaust gas temporarily introduced from the external atmosphere (only CO gas rises and no arc occurs), or to the 'transient fire characteristic of actual electrical coating carbonization fire'. When the fire probability result calculated in the final output layer of the LSTM model exceeds a preset confidence threshold and is finally confirmed to be in an 'actual fire precursor state', the control unit immediately performs a series of fire alarm control processes, such as turning on or flashing the status indicator lamp (LED), transmitting fire alarm data to an upper master system through the RS485 communication interface, or driving an external buzzer through the status output terminal.

[0070] Meanwhile, although it is a known fact that the present invention uses the Modbus protocol, the originality of the present invention lies in a scheduling technique that synchronizes upper and lower systems without breaking the 'temporal continuity' of data for high-capacity time-series AI (LSTM) computation in a low-speed RS485 environment with limited bandwidth (9,600 to 19,200 bps). While general Modbus transmission does not pose a problem for simple numerical verification even if data is lost or out of order, the LSTM model has a fatal problem in that the fire inference result is completely distorted if the data order is twisted by even one second.

[0071] Accordingly, the present invention provides an optimization algorithm that organically links physical hardware specifications (DIP switch speed settings) and a software register map so that jitter or packet disconnection does not occur when the master system reads the 60-second accumulated data at address 40102, using the index pointer information at address 40108 as a milestone.

[0073] In addition, the communication address and communication baud rate of the control unit are set by a combination of contacts of a physical dip switch module exposed on one side of the housing, or are variably set by software-recording a specific Modbus setting packet transmitted from an external master system through the RS485 communication line to a non-volatile memory (EEPROM) within the control unit.

[0075] (5) Calculation of composite risk based on temperature and humidity correction

[0076] The output stability and gas diffusion response of the CO detection sensor inside the distribution panel may vary depending on ambient temperature or humidity conditions. Therefore, the control unit of the present invention can calculate a correction coefficient for the CO concentration value or the rate of change in CO concentration using the measured values ​​of the temperature and humidity sensors, and reflect the corrected CO concentration value in the calculation of fire risk. For example, if the temperature value deviates from a predetermined reference temperature range or the relative humidity deviates from a predetermined reference humidity range, the control unit can suppress false alarms caused by environmental changes by adjusting the CO concentration threshold or the CO increase rate threshold upward or downward.

[0077] The above composite risk level may be calculated by weighted summing the moving sum of the number of arcs generated, the instantaneous peak value of the number of arcs generated, the CO concentration value, the unit-time increase rate of the CO concentration, the temperature value, the temperature increase rate, and the humidity correction coefficient, or by a rule-based judgment unit or an artificial intelligence judgment unit that uses the above values ​​as input values. In this case, the control unit may be configured to increase the risk weight when the number of arcs generated increases first and then the CO concentration or the CO increase rate increases within a certain delay time, and to decrease the risk weight when only the CO concentration increases alone or only the number of arcs increases alone.

[0079] (6) Step-by-step alarm and alarm confirmation conditions

[0080] The control unit of the present invention can classify fire risk levels into multiple stages. For example, the control unit may output a state requiring inspection or a first warning state when the moving sum of the number of arc occurrences exceeds a first arc threshold but the CO concentration or CO growth rate is below a reference value, and may output a state confirming fire precursor or a second severe warning state when the moving sum of the number of arc occurrences exceeds a second arc threshold or the temporal correlation sensitivity between the arc leading pattern and the CO trailing pattern is greater than or equal to a reference value. In addition, if the temperature value or the temperature growth rate exceeds a reference value, the fire risk level may be further increased to simultaneously determine poor contact or heat accumulation due to overload.

[0081] The above temporal correlation sensitivity can be calculated from a judgment variable including at least one of the following: a delay time between the arc generation quantity time series and the CO concentration time series, the slope of the CO concentration rise within the delay time, the cumulative number of arc generation quantities, and the duration of the CO concentration rise. Preferably, the control unit uses a circular ring buffer including 60 storage slots that store the arc generation quantity of the last 60 seconds and an index pointer that indicates the current storage slot, and calculates the temporal correlation sensitivity by sequentially rearranging the data from the most recent data to the past data based on the index pointer.

[0083] (7) Sensor placement and optical selectivity

[0084] The arc sensor may be positioned on the back of the light-transmitting window of the housing so as to face a terminal block, circuit breaker, busbar, cable connection part, or wire sheath inside the switchboard. The light-transmitting window may include a filter or optical structure that passes ultraviolet wavelength bands of 185 nm to 260 nm and suppresses the influence of visible light and infrared bands. Accordingly, the influence of sunlight, fluorescent lamps, LED lighting, or external lighting of the switchboard is reduced, and ultraviolet signals accompanying electrical arc discharge can be selectively detected.

[0085] A CO detection sensor can be positioned in a ventilation section where internal air enters the switchboard or in a gas diffusion path inside the housing, and temperature and humidity sensors can be positioned adjacent to the CO detection sensor to measure environmental conditions around the sensor in real time. This arrangement provides reference values ​​for CO concentration correction and sensor malfunction detection, and forms a complex detection structure that utilizes the physical correlation between sensor outputs rather than a simple parallel arrangement of sensors.

[0087] (8) Communication synchronization and remote maintenance

[0088] The control unit can transmit the alarm status, the number of accumulated arcs over the past 60 seconds, the number of arcs generated by each index, CO concentration, temperature, humidity, current index position, communication address, communication speed, and reset status to an external master system via the RS485 communication unit. The external master system can reconstruct the chronological order of data for each index based on the current index position to analyze the sequence and cause of fire precursor events occurring in multiple distribution panels.

[0089] In addition, the control unit can calculate a sensor abnormality if the sensor output deviates from a predetermined normal range or remains unchanged for a certain period, and transmit this sensor abnormality to an external master system, distinguishing it from an alarm state. In this case, maintenance personnel can differentiate between an actual fire precursor alarm and a sensor failure or communication anomaly, thereby reducing unnecessary on-site dispatches and increasing the reliability of the switchboard monitoring system.

[0091] (9) Example

[0092] In one embodiment, the control unit counts the number of arcs generated at a 1-second interval and sequentially stores the number of arcs generated in each 1-second interval in a circular ring buffer having 60 index storage areas. The control unit updates the current index position every 1 second and calculates the sum of the number of arcs generated over the last 60 seconds based on the current index position. At this time, the CO concentration is acquired at the same 1-second interval or multiple-second intervals and is aligned as time-series data on the same time axis.

[0093] In another embodiment, if the number of arcs exceeds a first set value but the CO concentration growth rate is less than a reference value, a Light Alarm is output, and if the number of arcs exceeds a second set value or if the CO concentration growth rate becomes greater than or equal to the reference value within a predetermined delay time after arc generation, a Heavy Alarm is output. On the other hand, if only the CO concentration rises due to external exhaust gas inflow, etc., and there is no arc leading pattern, it can be processed as a CO-only alarm or inspection alarm, and the Heavy Alarm can be withheld. Explanation of the symbols

[0095] 1 : Power lamp 2 : Communication lamp 3 : Operation lamp 4: Partial discharge detection sensor 5: CO gas detection sensor 6 : Temperature detection sensor 7 : Debug port 8 : Communication ID set, Communication speed set

Claims

Claim 1 An electric fire detection alarm for detecting signs of an electric fire inside a switchboard, comprising: an electrochemical CO detection sensor for detecting carbon monoxide (CO) gas generated by the carbonization of the electrical coating inside the switchboard; a temperature and humidity sensor for detecting the temperature and humidity inside the switchboard; an arc sensor for detecting ultraviolet wavelengths resulting from the occurrence of an arc inside the switchboard; and a control unit for acquiring the number of arc occurrences per unit time from the arc sensor, acquiring a CO concentration value from the electrochemical CO detection sensor, and acquiring a temperature value and a humidity value from the temperature and humidity sensor. The electric fire detection alarm comprises: an RS485 communication unit that transmits the operation data and alarm status of the control unit to an external master system; wherein the control unit stores the number of arcs generated per unit time during a recent period in a circular ring buffer memory having a plurality of storage slots, calculates a moving sum of the number of arcs generated per unit time based on an index pointer indicating the current storage slot, calculates a temporal correlation sensitivity using the time delay and slope relationship between the leading increase pattern of the number of arcs generated per unit time and the trailing increase pattern of the CO concentration value or the CO concentration increase rate, and generates a fire alarm while excluding false alarms caused by simple line noise or temporary external gas inflow based on the moving sum and the temporal correlation sensitivity, wherein the control unit calculates a sensor abnormality state when the output value of the electrochemical CO detection sensor, the temperature and humidity sensor, or the arc sensor deviates from the normal range or does not change for a certain period of time, and transmits the sensor abnormality state to the external master system, distinguishing it from the fire alarm. Claim 2 An electric fire detection alarm according to claim 1, wherein the arc sensor is configured to detect an ultraviolet wavelength band of 185 nm to 260 nm and includes an optical filter or a light-transmitting window that suppresses a response to visible light and infrared bands. Claim 3 An electric fire detection alarm according to claim 1, wherein the control unit generates a warning (Light Alarm) that flashes a status indicator lamp when the total movement exceeds a first set value, and generates a severe (Heavy Alarm) that keeps the status indicator lamp lit when the total movement exceeds a second set value higher than the first set value or when the temporal correlation sensitivity exceeds a reference value. Claim 4 An electric fire detection alarm according to paragraph 3, characterized in that the control unit is configured to operate in either an automatic return (Auto) mode, in which the alarm is automatically released when a dangerous situation is released, or a manual hold (Latch) mode, in which the alarm is released only after receiving a reset command from the RS485 communication unit even when a dangerous situation is released. Claim 5 An electric fire detection alarm according to claim 1, wherein the control unit corrects the CO concentration value or the CO concentration increase rate using the temperature value and humidity value, and reflects the corrected CO concentration value or the corrected CO concentration increase rate in the temporal correlation sensitivity or fire risk calculation. Claim 6 delete

Images