Flame detection method, device, electronic device and computer-readable storage medium

Abstract

Embodiments of the present application provide a flame detection method, apparatus, electronic device, and computer-readable storage medium. In this embodiment, the first flame band data and second flame band data required for flame detection are not obtained in a fixed gain mode, but rather are obtained through a dynamically adjusted gain mode. Compared to directly using flame band data in a fixed gain mode for flame detection, this method of using flame band data in a dynamically adjusted gain mode for flame detection avoids the problems of missing detection of close-range flames due to excessive gain amplitude in a fixed gain mode, and failing to identify distant flames due to excessive gain amplitude in a fixed gain mode. This allows flame detection using the method of this embodiment to take into account both close-range and long-range flame detection, effectively reducing the probability of missed detection, and thus improving the accuracy of flame detection.

Description

Technical Field

[0001] The present application relates to the field of security, and in particular to a flame detection method, device, electronic device, and computer-readable storage medium. Background Art

[0002] Fire is currently one of the main causes of property loss. Accurately detecting flames and issuing alarms helps to detect fires early. Currently, given that flames exhibit different characteristics in different infrared bands, flame detection is achieved by analyzing the relationship between the flame band data collected by flame detectors in multiple infrared bands.

[0003] However, this transformation is difficult to accurately analyze and can easily lead to false alarms, resulting in low flame detection accuracy. Furthermore, in this flame detection method, the flame detector collects flame band data in a fixed-gain mode. The fixed-gain mode has a fixed gain amplitude, which can easily miss flames that are close or far from the flame detector, further reducing flame detection accuracy. Summary of the Invention

[0004] In view of this, embodiments of the present application provide a flame detection method, device, electronic device, and computer-readable storage medium to reduce the false alarm rate and thus improve the accuracy of flame detection.

[0005] The present invention provides an infrared flame detection method, which is applied to a flame detector and includes:

[0006] Collect flame band data corresponding to the flame detector in the reference gain mode and the non-reference gain mode in the current time window; and determine the reference gain mode corresponding to the next time window based on the flame band data collected in the reference gain mode in the current time window, so as to collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window;

[0007] Obtain first flame band data; the first flame band data includes flame band data respectively collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window depends on whether the reference gain modes corresponding to the current time window and the next time window are the same. If the reference gain modes corresponding to the current time window and the next time window are the same, then the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window; if the reference gain modes corresponding to the current time window and the next time window are different, then the target gain mode corresponding to the current time window is other gain modes corresponding to the current time window, and the other gain modes are reference gain modes corresponding to the next time window;

[0008] Obtaining second flame band data; the second flame band data is a subset of the first flame band data;

[0009] Flame detection is performed using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data.

[0010] The present application also provides an infrared flame detection device, which is applied to a flame detector and includes:

[0011] A determination module is configured to respectively collect flame band data corresponding to the flame detector in a reference gain mode and a non-reference gain mode in a current time window; and determine a reference gain mode corresponding to a next time window based on the flame band data collected in the reference gain mode in the current time window, so as to respectively collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window;

[0012] A first acquisition module is used to obtain first flame band data; the first flame band data includes flame band data respectively collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window is determined depending on whether the reference gain modes corresponding to the current time window and the next time window are the same; if the reference gain modes corresponding to the current time window and the next time window are the same, then the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window; if the reference gain modes corresponding to the current time window and the next time window are different, then the target gain mode corresponding to the current time window is other gain modes corresponding to the current time window, and the other gain modes are the reference gain modes corresponding to the next time window;

[0013] A second obtaining module is configured to obtain second flame band data; the second flame band data is a subset of the first flame band data;

[0014] The detection module is configured to perform flame detection using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data.

[0015] An embodiment of the present application further provides an electronic device, comprising: a processor and a memory for storing computer program instructions, wherein the computer program instructions, when executed by the processor, enable the processor to execute the steps of the above method.

[0016] An embodiment of the present application further provides a machine-readable storage medium, which stores computer program instructions. When the computer program instructions are executed, the steps of the above method can be implemented.

[0017] It can be seen from the above technical solution that in this embodiment, the reference gain mode corresponding to the next time window is determined based on the flame band data collected under the reference gain mode in the current time window, and the target gain mode of the current time window is determined based on whether the reference gain modes corresponding to the current time window and the next time window are the same, and the flame band data collected under the target gain modes corresponding to N time windows are used as the first flame band data, and the first flame band data and the second flame band data (i.e., a subset of the first flame band data) are selected as the flame band data required for subsequent detection.

[0018] This ensures that the flame band data required for subsequent detection (i.e., the first flame band data and the second flame band data) are not obtained in the fixed gain mode, but rather obtained through the dynamic gain adjustment mode. Compared to the method of directly using the flame band data in the fixed gain mode for flame detection, this method of using the flame band data in the dynamic gain adjustment mode for flame detection does not result in the situation where the gain amplitude in the fixed gain mode is too large, resulting in missed detection of close-range flames, or the gain amplitude in the fixed gain mode is too small, resulting in the inability to identify distant flames. This allows the method of this embodiment to be used for flame detection, allowing for both close-range and long-range flame detection, effectively reducing the probability of missed detection, thereby improving the accuracy of flame detection.

[0019] Furthermore, flame detection is performed using the flame detection model based on the time and frequency domain characteristics of the first flame band data and the second flame band data. This allows flame detection to simultaneously consider the temporal fluctuation frequency and the signal intensity distribution in the specified frequency domain, fully extracting flame characteristics. This helps the flame detection model accurately identify flames, thereby improving the accuracy of flame detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of the method flow provided in an embodiment of the present application.

[0021] Figure 2 A schematic diagram of a flow chart for determining a reference gain mode corresponding to a next time window provided in an embodiment of the present application.

[0022] Figure 3 A schematic diagram of a flow chart for determining a reference gain mode corresponding to a next time window provided in an embodiment of the present application.

[0023] Figure 4 A schematic diagram of the process of performing flame detection using a flame detection model provided in an embodiment of the present application.

[0024] Figure 5 A flowchart of another method provided in an embodiment of the present application.

[0025] Figure 6 A schematic diagram of the device structure provided in an embodiment of the present application.

[0026] Figure 7 A schematic diagram of the electronic device structure provided in an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the technical solutions provided by the embodiments of the present application, and to make the above-mentioned purposes, features and advantages of the embodiments of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application are further described in detail below with reference to the accompanying drawings.

[0028] See also Figure 1 , Figure 1 This is a flow chart of a method provided in an embodiment of the present application. This flow chart and this method are applied to a flame detector. As an embodiment, the flame detector may be a pyroelectric infrared detector.

[0029] like Figure 1 As shown, the process may include the following steps:

[0030] S101, respectively collecting flame band data corresponding to the flame detector in the reference gain mode and the non-reference gain mode in the current time window; and determining the reference gain mode corresponding to the next time window based on the flame band data collected in the reference gain mode in the current time window, so as to respectively collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window.

[0031] In this embodiment, the time window refers to a time period of a set length, such as 2 seconds or 3 seconds. The time window can be set according to actual business needs and is not specifically limited in this embodiment.

[0032] In this embodiment, at each sampling point within the current time window, an infrared sensor configured within the flame detector and operating within a specified wavelength band collects an infrared signal. A conversion module configured within the flame detector converts the infrared signal into an electrical signal, and a two-stage amplifier configured within the flame detector amplifies the electrical signal. The amplification factor of the first-stage amplifier is smaller than the amplification factor of the second-stage amplifier. The electrical signal amplified by the first-stage amplifier represents the flame value collected at that sampling point in the set low-gain mode, and the electrical signal amplified by the second-stage amplifier represents the flame value collected at that sampling point in the set high-gain mode. Optionally, the amplification factors corresponding to the set high-gain mode and the set low-gain mode are different, and the amplification factor corresponding to the set high-gain mode is greater than the amplification factor of the set low-gain mode.

[0033] From the above description, it can be seen that in the current time window, the flame values ​​at each sampling point in the high gain mode constitute the flame band data in the high gain mode in the current time window. The flame values ​​at each sampling point in the low gain mode constitute the flame band data in the low gain mode in the current time window.

[0034] In this embodiment, in each time window, the gain mode selected from the high gain mode and the low gain mode is the reference gain mode, and the unselected gain mode is the non-reference gain mode. For example, for the current time window, if the high gain mode is selected as the reference gain mode corresponding to the current time window, and the low gain mode is selected as the non-reference gain mode corresponding to the current time window, then the flame band data collected in the reference gain mode in the current time window is composed of the flame values ​​at each sampling point in the high gain mode, and the flame band data collected in the non-reference gain mode in the current time window is composed of the flame values ​​at each sampling point in the low gain mode. Vice versa, which will not be repeated here.

[0035] In this embodiment, the reference gain mode for the first time window is set to high gain mode. For each subsequent time window, the reference gain mode for the next time window is determined based on the flame band data collected under the reference gain mode during the current time window. The specific implementation of this determination of the reference gain mode for the next time window will be described later and will not be detailed here.

[0036] It should be noted that the above-mentioned designated wavelength bands may be 2.7 nm, 3.8 nm, 4.3 nm, and 5.0 nm. The flame wavelength data collected in these wavelength bands are beneficial for flame detection.

[0037] S102, obtaining first flame band data; the first flame band data includes flame band data collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window is determined based on whether the reference gain modes corresponding to the current time window and the next time window are the same.

[0038] In this embodiment, if the reference gain modes corresponding to the current time window and the next time window are the same, the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window. If the reference gain modes corresponding to the current time window and the next time window are different, the target gain mode corresponding to the current time window is the other gain mode corresponding to the current time window, and the other gain mode is the reference gain mode corresponding to the next time window. The target gain mode of the current time window remains consistent with the reference gain mode of the next time window.

[0039] For example, when the reference gain mode of this time window is the set high gain mode, if it is determined that the reference gain mode corresponding to the next time window is still the set high gain mode based on the flame band data collected in the set high gain mode within this time window, then the target gain mode corresponding to this time window is the set high gain mode; if it is determined that the reference gain mode corresponding to the next time window is the set low gain mode, then the target gain mode corresponding to this time window is the set low gain mode.

[0040] When the reference gain mode of this time window is the set low gain mode, if the reference gain mode corresponding to the next time window is determined to be the set low gain mode based on the flame band data collected in the set low gain mode within this time window, the target gain mode corresponding to this time window is still the set low gain mode; if the reference gain mode corresponding to the next time window is determined to be the set high gain mode, the target gain mode corresponding to this time window is the set high gain mode.

[0041] In this embodiment, the target gain modes corresponding to the N time windows are dynamically switched between a set high gain mode and a set low gain mode, which enables obtaining the first flame band data by dynamically adjusting the gain mode.

[0042] S103, obtaining second flame band data; the second flame band data is a subset of the first flame band data.

[0043] For example, if the time window is 2 seconds, the flame band data collected in the target gain mode within each 2-second period are spliced ​​together in the order of the time windows. For the spliced ​​flame band data, a sliding window with a step size of 1 second and a duration of 6 seconds is used. The flame band data within the current sliding window is the first flame band data. Starting from the same time point, a sliding window with a step size of 1 second and a duration of 3 seconds is used. The flame band data within the current sliding window is the second flame band data. In this case, the second flame band data is a subset of the first flame band data.

[0044] S104 , performing flame detection using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data.

[0045] In this embodiment, the specific implementation of the above step S104 will be described later and will not be described here in detail.

[0046] So far, completed Figure 1 The process shown.

[0047] pass Figure 1As can be seen from the illustrated process, the flame band data required for detection (i.e., the first flame band data and the second flame band data) is not obtained in a fixed gain mode, but rather through a dynamically adjusted gain mode. Compared to directly using flame band data in a fixed gain mode for flame detection, this method of using flame band data in a dynamically adjusted gain mode for flame detection avoids the problems of missing close-range flames due to excessive gain amplitude in a fixed gain mode, or failing to identify distant flames due to excessive gain amplitude in a fixed gain mode. This allows the method of this embodiment to be used for flame detection, effectively reducing the probability of missed detection and thus improving the accuracy of flame detection.

[0048] Furthermore, flame detection is performed using the flame detection model based on the time and frequency domain characteristics of the first flame band data and the second flame band data. This allows flame detection to simultaneously consider the temporal fluctuation frequency and the signal intensity distribution in the specified frequency domain, fully extracting flame characteristics. This helps the flame detection model accurately identify flames, thereby improving the accuracy of flame detection.

[0049] The specific implementation process of determining the reference gain mode corresponding to the next time window is described in detail below with reference to a specific embodiment:

[0050] Figure 2 A schematic diagram of a flow chart for determining a reference gain mode corresponding to a next time window provided in an embodiment of the present application.

[0051] In this embodiment, the reference gain mode of the current time window is set to a high gain mode, and the flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points. Figure 2 As shown, the process may include the following steps:

[0052] S201 , counting the number of target data meeting the conditions from the flame band data collected in the reference gain mode within the current time window.

[0053] In this embodiment, the above condition means that the flame value at at least K consecutive sampling points within the current time window is greater than or equal to the set flame upper limit. Alternatively, the flame value at at least P consecutive sampling points within the current time window is less than or equal to the set flame lower limit, where K is greater than 1 and P is greater than 1.

[0054] The above K and P are determined according to the sampling frequency and preset time length of the flame detector. For example, if the sampling frequency is 1000 Hz (sampling once every 1 ms) and the preset time length is 30 ms, then K and P can be 30.

[0055] In this embodiment, the number of target data meeting the conditions can be counted in real time starting from the beginning of the current time window, or can be counted after the end of the current time window.

[0056] S202, determining whether the number is greater than a set counting threshold.

[0057] If the judgment result of step S202 is yes, then the following step S203 is executed; if the judgment result of step S202 is no, then the following step S204 is executed.

[0058] S203: Determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0059] S204: Determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0060] In this embodiment, if the judgment result of the above step S202 is yes, it indicates that the flame band data collected in the set high gain mode is oversaturated and distorted, and therefore the set low gain mode is determined as the reference gain mode corresponding to the next time window. If the judgment result of the above step S202 is no, it indicates that the flame band data collected in the set high gain mode is not distorted, and therefore the set high gain mode is still determined as the reference gain mode corresponding to the next time window.

[0061] Figure 3 A schematic diagram of a flow chart for determining a reference gain mode corresponding to a next time window provided in an embodiment of the present application.

[0062] In this embodiment, the reference gain mode of the current time window is set to a low gain mode, and the flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points. Figure 3 As shown, the process may include the following steps:

[0063] S301, performing a specified operation on the flame value at each sampling point in the current time window to obtain an operation result.

[0064] As an embodiment, a specified operation is performed on the flame value at each sampling point in the current time window according to the following formula to obtain an operation result:

[0065] a=∑(avg-i) 2

[0066] Where i is the flame value collected at sampling point i in the current time window, and avg is the average flame value collected at each sampling point in the current time window.

[0067] S302: Determine whether the calculation result is less than or equal to a preset value.

[0068] If the judgment result of S302 is yes, then the following step S303 is executed; if the judgment result of S302 is no, then the following step S304 is executed.

[0069] S303: Determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0070] S304: Determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0071] In this embodiment, if the judgment result of the above step S202 is yes, it indicates that the flame band data collected in the set low gain mode cannot reflect the changes in the flame and has been distorted, and the reference gain mode needs to be adjusted. Therefore, the high gain mode is determined to be the reference gain mode corresponding to the next time window. If the judgment result of the above step S202 is no, it indicates that the flame band data collected in the set low gain mode can reflect the changes in the flame and has not been distorted. Therefore, the low gain mode is still determined to be the reference gain mode corresponding to the next time window.

[0072] The above describes in detail the specific implementation process of determining the reference gain mode corresponding to the next time window. The following describes in detail the specific implementation method of flame detection using the flame detection model:

[0073] Figure 4 The following is a flow chart of flame detection using a flame detection model according to an embodiment of the present application. Figure 4 As shown, the process may include the following steps:

[0074] S401, obtaining time domain features and frequency domain features of first flame band data, and obtaining time domain features and frequency domain features of second flame band data.

[0075] In this embodiment, as an example, the first flame band data is input into a trained time domain feature extraction model to obtain the time domain features of the first flame band data. A Fourier transform is performed on the first flame band data to achieve a time-to-frequency domain conversion. First flame band data in a specified frequency domain (e.g., 3-20 Hz) is selected from the converted first flame band data and input into the trained frequency domain feature extraction model to obtain the frequency domain features of the first flame band data. Similarly, the time domain features and frequency domain features of the second flame band data are obtained.

[0076] S402: The time domain features and frequency domain features of the first flame band data are concatenated to obtain a first concatenated feature. The first concatenated feature is subjected to data preprocessing and input into the trained first flame detection model to obtain a first flame detection result. The data preprocessing is used to eliminate differences in the frequency domain features caused by time length.

[0077] S403: Splicing the time domain features and frequency domain features of the second flame band data to obtain a second splicing feature, and inputting the second splicing feature into the trained second flame detection model to obtain a second flame detection result.

[0078] S404, determining a target flame detection result based on the first flame detection result and the second flame detection result; the target flame detection result is used to indicate whether a flame exists.

[0079] In this embodiment, as an embodiment, if the detection accuracy of the flame detector meets the set low accuracy requirement, when any flame detection result of the first flame detection result and the second flame detection result indicates the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

[0080] As another embodiment, if the detection accuracy of the flame detector meets the set high accuracy requirement, when both the first flame detection result and the second flame detection result indicate the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

[0081] The specific implementation method of flame detection using the flame detection model is described in detail above.

[0082] To explain in more detail, the following Figure 5 The solution provided in this application is described in more detail by way of specific embodiments.

[0083] In this embodiment, when a flame detector is configured with infrared sensors operating in four different wavelengths based on specific application scenarios, these four sensors collect flame wavelength data at 2.7 nm, 3.8 nm, 4.3 nm, and 5.0 nm, respectively. The 2.7 nm and 4.3 nm wavelengths are more sensitive to flames, while the 3.8 nm and 5.0 nm wavelengths are less sensitive. Because the 2.7 nm and 4.3 nm wavelengths are flame-sensitive, the 2.7 nm wavelength is selected as the target wavelength, and the remaining three wavelengths are non-target wavelengths.

[0084] It should be noted that the flame detector can be configured with infrared sensors operating in different bands according to different application scenarios.

[0085] like Figure 5 As shown, the process may include the following steps:

[0086] 1. For the 2.7nm band, the flame band data corresponding to the flame detector in the reference gain mode and non-reference gain mode in the 2.7nm band are collected in the current time window, and the reference gain mode corresponding to the next time window is determined based on the flame band data collected in the reference gain mode in the 2.7nm band in the current time window, so as to collect the flame band data corresponding to the reference gain mode and non-reference gain mode in the 2.7nm band in the next time window.

[0087] Specifically, the reference gain of the first time window is set to a high gain mode, and the flame band data under the set high gain mode within the time window is obtained. The number of target data that meet the conditions is counted from the flame band data collected under the reference gain mode within the current time window. If the above number is greater than the set counting threshold, the low gain mode is set to be determined as the reference gain mode for the second time window. Otherwise, the high gain mode is still set to be determined as the reference gain mode for the second time window.

[0088] If the reference gain mode of the second time window is set to low gain mode, the flame value at each sampling point in the second time window is calculated according to the following formula to obtain the calculation result:

[0089] a=∑(avg-i) 2

[0090] If the result of the above operation is less than or equal to the preset value, the high gain mode is set as the reference gain mode for the third time window. Otherwise, the low gain mode is still set as the reference gain mode for the third time window. And so on, the reference gain mode for each time window is determined.

[0091] 2. Obtain first flame band data in the 2.7 nm band; the first flame band data in the 2.7 nm band includes flame band data collected by the flame detector in the 2.7 nm band under target gain modes corresponding to N time windows.

[0092] Specifically, if the reference gain mode of the first time window is set to high gain mode and the reference gain mode of the second time window is set to low gain mode, since the two are different, the target gain mode of the first time window is set to low gain mode. If the reference gain mode of the first time window is set to high gain mode and the reference gain mode of the second time window is set to high gain mode, since the two are the same, the target gain mode of the first time window is set to high gain mode. And so on, the target gain mode of each time window is determined.

[0093] Each time window is 3s. The flame band data collected in the target gain mode in each time window are spliced ​​in the order of the time windows. For the spliced ​​flame band data, a sliding window with a sliding time of 6s is slid with a step size of 1s. The flame band data in the current sliding window is the first flame band data under the 2.7nm band.

[0094] 3. Obtain the second flame band data in the 2.7 nm band. The second flame band data in the 2.7 nm band is a subset of the first flame band data in the 2.7 nm band.

[0095] Specifically, starting from the same time point, a sliding window with a sliding time of 3 seconds is slid with a step size of 1 second. The flame band data in the current sliding window is the second flame band data under the 2.7 nm band.

[0096] Thus, through the above steps, the first flame band data and the second flame band data of the target band (ie, the 2.7 nm band) are obtained.

[0097] 4. After determining whether the target gain mode corresponding to each time window is set to high gain mode or low gain mode based on the flame band data in the 2.7nm band, for any of the three bands of 3.8nm, 4.3nm, and 5.0nm, the flame band data collected under the target gain mode in each time window of the band are processed according to the above steps 2 and 3 to obtain the first flame band data and the second flame band data in the band.

[0098] In this way, for non-target bands (ie, 3.8 nm, 4.3 nm, and 5.0 nm bands), the first flame band data and the second flame band data in each of the three bands are also obtained.

[0099] It should be noted that after determining the target gain mode for each time window based on the flame band data in the 2.7nm band, when obtaining the first and second flame band data for other bands, it is not necessary to determine the reference gain mode and target gain mode for each time window based on the flame band data in that band. Instead, the target gain mode corresponding to each time window in the 2.7nm band is reused. In other words, the target gain modes for each time window in the 3.8nm, 4.3nm, and 5.0nm bands switch in accordance with the switching of the target gain modes for each time window in the 2.7nm band.

[0100] At this point, the first flame band data and the second flame band data at the four bands of 2.7 nm, 3.8 nm, 4.3 nm, and 5.0 nm have been obtained.

[0101] 5. Input the above four first flame band data into the trained time domain feature extraction model A1 to obtain the first time domain feature. Perform a Fourier transform on the first flame band data in the 2.7 nm band, select the first flame band data in the 3-20 Hz range from the converted first flame band data, and input the first flame band data in the 3-20 Hz range into the trained frequency domain feature extraction model C1 to obtain the first frequency domain feature. Concatenate the first time domain feature with the first frequency domain feature to obtain the first concatenated feature. Perform data preprocessing on the first concatenated feature, and input the preprocessed first concatenated feature into the trained flame detection model B1 to obtain the first flame detection result.

[0102] 6. Input the aforementioned four second flame band data into the trained time-domain feature extraction model A2 to obtain the second time-domain feature. Perform a Fourier transform on the second flame band data at 2.7 nm, and input the converted second flame band data corresponding to 2.7 nm into the trained frequency-domain feature extraction model C2 to obtain the second frequency-domain feature. Concatenate the second time-domain feature with the second frequency-domain feature to obtain a second concatenated feature. Input the second concatenated feature into the trained flame detection model B2 to obtain the second flame detection result.

[0103] It should be noted that since the time length corresponding to the first flame band data is twice the time length corresponding to the second flame band data, it is known that the longer the time, the stronger the spectrum intensity, and the corresponding frequency domain features will be more prominent. In order to eliminate the detection error caused by the difference in frequency domain features caused by eliminating the time length, each parameter in the first splicing matrix can be reduced by half before inputting the first splicing matrix into the trained flame detection model B1.

[0104] 7. If the detection accuracy of the flame detector meets the set low accuracy requirement, then when either the first flame detection result or the second flame detection result indicates the presence of a flame, the target flame detection result is determined to indicate the presence of a flame. If the detection accuracy of the flame detector meets the set high accuracy requirement, then when both the first flame detection result and the second flame detection result indicate the presence of a flame, the target flame detection result is determined to indicate the presence of a flame.

[0105] The above describes the method provided in the embodiment of the present application. The following describes the device provided in the embodiment of the present application:

[0106] See also Figure 6 , Figure 6 This is a diagram of the structure of the device provided in the embodiment of the present application. The device is applied to flame detectors, such as Figure 6 As shown, the apparatus may include: a determination module 601 , a first obtaining module 602 , a second obtaining module 603 , and a detection module 604 .

[0107] Determination module 601 is used to collect flame band data corresponding to the flame detector in the reference gain mode and the non-reference gain mode in the current time window; and determine the reference gain mode corresponding to the next time window based on the flame band data collected in the reference gain mode in the current time window, so as to collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window;

[0108] The first acquisition module 602 is used to obtain first flame band data; the first flame band data includes flame band data respectively collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window is determined by whether the reference gain modes corresponding to the current time window and the next time window are the same; if the reference gain modes corresponding to the current time window and the next time window are the same, then the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window; if the reference gain modes corresponding to the current time window and the next time window are different, then the target gain mode corresponding to the current time window is the other gain mode corresponding to the current time window, and the other gain mode is the reference gain mode corresponding to the next time window;

[0109] The second obtaining module 603 is used to obtain second flame band data; the second flame band data is a subset of the first flame band data;

[0110] The detection module 604 is configured to perform flame detection using a flame detection model and based on the time domain features and frequency domain features of the first flame band data and the time domain features and frequency domain features of the second flame band data.

[0111] As an embodiment, the flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points;

[0112] The reference gain mode corresponding to the next time window is determined based on the flame band data collected under the reference gain mode in the current time window, including:

[0113] If the reference gain mode of the current time window is the set high gain mode, count the number of target data that meet the conditions from the flame band data collected in the reference gain mode within the current time window; the conditions are: the flame values ​​at at least K consecutive sampling points in the current time window are greater than or equal to the set flame upper limit value; or the flame values ​​at at least P consecutive sampling points in the current time window are less than or equal to the set flame lower limit value; K is greater than 1, and P is greater than 1;

[0114] Determine whether the number is greater than the set counting threshold. If so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to the low gain mode. If not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0115] As an embodiment, the flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points;

[0116] The reference gain mode corresponding to the next time window is determined based on the flame band data collected under the reference gain mode in the current time window, including:

[0117] If the reference gain mode of the current time window is the set low gain mode, perform the specified operation on the flame value at each sampling point in the current time window to obtain the operation result;

[0118] Determine whether the operation result is less than or equal to a preset value. If so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to a high gain mode; if not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

[0119] As an embodiment, performing flame detection using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data includes:

[0120] Splicing the time domain features and the frequency domain features of the first flame band data to obtain a first splicing feature;

[0121] The first splicing feature is subjected to data preprocessing and input into the trained first flame detection model to obtain a first flame detection result; the data preprocessing is used to eliminate the difference in frequency domain features caused by time length;

[0122] Splicing the time domain features and the frequency domain features of the second flame band data to obtain a second splicing feature; inputting the second splicing feature into the trained second flame detection model to obtain a second flame detection result;

[0123] A target flame detection result is determined based on the first flame detection result and the second flame detection result; the target flame detection result is used to indicate whether a flame exists.

[0124] As an embodiment, determining the target flame detection result according to the first flame detection result and the second flame detection result includes:

[0125] If the detection accuracy of the flame detector meets the set low accuracy requirement, then when any one of the first flame detection result and the second flame detection result indicates the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

[0126] As an embodiment, determining the target flame detection result according to the first flame detection result and the second flame detection result includes:

[0127] If the detection accuracy of the flame detector meets the set high accuracy requirement, then when both the first flame detection result and the second flame detection result indicate the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

[0128] So far, completed Figure 6 Structural description of the device shown.

[0129] See Figure 7 , Figure 7 This is a structural diagram of an electronic device provided in an embodiment of the present application. Figure 7 As shown, the hardware structure may include: a processor and a machine-readable storage medium, the machine-readable storage medium storing machine-executable instructions that can be executed by the processor; the processor is used to execute the machine-executable instructions to implement the method disclosed in the above example of this application.

[0130] Based on the same application concept as the above method, an embodiment of the present application also provides a machine-readable storage medium, on which a number of computer instructions are stored. When the computer instructions are executed by a processor, the method disclosed in the above example of the present application can be implemented.

[0131] Exemplarily, the machine-readable storage medium may be any electronic, magnetic, optical, or other physical storage device that may contain or store information, such as executable instructions, data, and the like. For example, the machine-readable storage medium may be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, a storage drive (such as a hard disk drive), a solid-state drive, any type of storage disk (such as a CD, DVD, etc.), or similar storage media, or a combination thereof.

[0132] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A flame detection method, characterized in that: The method is applied to a flame detector and comprises: Collecting flame band data corresponding to the flame detector in a reference gain mode and a non-reference gain mode in a current time window; and determining a reference gain mode corresponding to a next time window based on the flame band data collected in the reference gain mode in the current time window, so as to collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window; Obtain first flame band data; the first flame band data includes flame band data respectively collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window is determined depending on whether the reference gain modes corresponding to the current time window and the next time window are the same; if the reference gain modes corresponding to the current time window and the next time window are the same, then the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window; if the reference gain modes corresponding to the current time window and the next time window are different, then the target gain mode corresponding to the current time window is other gain modes corresponding to the current time window, and the other gain modes are the reference gain modes corresponding to the next time window; obtaining second flame band data; wherein the second flame band data is a subset of the first flame band data; Flame detection is performed using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data.

2. The method according to claim 1, characterized in that The flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points; Determining the reference gain mode corresponding to the next time window based on the flame band data collected under the reference gain mode in the current time window includes: If the reference gain mode of the current time window is the set high gain mode, the number of target data that meet the conditions is counted from the flame band data collected in the reference gain mode within the current time window; the conditions are: the flame values ​​at at least K consecutive sampling points in the current time window are greater than or equal to the set flame upper limit value; or the flame values ​​at at least P consecutive sampling points in the current time window are less than or equal to the set flame lower limit value; K is greater than 1, and P is greater than 1; Determine whether the number is greater than a set counting threshold; if so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to a low gain mode; if not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

3. The method according to claim 1, characterized in that The flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points; Determining the reference gain mode corresponding to the next time window based on the flame band data collected under the reference gain mode in the current time window includes: If the reference gain mode of the current time window is the set low gain mode, performing a specified operation on the flame value at each sampling point in the current time window to obtain an operation result; Determine whether the operation result is less than or equal to a preset value. If so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to a high gain mode; if not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window.

4. The method according to claim 1, wherein The flame detection using the flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data includes: splicing the time domain features and the frequency domain features of the first flame band data to obtain a first splicing feature; The first splicing feature is subjected to data preprocessing and input into a trained first flame detection model to obtain a first flame detection result; the data preprocessing is used to eliminate the difference in frequency domain features caused by time length; Splicing the time domain features and the frequency domain features of the second flame band data to obtain a second splicing feature; inputting the second splicing feature into the trained second flame detection model to obtain a second flame detection result; A target flame detection result is determined based on the first flame detection result and the second flame detection result; the target flame detection result is used to indicate whether a flame exists.

5. The method according to claim 4, characterized in that Determining a target flame detection result based on the first flame detection result and the second flame detection result includes: If the detection accuracy of the flame detector meets the set low accuracy requirement, when any one of the first flame detection result and the second flame detection result indicates the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

6. The method according to claim 4, characterized in that Determining a target flame detection result based on the first flame detection result and the second flame detection result includes: If the detection accuracy of the flame detector meets the set high accuracy requirement, then when both the first flame detection result and the second flame detection result indicate the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

7. A flame detection device, characterized in that: The device is applied to a flame detector and comprises: A determination module is configured to respectively collect flame band data corresponding to the flame detector in a reference gain mode and a non-reference gain mode in a current time window; and determine a reference gain mode corresponding to a next time window based on the flame band data collected in the reference gain mode in the current time window, so as to respectively collect flame band data corresponding to the reference gain mode and the non-reference gain mode in the next time window; A first acquisition module is used to obtain first flame band data; the first flame band data includes flame band data respectively collected by the flame detector under target gain modes corresponding to N time windows; the target gain mode corresponding to any time window is determined depending on whether the reference gain modes corresponding to the current time window and the next time window are the same; if the reference gain modes corresponding to the current time window and the next time window are the same, then the target gain mode corresponding to the current time window is the reference gain mode corresponding to the current time window; if the reference gain modes corresponding to the current time window and the next time window are different, then the target gain mode corresponding to the current time window is other gain modes corresponding to the current time window, and the other gain modes are the reference gain modes corresponding to the next time window; A second obtaining module is configured to obtain second flame band data; the second flame band data is a subset of the first flame band data; The detection module is configured to perform flame detection using a flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data.

8. The device according to claim 7, characterized in that The flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points; Determining the reference gain mode corresponding to the next time window based on the flame band data collected under the reference gain mode in the current time window includes: If the reference gain mode of the current time window is the set high gain mode, the number of target data that meet the conditions is counted from the flame band data collected in the reference gain mode within the current time window; the conditions are: the flame values ​​at at least K consecutive sampling points in the current time window are greater than or equal to the set flame upper limit value; or the flame values ​​at at least P consecutive sampling points in the current time window are less than or equal to the set flame lower limit value; K is greater than 1, and P is greater than 1; Determine whether the number is greater than a set counting threshold; if so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to a low gain mode; if not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window; or, The flame band data collected in the reference gain mode within the current time window includes flame values ​​at multiple different sampling points; Determining the reference gain mode corresponding to the next time window based on the flame band data collected under the reference gain mode in the current time window includes: If the reference gain mode of the current time window is the set low gain mode, performing a specified operation on the flame value at each sampling point in the current time window to obtain an operation result; Determine whether the operation result is less than or equal to a preset value; if so, determine the non-reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window, and the non-reference gain mode of the current time window is set to a high gain mode; if not, determine the reference gain mode corresponding to the current time window as the reference gain mode corresponding to the next time window; and / or, The flame detection using the flame detection model and based on the time domain characteristics and frequency domain characteristics of the first flame band data and the time domain characteristics and frequency domain characteristics of the second flame band data includes: splicing the time domain features and the frequency domain features of the first flame band data to obtain a first splicing feature; The first splicing feature is subjected to data preprocessing and input into a trained first flame detection model to obtain a first flame detection result; the data preprocessing is used to eliminate the difference in frequency domain features caused by time length; Splicing the time domain features and the frequency domain features of the second flame band data to obtain a second splicing feature; inputting the second splicing feature into the trained second flame detection model to obtain a second flame detection result; Determining a target flame detection result based on the first flame detection result and the second flame detection result; the target flame detection result is used to indicate whether a flame exists; and / or, Determining a target flame detection result based on the first flame detection result and the second flame detection result includes: If the detection accuracy of the flame detector meets the set low accuracy requirement, when any one of the first flame detection result and the second flame detection result indicates the presence of flame, determining that the target flame detection result indicates the presence of flame; or, Determining a target flame detection result based on the first flame detection result and the second flame detection result includes: If the detection accuracy of the flame detector meets the set high accuracy requirement, then when both the first flame detection result and the second flame detection result indicate the presence of flame, it is determined that the target flame detection result indicates the presence of flame.

9. An electronic device, characterized in that: include: processor; as well as A memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, which, when executed by a processor, enable the processor to perform the steps of the method according to any one of claims 1 to 6.

Images