Flame detection system and method thereof

Abstract

A flame detection system includes: at least one flame detector for detecting one or more radiations within a field of view (FOV); and at least one analog-to-digital converter for converting the one or more signals corresponding to the detected one or more radiations into one or more analog-to-digital converted (ADC) signals. Further, the at least one processor is configured to receive ADC signals corresponding to the detected one or more radiations, and determine a plurality of threshold conditions of a reflected torch chimney associated with the one or more ADC signals. The at least one processor is configured to determine a parameter associated with the one or more ADC signals, compare the parameter to each of a plurality of threshold conditions, and determine, based at least on the comparison, whether the one or more ADC signals are a flare chimney indicative of a flame or a reflected flare chimney.

Description

Technical Field

[0001] The exemplary embodiments disclosed herein relate generally to flame detection systems, and more specifically to flame detection systems for determining false alarms during flame detection. Background Technology

[0002] Flame detectors are safety devices designed to identify flames produced by the combustion of different fuel sources in various industries, including oil and gas, chemical processing, and manufacturing. However, false alarms pose a significant challenge to the efficiency and effectiveness of flame detectors. False alarms lead to unnecessary downtime, costly maintenance, and operational inefficiency. Furthermore, flame detectors often struggle to distinguish between actual fire events and false triggers caused by environmental factors such as arc welding, reflected sunlight, LED or halogen lamps, heaters, and reflecting torches / chimneys. Flame detectors equipped with advanced false alarm processing capabilities can effectively differentiate between various sources of false alarms. However, distinguishing between reflected torches / chimneys and actual flames remains a challenge. The wavelengths and intensities of reflected torches are very similar to those of actual flames, making it difficult for existing flame detectors to accurately distinguish between the two.

[0003] The inventors have identified numerous areas for improvement in the prior art and methods, which are the subject of the embodiments described herein. Through effort, ingenuity, and innovation, many of these deficiencies, challenges, and problems have been addressed by developing solutions included in the embodiments of this disclosure, some examples of which are described in detail herein. Summary of the Invention

[0004] The following is a simplified overview to provide a basic understanding of some aspects of this disclosure. This invention is not an exhaustive summary and is neither intended to identify key or essential elements nor to describe a range of such elements. Its purpose is to provide, in a simplified form, some concepts of the described features as a prelude to the more detailed description provided later.

[0005] In an example embodiment, a flame detection system is disclosed. The flame detection system includes at least one flame detector configured to detect one or more types of radiation within a field of view (FOV). Furthermore, at least one analog-to-digital converter (ADC) is configured to convert one or more signals corresponding to the detected one or more types of radiation into one or more analog-to-digital converter (ADC) signals. Additionally, at least one processor is operatively coupled to the at least one flame detector. The at least one processor is configured to receive one or more ADC signals from the at least one flame detector. Furthermore, the at least one processor is configured to determine multiple threshold conditions for a reflected torch chimney associated with the one or more ADC signals. Furthermore, the at least one processor is configured to determine parameters associated with the one or more ADC signals. The parameters are determined based at least on spectral density and frequency ratio. Furthermore, the at least one processor is configured to compare the parameters with multiple threshold conditions of the one or more ADC signals, wherein the parameters satisfy the multiple threshold conditions. Subsequently, the at least one processor is configured to determine, at least based on the comparison, whether the one or more ADC signals indicate a flame or a reflected torch.

[0006] In some implementations, the parameters correspond at least to the skewness and kurtosis ratio (SK ratio) of one or more ADC signals. In some implementations, at least one processor is configured to compare the skewness and kurtosis ratio with multiple threshold conditions using a Kolmogorov-Smirnov test.

[0007] In some embodiments, at least one processor is configured to compare a parameter with each of a plurality of threshold conditions for one or more ADC signals to determine whether the parameter associated with the one or more ADC signals satisfies each of the determined plurality of threshold conditions. Furthermore, at least one processor is configured to determine whether the one or more ADC signals indicate a flame when it is determined that the parameter associated with the one or more ADC signals does not satisfy each of the determined plurality of threshold conditions. Additionally, at least one processor is configured to determine whether the one or more ADC signals indicate a reflected torch when it is determined that the parameter associated with the one or more ADC signals satisfies each of the determined plurality of threshold conditions.

[0008] In some implementations, at least one processor is configured to determine multiple threshold conditions associated with one or more ADC signals, based at least on flag conditions associated with those signals. Furthermore, the flag conditions correspond to one or more fluctuations and modulations in the one or more ADC signals received by the at least one processor over a certain time period.

[0009] In some implementations, at least one processor is configured to generate a signal in response to determining that one or more ADC signals indicate a flame, and to send the signal to a communication device for warning a user.

[0010] In some implementations, at least one flame detector includes at least one of an infrared (IR) sensor, a photodiode, or a combination of an IR sensor and a photodiode.

[0011] In another example implementation, a method is disclosed. The method includes the steps of: receiving, via at least one processor, one or more analog-to-digital converter (ADC) signals corresponding to one or more types of radiation detected within the field of view (FOV) of at least one flame detector; determining, via at least one processor, multiple threshold conditions for a reflected torch chimney associated with the one or more ADC signals; determining, via at least one processor, parameters associated with the one or more ADC signals, wherein the parameters are determined at least based on spectral density and frequency ratio; comparing, via at least one processor, the parameters to the multiple threshold conditions of the one or more ADC signals, wherein the parameters satisfy the multiple threshold conditions; and determining, via at least one processor, whether the one or more ADC signals indicate a flame or a reflected torch, based at least on the comparison.

[0012] The above description of the invention is provided merely to outline some exemplary embodiments to provide a basic understanding of some aspects of the invention. Therefore, it should be understood that the above embodiments are merely illustrative and should not be construed as limiting the scope or nature of the invention in any way. It should be understood that, in addition to those described herein, the scope of the invention covers many possible embodiments, some of which will be further described below. Attached Figure Description

[0013] Therefore, some exemplary embodiments of this disclosure have been described in general terms, and reference will be made below to the accompanying drawings, which are not necessarily drawn to scale, and in which:

[0014] Figure 1 A block diagram of a flame detection system according to an example embodiment of the present disclosure is shown;

[0015] Figure 2 A graphical representation of the spectrum of one or more radiations according to an example embodiment of the present disclosure is illustrated;

[0016] Figure 3 An image representation of the field of view (FOV) of at least one flame detector according to an example embodiment of the present disclosure is illustrated;

[0017] Figure 4 An exemplary scenario of a flame detection system according to an example embodiment of the present disclosure is illustrated;

[0018] Figure 5 A table illustrating multiple threshold conditions according to example embodiments of this disclosure is provided;

[0019] Figures 6A to 6B A table illustrating the verification results of a flame detection system according to an example embodiment of the present disclosure is provided; and

[0020] Figure 7 A flowchart illustrating a method for flame detection according to an example embodiment of the present disclosure is shown. Detailed Implementation

[0021] Some embodiments will now be described more fully below with reference to the accompanying drawings, which illustrate some embodiments, but not all embodiments. In fact, various embodiments can be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will meet applicable legal requirements.

[0022] The components illustrated in the accompanying drawings represent components that may or may not be present in the various embodiments of the invention described herein, such that embodiments may include fewer or more components than those shown in the figures without departing from the scope of the invention. Some components may be omitted from one or more figures, or shown with dashed lines to make the components below visible.

[0023] As used herein, the term “comprising” means including but not limited to, and should be interpreted in the manner in which it is typically used in the patent context. The use of broader terms such as “comprising,” “including,” and “having” should be understood to provide support for narrower terms such as “consisting of,” “substantially composed of,” and “substantially constituted by.”

[0024] The phrases “in various embodiments,” “in one embodiment,” “according to one embodiment,” “in some embodiments,” etc., generally mean that the specific feature, structure, or characteristic following the phrase may be included in at least one embodiment of this disclosure, and may be included in more than one embodiment of this disclosure (importantly, such phrases do not necessarily refer to the same embodiment).

[0025] As used herein, the terms “example” or “exemplary” mean “serving as an example, instance, or illustration.” Any specific implementation described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other specific implementations.

[0026] If this specification states that a component or feature is "may", "can", "may", "should", "will", "preferably", "possibly", "usually", "optionally", "for example", "often", or "maybe" (or other such language) included or has a characteristic, then the specific component or feature does not need to be included or have that characteristic. Such components or features may be optionally included in some embodiments or may be excluded.

[0027] This disclosure provides various embodiments of a flame detection system and method thereof. Embodiments may include at least one flame detector. Embodiments may be configured to detect one or more types of radiation within a field of view (FOV). Embodiments may include at least one analog-to-digital converter (ADC) configured to convert one or more signals corresponding to the detected one or more types of radiation into one or more analog-to-digital converter (ADC) signals. Embodiments may include at least one processor operatively coupled to at least one flame detector. Embodiments may be configured to receive one or more ADC signals from at least one flame detector. Embodiments may be configured to determine multiple threshold conditions for a reflected torch chimney associated with one or more ADC signals. Embodiments may be configured to determine parameters associated with one or more ADC signals. Embodiments may be configured to compare the parameters with multiple threshold conditions of one or more ADC signals, wherein the parameters satisfy the multiple threshold conditions. Embodiments may be configured to determine, at least based on the comparison, whether one or more ADC signals indicate a flame or a reflected torch.

[0028] Figure 1 A block diagram of a flame detection system 100 according to an example embodiment of the present disclosure is shown.

[0029] The flame detection system 100 may include at least one flame detector 104, an analog-to-digital converter 106, at least one processor 108, a memory 110, input / output circuitry 112, a communication device 114, and a communication circuitry 116. In some embodiments, at least one flame detector 104 may be configured to detect one or more types of radiation within a field of view (FOV) 300. Figure 3 (As shown). One or more types of radiation can be emitted from the flame source 102 within the FOV 300. In some embodiments, the flame source 102 may correspond to at least a source of hydrogen, hydrocarbon gas, methane gas, or other combustible fuel. In some embodiments, the flame source 102 may emit one or more types of radiation during combustion. Furthermore, one or more types of radiation can be detected in the form of one or more infrared (IR) signals, one or more optical signals, or one or more thermal signals.

[0030] In some embodiments, at least one flame detector 104 may be configured to detect one or more types of radiation within the FOV 300 and convert the detected radiation into one or more analog-to-digital converter (ADC) signals. In one example, at least one flame detector 104 includes at least one of an infrared (IR) sensor, a photodiode, or a combination of an IR sensor and a photodiode. In some embodiments, the IR sensor and photodiode may be configured to capture one or more types of radiation. In some embodiments, the IR sensor and photodiode may be configured to capture one or more types of radiation in the form of one or more analog signals. Furthermore, at least one flame detector 104 may be made of one or more active and passive electronic components that enable at least one flame detector 104 to detect one or more types of radiation.

[0031] In some embodiments, at least one flame detector 104 may be integrated with an analog-to-digital converter (ADC) 106. In some embodiments, the ADC may be configured to convert one or more analog signals corresponding to one or more types of radiation into one or more digital signals. Furthermore, the one or more digital signals may also be referred to as one or more ADC signals, such as one or more ADC counts. In some embodiments, at least one flame detector 104 may be operatively coupled to at least one processor 108, such that the ADC may be configured to feed one or more ADC signals to at least one processor 108.

[0032] In some embodiments, at least one processor 108 may be configured to receive one or more ADC signals from at least one flame detector 104. Furthermore, at least one processor 108 may be configured to determine multiple threshold conditions for a reflected flare chimney associated with the one or more ADC signals. The threshold conditions correspond to specific threshold conditions tailored for the flame and the reflected flare, and are represented as one or more columns, such as characteristic only fire 502, SK ratio average of characteristic only fire 502 504, characteristic reflected flare chimney 506, and SK ratio average of characteristic reflected flare chimney 506 508 (e.g.,...). Figure 5 (As shown). In some implementations, threshold conditions can be calculated using the Kolmogorov-Smirnov test (KS test) and the p-value test. In one example, if the p-value is less than the significance level (α), the KS statistic is considered a threshold condition for one or more characteristics of the flame and the reflected torch. Furthermore, at least one processor 108 can be configured to determine parameters associated with one or more ADC signals. The parameters can be determined based at least on spectral density and frequency ratio. In some implementations, the parameters can at least correspond to the skewness and kurtosis ratio (SK ratio) of one or more ADC signals.

[0033] In one example, the frequency domain ratio derived from one or more ADC signals (e.g., one or more ADC signals of AC / DC WB, LB, and NB) includes PSDLoBandAWbLb: the power spectral density (PSD) ratio, which is the ratio of the sum of the low-frequency band [2Hz–5Hz] of the wideband signal (WBAC) to the sum of the low-frequency band [2Hz–5Hz] of the narrowband signal (LBAC). In some embodiments, the power spectral density (PSD) is a measure of the power distributed across one or more ADC signals at different frequencies. In some embodiments, WBAC typically refers to one or more ADC signals encompassing a wide frequency range. In some embodiments, LBAC refers to one or more ADC signals limited to a narrower frequency range.

[0034] In some implementations, the low-frequency band summation (the sum of PSD values ​​over 2Hz to 5Hz) corresponds to summing the PSD values ​​over a specific frequency range, in this case, 2Hz to 5Hz. The low-frequency band summation focuses on the low-frequency components of the signal. In some implementations, the ratio of the low-frequency band PSD summation of the WBAC to the low-frequency band PSD summation of the LBAC compares the power content in the low-frequency range of the broadband signal with the power content in the low-frequency range of the narrowband signal. The numerator involves calculating the sum of PSD values ​​in the low-frequency range of the broadband signal. The denominator involves calculating the sum of PSD values ​​in the same low-frequency range of the narrowband signal.

[0035] In some embodiments, at least one processor 108 may be configured to compare parameters against each of a plurality of threshold conditions for one or more ADC signals. Furthermore, at least one processor 108 may be configured to compare skewness and kurtosis ratio against the plurality of threshold conditions using a Kolmogorov-Smirnov test. In some embodiments, the distribution of skewness and kurtosis ratio can be compared for various frequency domain features between flame and reflected torch scenes using a Kolmogorov-Smirnov test (KS test). The KS test is used to compare the distributions of two samples. The KS test checks whether the distribution of skewness / kurtosis ratio for each feature is significantly different between flame and reflected torch scenes. Furthermore, a significance level (α) set to 0.01 can be defined. The significance level means that if the p-value from the KS test is less than 0.01, the difference between the distributions is considered statistically significant. In one example, the results could correspond to: Feature: PSA_LbLo_LbHi, KS statistic: 0.09, P-value: 0.0, indicating that the test is significant (different distribution); Feature: PSA_WbLo_LbHi, KS statistic: 0.14, P-value: 0.0, indicating that the test is significant (different distribution).

[0036] In some embodiments, at least one processor 108 may be configured to determine whether one or more ADC signals indicate a flame or a reflecting torch, at least based on comparison. In some embodiments, at least one processor 108 may be configured to train an ML model, at least based on multiple threshold conditions extracted from one or more ADC signals over a time period. Furthermore, at least one processor 108 may be configured to deploy the trained ML model to determine whether it is a flame or a reflecting torch. The trained ML model may include an aggregate simulation of multiple models trained by at least one processor 108 over a time period.

[0037] In some embodiments, at least one processor 108 may be configured to determine multiple threshold conditions associated with one or more ADC signals, based at least on flag conditions associated with one or more ADC signals. In some embodiments, the threshold conditions may be calculated using the Kolmogorov-Smirnov test (KS test) and a p-value test. In one example, if the p-value is less than the significance level (α), the KS statistic is considered a threshold condition for one or more characteristics of the flame and the reflected torch. In some embodiments, at least one processor 108 may be configured to employ flag conditions based on one or more fluctuations and modulations of the one or more ADC signals received by at least one processor 108 over a time period.

[0038] At least one processor 108 may be configured to detect fluctuations or modulations in a plurality of threshold conditions determined based on one or more ADC signals to determine whether the one or more ADC signals indicate a flame as a fire, an intentional flame, or a reflected torch. In some embodiments, at least one processor 108 may be configured to generate a signal in response to the determined one or more ADC signals indicating a flame. Thereafter, at least one processor 108 may be configured to transmit the signal to communication device 114 for warning a user. In some embodiments, at least one processor 108 may transmit the signal to communication device 114 via input / output circuitry 112 and communication circuitry 116.

[0039] In some embodiments, at least one processor 108 may include suitable logic components, circuitry, and / or interfaces operable to execute one or more instructions stored in memory 110 to perform predetermined operations. In one embodiment, at least one processor 108 may be configured to decode and execute any instructions received from one or more other electronic devices or servers. At least one processor 108 may be configured to execute one or more computer-readable program instructions, such as program instructions for performing any of the functions described herein. Furthermore, at least one processor 108 may be implemented using one or more processor technologies known in the art. Examples of at least one processor 108 include, but are not limited to, one or more general-purpose processors (e.g., or Advanced Micro (AMD) microprocessors) and / or one or more dedicated processors (e.g., digital signal processors or System-on-a-Chip (SOC) Field Programmable Gate Array (FPGA) processor.

[0040] In some embodiments, at least one processor 108 can validate and test multiple threshold conditions via an ML model by comparing multiple threshold conditions with multiple scenarios based on one or more ADC signals over a time period. Furthermore, the multiple scenarios may correspond to instances of flames generated from different flame sources 102 and torch reflections. Additionally, at least one processor 108 can be configured via a trained ML model to generate one or more results based on at least multiple characteristics of the validation and testing. In some embodiments, one or more results may indicate between flames or between reflecting torches. In some embodiments, the results of the validation and testing may be stored in memory 110. Furthermore, in some embodiments, at least one processor 108 can be configured to train the ML model based at least on multiple threshold conditions extracted from one or more ADC signals over a time period. Furthermore, at least one processor 108 can be configured to deploy the trained ML model to determine flames or reflecting torches.

[0041] In some embodiments, memory 110 may be configured to store a set of instructions and data that can be executed by one or more processors 108. Memory 110 may include one or more instructions that can be executed by at least one processor 108 to perform a specific operation. It will be apparent to those skilled in the art that one or more instructions stored in memory 110 enable the hardware of flame detection system 100 to perform predetermined operations. Some known embodiments of memory 110 include, but are not limited to, fixed (hard) drives, magnetic tape, floppy disks, optical disks, compact disc read-only memory (CD-ROM) and magneto-optical disks, semiconductor memories (such as ROM), random access memory (RAM), programmable read-only memory (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), flash memory, magnetic cards or optical cards, or other types of media / machine-readable media suitable for storing electronic instructions.

[0042] In some embodiments, the flame detection system 100 may include input / output circuitry 112. Input / output circuitry 112 enables one or more users to communicate or interact with at least one processor 108 via communication device 114. In some embodiments, input / output circuitry 112 may act as a medium for sending input from and from communication device 114 to the flame detection system 100. In some embodiments, input / output circuitry 112 may refer to hardware and software components that facilitate the exchange of information between the user and the flame detection system 100. Input / output circuitry 112 may include various input devices, such as a keyboard, barcode scanner, and graphical user interface (GUI) for providing data to the user, and various output devices, such as alarm units for alerting the user in the event that one or more ADC signals indicate a flame. In one example, communication circuitry 116 may include N user devices. In some embodiments, communication circuitry 116 may include a graphical user interface (GUI) (not shown) as input circuitry to allow the user to input data. In some embodiments, communication circuitry 116 may include at least one of one or more mobile phones, laptop computers, etc., for generating notification alarms or audible alarms for the user.

[0043] In some embodiments, at least one processor 108 may further include communication circuitry 116. Communication circuitry 116 allows at least one processor 108 to exchange data or information with other systems or devices. Furthermore, communication circuitry 116 may include network interfaces, protocols, and software modules responsible for transmitting and receiving data or information. In some embodiments, communication circuitry 116 may include an Ethernet port, a Wi-Fi adapter, or a communication protocol such as HTTP or MQTT for connecting to other systems. Communication circuitry 116 may also include components for exchanging data with other systems or network devices, such as communication modules (e.g., Wi-Fi, Ethernet, cellular), transceivers, antennas, and protocols (e.g., TCP / IP, MQTT, SNMP). Communication circuitry 116 allows at least one processor 108 to stay up-to-date and accurately determine whether one or more ADC signals indicate a flame or a reflected torch.

[0044] It will be apparent to those skilled in the art that the aforementioned components of the flame detection system 100 are provided for illustrative purposes only without departing from the scope of this disclosure.

[0045] Figure 2 A graphical representation of the spectrum 200 of one or more radiations according to an example embodiment of the present disclosure is illustrated.

[0046] like Figure 1 As shown, at least one flame detector 104 can be configured to detect one or more types of radiation. The one or more types of radiation can be detected in the form of electromagnetic signals. In some embodiments, the one or more electromagnetic signals may correspond to one or more electrical signals. In some embodiments, the one or more electromagnetic signals may have different wavelengths. A graphical representation of the spectrum 200 of one or more types of radiation emitted from the flame source 102 is shown in [the diagram]. Figure 2 The illustration is shown below. The graphic representation can show the spectrum 200 of one or more wavelengths of radiation emitted by the flame source 102. Furthermore, the spectrum 200 can include wavelengths corresponding to the ultraviolet C (UVC) range (0.1 μm–0.3 μm), the visual range (0.3 μm–0.7 μm), the broadband range (0.7 μm–4.5 μm), and the long-wavelength range (3.0 μm–5.0 μm).

[0047] In one example embodiment, spectrum 200 may include wavelengths (in micrometers) of one or more types of radiation emitted due to the combustion of ethylene, as exemplified by 202. Furthermore, spectrum 200 may include wavelengths of one or more types of radiation emitted due to sunlight, as exemplified by 204. In another example embodiment, spectrum 200 may include wavelengths of one or more types of radiation emitted due to the combustion of hydrogen, as exemplified by 206. In some embodiments, spectrum 200 represents multiple peaks around different wavelength ranges. In one example, the peaks in the wavelength range depicted by 208 correspond to water emission peaks indicating a hydrogen flame. Furthermore, the peaks in the wavelength range depicted by 210 correspond to carbon dioxide emission peaks indicating a hydrocarbon (HC) flame. In some embodiments, it is evident from spectrum 200 that the wavelengths and intensities reflected by the torch are very similar to the wavelengths and intensities of a real flame, making it difficult for at least one flame detector 104 to accurately distinguish between the two.

[0048] In some embodiments, at least one flame detector 104 may be configured to detect one or more types of radiation in the form of analog signals. Furthermore, the one or more types of radiation may correspond to one or more types of electromagnetic radiation in the form of infrared (IR), visible light, and ultraviolet (UV) wavelengths. Furthermore, the wavelength of the one or more types of radiation depends on the type of fuel source. Furthermore, at least one processor 108 operatively coupled to at least one flame detector 104 may process the one or more types of radiation into one or more ADC signals. Furthermore, at least one processor 108 may be configured to determine whether the one or more ADC signals indicate a real flame or a reflected torch based on various comparisons between parameters (skewness and kurtosis ratio (SK ratio) thresholds of one or more ADC signals) and multiple threshold conditions.

[0049] Figure 3 An image representation of the field of view (FOV) 300 of at least one flame detector 104 according to an example embodiment of the present disclosure is illustrated.

[0050] In some implementations, FOV 300 may correspond to an area from which at least one flame detector 104 can detect one or more types of radiation. FOV 300 of at least one flame detector 104 may also refer to the angular range from which at least one flame detector 104 receives one or more types of radiation. FOV 300 of at least one flame detector 104 may define a spatial range within which at least one flame detector 104 can capture one or more types of radiation converted into one or more ADC signals. Furthermore, one or more ADC signals may be received by at least one processor 108 and compared with multiple predefined threshold conditions specific to the reflected torch chimney. In one example, FOV 300 may include one or more flames as fires, such as a first fire 302, a second fire 304, a third fire 306, a fourth fire 308, a fifth fire 310, and a sixth fire 312. At least one flame detector 104 may detect one or more types of radiation from the first fire 302, the second fire 304, the third fire 306, the fourth fire 308, the fifth fire 310, and the sixth fire 312. The FOV 300 ensures that at least one flame detector 104 can effectively capture one or more types of radiation from a designated area, thereby allowing a trained ML model to accurately determine whether one or more ADC signals correspond to a real flame or a reflected torch.

[0051] Figure 4 An exemplary scenario 400 of a flame detection system 100 according to an example embodiment of the present disclosure is illustrated.

[0052] In some embodiments, the flame detection system 100 may include at least one flame detector 104. In exemplary scenario 400, the flame zone 402 may emit one or more types of radiation 404. Furthermore, at least one flame detector 104 may detect additional one or more types of radiation 406 from a monitoring area 408 within the field of view (FOV) 300. One or more types of radiation 404 from the flame zone 402 may be reflected by an object 410 present in the monitoring area. For example, one or more types of radiation 404 from the flame zone 402 may be reflected from a high-gloss object 410 (such as a steel pipe used in many industrial plants). At least one flame detector 104 may detect one or more types of radiation 404 reflected by the object 410. Additionally, at least one analog-to-digital converter 106 may convert signals from the flame detector 104 corresponding to one or more types of radiation 404 into one or more ADC signals.

[0053] In some embodiments, at least one processor 108 may receive one or more ADC signals. Furthermore, at least one processor 108 may determine multiple threshold conditions for the reflected flame chimney 412 associated with the one or more ADC signals. In some embodiments, the threshold conditions may include one or more characteristics tailored for the flame and the reflected flame chimney, such as… Figure 5 As shown, the features are fire-only 502, the average SK ratio of fire-only 502 504, the reflected flare chimney 506, and the average SK ratio of reflected flare chimney 506 508. Fire-only 502 can represent multiple threshold conditions specifically tailored to identify characteristics associated with flame detection. The average SK ratio 504 can calculate the average spectral-SK ratio related to fire-related characteristics, thus providing a baseline for comparison. The reflected flare chimney 506 can correspond to multiple threshold conditions designed to distinguish one or more radiations indicative of a reflected flare chimney. The average SK ratio 508 can calculate the average SK ratio related to the characteristics of the reflected flare chimney, enabling comparison with multiple predefined thresholds.

[0054] Furthermore, in some embodiments, at least one processor 108 may be configured to determine parameters associated with one or more ADC signals. The parameters may be determined based at least on spectral density and frequency ratio. In some embodiments, the parameters may at least correspond to the skewness and kurtosis ratio (SK ratio) of one or more ADC signals. Furthermore, at least one processor 108 may be configured to compare the parameters associated with one or more ADC signals with multiple threshold conditions for one or more ADC signals. Based on the comparison, at least one processor 108 may determine one or more types of radiation 404 of the flame zone 402 reflected by the object 202 within the monitoring area 408. Subsequently, at least one processor 108 may determine, at least based on comparisons performed using a trained ML model, that one or more ADC signals indicate a reflected torch chimney 412. For example, analyzed radiation with an SK ratio of 2.1149 having a threshold condition higher than 2.1147 may indicate that one or more types of radiation correspond to a reflected torch chimney 412.

[0055] Figure 5 Table 500 illustrates a plurality of threshold conditions according to an example embodiment of the present disclosure.

[0056] In some implementations, Table 500 represents multiple threshold conditions determined by at least one processor 108. Table 500 includes multiple threshold conditions for flames and reflecting flare chimneys, represented as one or more columns, as feature-only fire 502, SK ratio average of feature-only fire 502 504, feature-reflecting flare chimney 506, and SK ratio average of feature-reflecting flare chimney 506 508. Feature-only fire 502 may represent multiple threshold conditions specifically tailored to identify characteristics associated with flame detection. SK ratio average 504 may calculate an average spectral-SK ratio related to fire-related characteristics, thereby providing a baseline for comparison. Feature-reflecting flare chimney 506 may correspond to multiple threshold conditions designed to distinguish one or more types of radiation indicative of reflecting flare chimneys. SK ratio average 508 may calculate an average SK ratio related to characteristics of reflecting flare chimneys, thereby enabling comparison with multiple predefined thresholds.

[0057] In one example, Table 500 may include a fire-only feature as “LBDC” 502, an average SK ratio as “1.6384310786983747” 504, a reflected flare chimney as “LBDC” 506, and an average SK ratio as “0.7497453729568373” 508. In another example, Table 500 may include a fire-only feature as “PHiSum[NBAC]” 502, an average SK ratio as “1.2443769347355176” 504, a reflected flare chimney as “PHiSum[NBAC]” 506, and an average SK ratio as “1.6933904839481753” 508. In yet another example, Table 500 may include the feature fire-only 502 as “PSDLoBandANbWb”, the SK ratio average as “1.6457600953608769” 504, the feature reflectedflarestack as “PSDLoBandANbWb” 506, and the SK ratio average as “2.569718075806131” 508.

[0058] In another example, Table 500 may include the feature "PSDLoBandBNbLb" (fire-only) 502, the SK ratio average as "1.4212939824639639" 504, the feature "PSDLoBandBNbLb" (reflectedflarestack) 506, and the SK ratio average as "2.316948061016309" 508. In yet another example, Table 500 may include the feature "PSDRatioHiSum[NBAC]" (fire-only) 502, the SK ratio average as "0.7125199251699511" 504, the feature "PSDRatioHiSum[NBAC]" (reflectedflarestack) 506, and the SK ratio average as "2.1147001647499564" 508.

[0059] In another example, Table 500 may include a fire-only feature 502 as “PSDRatioHiSum2[NBAC]”, an average SK ratio 504 as “0.551574422682729”, a reflected torch chimney 506 as “PSDRatioHiSum2[NBAC]”, and an average SK ratio 508 as “1.893154389406503”. In yet another example, Table 500 may include a fire-only feature 502 as “PSDRatioLowSum2[NBAC]”, an average SK ratio 504 as “0.5515746441022596”, a reflected torch chimney 506 as “PSDRatioLowSum2[NBAC]”, and an average SK ratio 508 as “1.8931447742083798”.

[0060] In another example, Table 500 may include a fire-only feature as “PSDMRatioNbLb” 502, an average SK ratio as “2.273277302535659” 504, a reflected flare chimney as “PSDMRatioNbLb” 506, and an average SK ratio as “2.538761980089251” 508. In yet another example, Table 500 may include a fire-only feature as “PSum_WbLo_LbHi” 502, an average SK ratio as “2.752514362148871” 504, a reflected flare chimney as “PSum_WbLo_LbHi” 506, and an average SK ratio as “2.4905566781486606” 508.

[0061] In another example, Table 500 may include a fire-only feature as “PSDLoBandAWbLb” 502, an average SK ratio as “1.6194698605323177” 504, a reflected flare chimney as “PSDLoBandAWbLb” 506, and an average SK ratio as “2.0896854505722855” 508. In yet another example, Table 500 may include a fire-only feature as “FWbacPPSRL” 502, an average SK ratio as “1.0763682539902057” 504, a reflected flare chimney as “FWbacPPSRL” 506, and an average SK ratio as “0.759141339706597” 508.

[0062] In another example, Table 500 may include a fire-only feature 502 as “NBDC”, an average SK ratio 504 as “1.7765228882847788”, a reflected flare chimney 506 as “NBDC”, and an average SK ratio 508 as “1.4969996311752503”. In yet another example, Table 500 may include a fire-only feature 502 as “PSDLoBandANbLb”, an average SK ratio 504 as “1.6540940615785718”, a reflected flare chimney 506 as “PSDLoBandANbLb”, and an average SK ratio 508 as “2.717360295482698”.

[0063] In another example, Table 500 may include fire-only 502 as “WBDC”, SK ratio average 504 as “1.6963037879885439”, flare chimney as “WBDC”, and SK ratio average 508 as “1.3811811410134025”. In yet another example, Table 500 may include fire-only 502 as “PSDRatioLowHigh[NBAC]”, SK ratio average 504 as “2.226514343018175”, flare chimney as “PSDRatioLowHigh[NBAC]”, and SK ratio average 508 as “2.6545885560072398”.

[0064] Figures 6A to 6B Tables 600 and 608 illustrate the verification results of a flame detection system 100 according to an example embodiment of this disclosure. (In conjunction with...) Figures 1 to 5 right Figures 6A to 6B Describe it.

[0065] refer to Figure 6A Table 600 may represent validation results obtained by deploying a trained ML model. In one example, Table 600 may include validation results for one or more files containing data related to fires (represented as “flames”) in one or more radiation sources at different industrial sites. Table 600 may include validation results represented as one or more columns, such as filename 602, “labeled (based on SK-Ratio)” 604, and ground facts 606. Filename 602 may identify a specific file or dataset containing one or more radiation sources collected from various industrial sites, each file or dataset designed for analysis of signals related to fires (“flames”). “labeled (based on SK-Ratio)” 604 may indicate the classification output of the trained ML model, indicating whether each file is labeled as containing a fire based on the SK-Ratio. The classification output can provide insight into the ability of the trained ML model to detect potential fires within the dataset. Ground facts 606 can provide actual ground facts or known results for comparison, thereby validating the accuracy of the trained ML model’s classification output relative to known fire occurrences.

[0066] In one example, table 600 may include filename 602 as "File 1", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In another example, table 600 may include filename 602 as "File 2", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 3", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 4", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame".

[0067] In yet another example, table 600 may include filename 602 as "File 5", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 6", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 7", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 8", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame".

[0068] In yet another example, table 600 may include filename 602 as "File 9", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 10", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 11", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame". In yet another example, table 600 may include filename 602 as "File 12", "marked as (SK-Ratio based)" as "Flame" 604, and ground view 606 as "Flame".

[0069] Furthermore, based on the classification results of files 1 to 12, all filenames 602 were labeled as "Flame," the same as the ground reality 606 "Flame." As a result, the percentage (%) correctly labeled as "Flame (based on SK_Ratio)" was 100.0%.

[0070] refer to Figure 6BTable 608 may represent validation results obtained by deploying a trained ML model. In one example, Table 608 may include validation results for one or more files containing data related to reflected flare stacks (denoted as “reflected flare stacks”) in one or more types of radiation from different industrial sites. Table 608 may include validation results represented as one or more columns, such as filename 610, “labeled (based on SK-Ratio)” 612, and ground fact 614. Filename 610 may identify a specific file or dataset containing one or more types of radiation collected from various industrial sites, each file or dataset designed for analysis of signals related to reflected flare stacks (“reflected flare stacks”). “labeled (based on SK-Ratio)” 612 may indicate the classification output of the trained ML model, indicating whether each file is labeled as containing a reflected flare stack based on SK-Ratio. The classification output can provide insight into the ability of the trained ML model to detect reflected flare stacks within the dataset. Ground Facts 614 can provide actual ground facts or known results for comparison, thereby verifying the accuracy of the classification output of the trained ML model relative to known reflections of flare chimney occurrences.

[0071] In one example, table 608 may include a filename 610 as "File_1", a "labeled as (SK-Ratio based)" 612 as "ReflectedFlareStack", and ground conditions 614 as "ReflectedFlareStack". In another example, table 608 may include a filename 610 as "File_2", a "labeled as (SK-Ratio based)" 612 as "ReflectedFlareStack", and ground conditions 614 as "ReflectedFlareStack". In yet another example, table 608 may include a filename 610 as "File_3", a "labeled as (SK-Ratio based)" 612 as "ReflectedFlareStack", and ground conditions 614 as "ReflectedFlareStack".

[0072] Furthermore, based on the classification results of File_1 to File_3, all filenames 602 were labeled as "ReflectedFlareStack," the same as the ground reality 606 "ReflectedFlareStack." As a result, the percentage (%) of those correctly labeled as "ReflectedFlareStack (based on SK_Ratio)" was 100.0%.

[0073] Figure 7 A flowchart illustrating a method 700 for flame detection according to an example embodiment of the present disclosure is shown.

[0074] At operation 702, at least one processor 108 may be configured to receive, via at least one processor 108, one or more analog-to-digital conversion (ADC) signals corresponding to one or more types of radiation detected within the field of view (FOV) of at least one flame detector. In some embodiments, at least one flame detector 104 may include at least one of an IR sensor, a photodiode, or a combination of an IR sensor and a photodiode. In some embodiments, at least one processor 108 may receive one or more analog-to-digital conversion (ADC) signals from at least one analog-to-digital converter 106.

[0075] At operation 704, at least one processor 108 may be configured to determine multiple threshold conditions for a reflected torch chimney associated with one or more ADC signals. In some embodiments, at least one processor 108 may be configured to extract multiple threshold conditions in both low-frequency and high-frequency ranges of the flame or the reflected torch.

[0076] At operation 706, at least one processor 108 may be configured to determine parameters associated with one or more ADC signals. The parameters may be determined based at least on spectral density and frequency ratio. In some embodiments, the parameters may correspond at least to the SK ratio of one or more ADC signals.

[0077] At operation 708, at least one processor 108 may be configured to compare parameters against multiple threshold conditions of one or more ADC signals. The parameters may satisfy multiple threshold conditions. In some embodiments, at least one processor 108 is configured to compare skewness and kurtosis ratio against multiple threshold conditions using a Kolmogorov-Smirnov test.

[0078] In operation 710, at least one processor 108 may be configured to determine, at least based on comparison, whether one or more ADC signals indicate a flame or a reflecting torch. In some embodiments, at least one processor 108 may be configured to detect fluctuations or modulations of multiple threshold conditions determined based on one or more ADC signals to determine whether the one or more ADC signals indicate a flame or a reflecting torch. In some embodiments, the trained ML model may include aggregate simulations of multiple models trained by at least one processor 108 over a period of time.

[0079] In some implementations, method 700 may include training an ML model via at least one processor 108 based on at least a plurality of threshold conditions extracted from one or more ADC signals over a period of time. Method 700 may further include deploying the trained ML model via at least one processor 108 to determine a flame or a reflecting torch.

[0080] In some embodiments, method 700 may further include generating a signal via at least one processor 108 in response to determining that one or more ADC signals indicate a flame. Thereafter, method 700 may include sending the signal via at least one processor 108 to communication circuitry 116 for warning a user.

[0081] The embodiments disclosed herein improve detection accuracy by converting radiation in the field of view into a precise ADC signal, thereby ensuring reliable data acquisition. By establishing multiple threshold conditions specifically tailored for reflecting flare chimneys, the flame detection system can effectively distinguish between actual fires and benign reflections, minimizing false alarms. Furthermore, the determination of parameters such as spectral density and frequency ratio increases the robustness of the analysis, providing deeper insights into the nature of the detected signal. The use of a trained ML model enhances classification capabilities, enabling the flame detection system to distinguish between actual fires, intentional flames, and reflecting flare chimneys with high accuracy. Integrating these features into a unified framework optimizes the operational efficiency and reliability of the flame detection system under various environmental conditions, thereby enhancing overall safety and reducing unnecessary interruptions.

[0082] Those skilled in the art to which this invention pertains will, upon benefiting from the teachings presented in the foregoing description and associated drawings, contemplate numerous modifications and other embodiments of the invention set forth herein. Therefore, it should be understood that the invention is not limited to the specific embodiments disclosed, and that modifications and other embodiments are intended to be included within the scope of the appended claims. Furthermore, although the foregoing description and associated drawings have described exemplary embodiments in the context of certain example combinations of elements and / or functions, it should be understood that different combinations of elements and / or functions may be provided by alternative embodiments without departing from the scope of the appended claims. In this regard, for example, different combinations of elements and / or functions explicitly described above are also contemplated, as may be set forth in some of the appended claims. Although specific terms are used herein, they are used only in a general and descriptive sense and not for limiting purposes.

Claims

1. A flame detection system, the flame detection system comprising: At least one flame detector, the at least one flame detector being configured to detect one or more types of radiation within a field of view (FOV); At least one analog-to-digital converter, the at least one analog-to-digital converter being configured to convert one or more signals corresponding to one or more detected radiations into one or more analog-to-digital conversion (ADC) signals; and At least one processor, operatively coupled to the at least one flame detector, the at least one processor being configured to: Receive the one or more ADC signals from the at least one analog-to-digital converter; Determine multiple threshold conditions for the reflected torch chimney associated with the one or more ADC signals; Determine the parameters associated with the one or more ADC signals; The parameter is compared with each of the plurality of threshold conditions of the one or more ADC signals; as well as The comparison is used to determine whether the one or more ADC signals indicate a flame or a reflected torch.

2. The flame detection system according to claim 1, wherein the parameters correspond to the skewness and kurtosis ratio (SK ratio) thresholds of the one or more ADC signals; and The at least one processor is configured to compare the skewness and kurtosis ratio with each of the plurality of threshold conditions using a Kolmogorov-Smirnov test.

3. The flame detection system of claim 1, wherein the at least one processor is configured to compare the parameter with each of the plurality of threshold conditions of the one or more ADC signals to determine whether the parameter associated with the one or more ADC signals satisfies each of the determined plurality of threshold conditions.

4. The flame detection system according to claim 3, wherein the at least one processor is configured to: When it is determined that the parameter associated with the one or more ADC signals satisfies each of the determined plurality of threshold conditions, it is determined whether the one or more ADC signals indicate a reflected torch; When it is determined that the parameter associated with the one or more ADC signals does not meet the determined plurality of threshold conditions, it is determined whether the one or more ADC signals indicate a flame; A signal is generated in response to determining that the one or more ADC signals indicate the flame; as well as The signal is sent to the communication device to warn the user.

5. The flame detection system of claim 1, wherein the at least one processor is configured to determine the plurality of threshold conditions associated with the one or more ADC signals based at least on flag conditions associated with the one or more ADC signals; and The flag condition refers to the one or more ADC signals received by the at least one processor having one or more fluctuations and modulations within a certain time period.

6. A method, the method comprising: Receive one or more analog-to-digital conversion (ADC) signals corresponding to one or more types of radiation detected within the field of view (FOV) of at least one flame detector via at least one processor; Multiple threshold conditions for the reflected torch chimney associated with the one or more ADC signals are determined via the at least one processor; Parameters associated with the one or more ADC signals are determined via the at least one processor; The parameters are compared with the plurality of threshold conditions of the one or more ADC signals via the at least one processor; as well as The processor determines, at least based on the comparison, whether the one or more ADC signals indicate a flame or a reflected torch.

7. The method of claim 6, wherein the parameters correspond to the skewness and kurtosis ratio (SK ratio) thresholds of the one or more ADC signals; and The method further includes comparing the skewness and kurtosis ratio with each of the plurality of threshold conditions via the at least one processor using a Kolmogorov-Smirnov test.

8. The method of claim 6, wherein the parameter is compared with each of the plurality of threshold conditions of the one or more ADC signals to determine whether the parameter associated with the one or more ADC signals satisfies each of the determined plurality of threshold conditions.

9. The method according to claim 6, further comprising: When it is determined that the parameters associated with the one or more ADC signals satisfy each of the determined plurality of threshold conditions, it is determined via the at least one processor whether the one or more ADC signals indicate a reflected torch; When it is determined that the parameter associated with the one or more ADC signals does not meet the determined plurality of threshold conditions, it is determined via the at least one processor whether the one or more ADC signals indicate a flame; A signal is generated via the at least one processor in response to determining that the one or more ADC signals indicate the flame; as well as The signal is sent to the communication device via the at least one processor to warn the user.

10. The method according to claim 6, further comprising: The plurality of threshold conditions associated with the one or more ADC signals are determined, at least based on flag conditions associated with the one or more ADC signals, via the at least one processor. The flag condition refers to the one or more ADC signals received by the at least one processor having one or more fluctuations and modulations within a certain time period.

Images