Infrared flame detection method and system based on fractional Fourier transform

Abstract

The invention relates to an infrared flame detection method and system based on fractional Fourier transform, belongs to the field of flame detection, and solves the problem that in the prior art, multi-band signal detection processing hardware is more, and time domain features are lacked. The method comprises the following specific steps: acquiring an infrared signal of a main flame detection wave band by using an infrared sensor to obtain a time sequence signal; the time sequence signal is preprocessed; performing specified-order fractional Fourier transform on the pre-processed signal, and extracting amplitude characteristics of the time-frequency domain signal based on a transform result; and obtaining a flame detection result based on the amplitude characteristics and the trained flame judgment model, thereby realizing flame detection based on the single-band infrared signal, reducing hardware and obtaining time-frequency characteristics at the same time.

Description

Technical Field

[0001] The invention relates to the field of flame detection, and in particular to an infrared flame detection method and system based on fractional-order Fourier transform. Background Art

[0002] When a fire occurs in an environment, the flame will generate obvious infrared radiation. Due to the turbulence effect during the flame combustion process, the intensity of these infrared radiation will show flickering characteristics. This change is not irregular. Some studies have shown that the flame flicker frequency is distributed between 3-25Hz, and the main frequency is in the range of 7-12Hz. Figure 2 As shown. And the flame flicker frequency has nothing to do with the flame size and the distance of the local infrared receiving device. Although this signal has the characteristics of a certain frequency range distribution, it is a non-stationary signal.

[0003] Traditional flame judgment algorithms are generally based on Fourier transform, which extracts the frequency domain characteristics of three-band time series signals from 2.8um, 4.3um, and 5.3um multi-band infrared signals, combines the different characteristics of the three-band signals, analyzes the frequency domain characteristics of the flame signal, and finally outputs the judgment result. Among them, each band signal corresponds to the background, flame, and interference bands respectively. Traditional flame detection not only requires the collection of multi-band infrared signals, but also requires a lot of sensor hardware and a large amount of calculation; and Fourier transform can only extract the frequency domain information of the sequence, and the extraction of time-frequency features in the sequence is poor. Summary of the invention

[0004] In view of the above analysis, the embodiments of the present invention aim to provide an infrared flame detection method and system based on fractional Fourier transform, so as to solve the defects of the prior art of using multi-band signal detection and processing hardware and lack of time domain characteristics.

[0005] The purpose of the present invention is mainly achieved through the following technical solutions:

[0006] On the one hand, an embodiment of the present invention provides an infrared flame detection method based on fractional-order Fourier transform, comprising the following steps:

[0007] The infrared sensor is used to collect the infrared signal of the main flame detection band to obtain a time series signal;

[0008] Preprocessing the time series signal;

[0009] Perform a specified order fractional Fourier transform on the preprocessed signal, and extract the amplitude characteristics of the time-frequency domain signal based on the transform result;

[0010] Based on the amplitude characteristics and the trained flame judgment model, a flame detection result is obtained.

[0011] Further, the specified order is predetermined by the following steps:

[0012] Generate a target flame signal, an interference signal and a background signal as detection signals respectively;

[0013] Using infrared sensors to collect the detection signals respectively and perform pre-processing;

[0014] Calculating fractional Fourier transforms of different orders for each preprocessed detection signal;

[0015] Based on the transformation results, the correlation coefficients between the target flame signal and the interference signal and the background signal at different orders are calculated;

[0016] The order corresponding to the minimum correlation coefficient is used as the designated order.

[0017] Furthermore, the different orders are obtained by dividing the transformation order interval of the fractional Fourier transform by step size.

[0018] Furthermore, the amplitude feature includes a support area width feature and a positive value ratio feature.

[0019] Furthermore, the expression of the support area width feature is:

[0020]

[0021] Wherein, find(A) means counting the number of sequence points that satisfy the condition A greater than 0 among N sampling points; f p (u) represents the p-order fractional Fourier transform result of the preprocessed signal.

[0022] Furthermore, the positive value ratio feature is to count the number of points in the transformation result that are greater than a threshold value.

[0023] Furthermore, the preprocessing includes performing a first-order difference on the time series signal.

[0024] Furthermore, the formula of the fractional Fourier transform is:

[0025]

[0026] in, u is the fractional Fourier transform domain; p is the fractional transform order; n is a positive number; σ() represents the unit impulse function.

[0027] Furthermore, the flame judgment model is obtained by training based on a support vector machine algorithm.

[0028] On the other hand, an embodiment of the present invention provides an infrared flame detection system based on fractional-order Fourier transform, comprising:

[0029] Infrared sensor, used to collect infrared signals in the main flame detection band to obtain time series signals;

[0030] A preprocessing module, used for preprocessing the time series signal;

[0031] A fractional Fourier processing module is used to perform a fractional Fourier transform of a specified order on the preprocessed signal, and extract the amplitude characteristics of the time-frequency domain signal based on the transform result;

[0032] The flame judgment module is used to obtain the flame detection result by using the amplitude feature and the trained flame judgment model.

[0033] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0034] 1. The present invention proposes a method for realizing flame detection based only on the main band signal, extracting time-frequency features through fractional Fourier transform processing, and realizing flame detection through a flame detection model, without the need to use different features of multi-band infrared signals, reducing the amount of calculation and making flame detection more direct;

[0035] 2. Because only single-band signals need to be collected, the number of sensors is reduced, making the detection system lighter and saving hardware costs;

[0036] 3. The fractional Fourier transform is used to extract the time-frequency domain features of the infrared sequence, which is suitable for non-stationary signal processing and can more effectively extract the features in the sequence.

[0037] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0039] Figure 1 This is a flow chart of an infrared flame detection method based on fractional-order Fourier transform according to an embodiment of the present invention;

[0040] Figure 2 It is the spectrum distribution diagram of the flame time series of the present invention;

[0041] Figure 3 Schematic diagram of the FrFT transformation result of the flame signal according to an embodiment of the present invention;

[0042] Figure 4 A schematic diagram of the FrFT transformation result of the interference signal match fire according to an embodiment of the present invention;

[0043] Figure 5 A schematic diagram of different transformation orders and correlation coefficients of flame and interference signals according to an embodiment of the present invention;

[0044] Figure 6 It is a schematic diagram of the FrFT transformation results of the flame signal collected at different distances according to an embodiment of the present invention;

[0045] Figure 7 It is a schematic diagram of different transformation orders and correlation coefficients of flame signals collected at different distances according to an embodiment of the present invention;

[0046] Figure 8 is the FrFT transformation result of the background signal in the embodiment of the present invention;

[0047] Fig. 9 Schematic diagram of different transformation orders and correlation coefficients of flame signals and background signals according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0049] Example 1

[0050] A specific embodiment of the present invention discloses an infrared flame detection method based on fractional Fourier transform, such as Figure 1 As shown, the following steps are included:

[0051] Step S1, using an infrared sensor to collect infrared signals in the main flame detection band to obtain a time series signal;

[0052] Step S2, preprocessing the time series signal;

[0053] Step S3, performing a fractional Fourier transform of a specified order on the preprocessed signal, and extracting the amplitude characteristics of the time-frequency domain signal based on the transform result;

[0054] Step S4: obtaining a flame detection result based on the amplitude feature and the trained flame judgment model.

[0055] Through the above method, only by collecting single-band infrared time series signals, performing data processing to extract amplitude features, and using the flame judgment model, the flame detection results are obtained. There is no need to rely on the different characteristics of multi-band infrared signals, which reduces the sensor hardware, makes flame detection more direct, and obtains time-frequency features at the same time.

[0056] Specifically, in step S1, the flame main detection band refers to the main infrared band for determining the generation of flames. Usually, infrared radiation near 4.3 μm is consistent with the radiation characteristics of CO2 gas in the flame, contains a large number of radiation characteristics in oil and gas fires, is the main basis for determining whether a fire has occurred in the environment, and is the flame main detection band for oil and gas fires; according to application needs, infrared signals near 4.6 μm can also be used as the flame main detection band. The specific situation depends on the actual application.

[0057] Specifically, in step S2, the time series signal is preprocessed based on the first-order difference;

[0058] It should be noted that the original sampling value of the main flame detection band changes based on a certain background value, and this background value varies due to the hardware characteristics of the sensor and the environment in which it is located.

[0059] In order to eliminate the influence of background values, the signal needs to be preprocessed. By calculating the first-order difference of the original sampling value sequence, the changes in the sampling values ​​at adjacent moments can be obtained. It is these changes that contain the typical characteristics of flame combustion.

[0060] Specifically, in step S3, a fractional Fourier transform of a specified order is performed on the preprocessed signal, wherein the specified order is predetermined by the following steps:

[0061] Step S31, respectively generating a target flame signal, an interference signal and a background signal as detection signals; wherein the background signal is usually an indoor background under an energy-saving lamp, and the interference signal is generally a match, an artificial light source, etc.;

[0062] Step S32, using an infrared sensor to collect each of the detection signals respectively, and perform preprocessing; wherein the preprocessing is the same as step S2;

[0063] Step S33, calculating fractional Fourier transforms of different orders for each preprocessed detection signal;

[0064] Specifically, for the time series signal f(t), the fractional Fourier transform can be expressed as:

[0065]

[0066] in, u is the fractional Fourier transform domain; p is the fractional transform order; n is a positive number; σ() represents the unit impulse function.

[0067] It should be noted that at different orders, the signal will show different energy distributions. Since the kernel function is periodic and the transformation result is symmetrical, the transformation order is in the interval [0,1] in actual calculations. In particular, when p = 0, it is the original time series signal; when p = 1, it is the result after Fourier transformation. Since the value of the order is a fraction, the number of decimal places and the number of different orders are determined by the specific application.

[0068] Exemplarily, a step size of 0.1 is selected to divide the transformation order, and the different orders are 0.1, 0.2, ..., 0.9. The fractional Fourier transforms of the different orders of the preprocessed target flame signal, the fractional Fourier transforms of the different orders of the preprocessed interference signal, and the fractional Fourier transforms of the different orders of the preprocessed background signal are calculated respectively to obtain transformation results.

[0069] Step S34, based on the transformation result, calculating the correlation coefficients between the target flame signal and the interference signal and the background signal at different orders;

[0070] Step S35: Based on the correlation coefficient and the corresponding fractional Fourier transform order, select the order corresponding to the minimum correlation coefficient as the designated order.

[0071] For example, a fire in a brazier with a diameter of 0.1 m and a distance of 2 m is used to generate a target flame signal, and a match fire is used to simulate an interference signal. Figure 3 , is the FrFT transformation result of the target flame signal, Figure 4 is the FrFT transform result of the interference signal, Figure 5 is the correlation coefficient of the two at different orders. It can be seen that when the transformation order is 1, the correlation coefficient is high; when the transformation order is between [0.5, 0.9], the correlation coefficient is stable below 0.1. This shows that the fractional Fourier transform can better distinguish different feature signals than the Fourier transform; the fractional transform order within a certain interval is more conducive to signal classification. Figure 8 is the FrFT transformation result of the background signal, Fig. 9 is the correlation coefficient between the two at different orders. It can be seen that when the transformation order is between [0.5, 0.8], the correlation coefficient is close to 0. To verify the versatility, flame signals at different distances are selected for comparison, such as Figure 6 , FrFT transformation result of the flame signal collected at a distance of 60cm, Figure 7is the correlation coefficient between the two at different orders. It can be seen that when the transformation order is between [0.5, 0.8], the correlation coefficient is between 0.15, which is much larger than the correlation coefficient between the flame and the interference signal. According to the above calculation results, p = 0.7 corresponding to the minimum value of the correlation coefficient is used as the transformation order for judging the flame.

[0072] Exemplarily, in step S3, a 0.7-fold fractional Fourier transform is performed on the preprocessed signal.

[0073] Assume that the acquisition frequency of the infrared sensor is 4000 Hz, and perform fractional Fourier transform on 4000 input signals each time. The above point selection method ensures that the extracted features have sufficient resolution.

[0074] It should be noted that by observing Figure 3-Figure 9 After the fractional order transformation, the flame signal has several prominent peaks and a higher amplitude relative to the background and interference signals, while the background and interference signals have less obvious peaks and a lower amplitude, which is stable below 50. Therefore, based on the transformation results, the support area width and positive value ratio characteristics of the time-frequency domain signal are extracted as amplitude characteristics, the support width is used to identify the peak width of the flame signal characteristics, and the proportion of the amplitude greater than a certain value in the transformation results is obtained based on the positive value ratio, the amplitude characteristics of the flame signal are identified, and the flame and background and interference signals are effectively distinguished.

[0075] Specifically, the support area width feature refers to the lateral width of the support area; the peak width is reflected by the number of samples whose amplitude of the statistical transformation result is greater than the average amplitude; the calculation formula is:

[0076]

[0077] Wherein, find(A) means counting the number of sequence points that satisfy the condition A greater than 0 among N sampling points; f p (u) represents the p-order fractional Fourier transform.

[0078] The Proportion of Positive Values ​​(PPV) feature usually indicates the proportion of positive values ​​in a data sequence or feature vector. For example, the number of points in the transformation result whose amplitude is greater than the threshold value is counted to obtain the positive value proportion, wherein the threshold value is a value pre-set according to the flame characteristics, and by analyzing the experimental data of the peak amplitude of the transformation results of the flame, background and interference signals, the amplitude of the amplitude range that excludes the background and interference signals as much as possible is selected as the threshold value, and the specific value depends on the application.

[0079] Specifically, in step S4, the flame signal detection result includes two classification situations: flame and non-flame. The amplitude feature is used as the training data set, and the entire training data set is divided into a training set and a validation set, which are used for model training and model performance evaluation respectively; the model is trained by the support vector machine (SVM) algorithm to generate a flame judgment model.

[0080] Compared with the prior art, the present embodiment provides an infrared flame detection method based on fractional Fourier transform, which is based on a single-band signal, extracts time-frequency features by fractional Fourier transform, and uses a machine learning model to implement a flame detection method. On the one hand, the method does not need to rely on the different features of multi-band infrared signals, reduces the amount of calculation and sensor hardware, and makes flame detection more direct; on the other hand, the fractional Fourier transform is used to extract the time-frequency domain features of the infrared sequence, which is suitable for non-stationary signal processing and can more effectively extract features in the sequence.

[0081] Example 2

[0082] Another specific embodiment of the present invention discloses an infrared flame detection system based on fractional-order Fourier transform, comprising:

[0083] Infrared sensor, used to collect infrared signals in the main flame detection band to obtain time series signals;

[0084] A preprocessing module, used for preprocessing the time series signal;

[0085] A fractional Fourier processing module is used to perform a fractional Fourier transform of a specified order on the preprocessed signal, and extract the amplitude characteristics of the time-frequency domain signal based on the transform result;

[0086] The flame judgment module is used to obtain the flame detection result by using the amplitude feature and the trained flame judgment model.

[0087] The system can perform flame detection according to the infrared flame detection method based on fractional Fourier transform described in any one of the schemes in Embodiment 1. The relevant parts are referenced from each other and are not described repeatedly in this embodiment.

[0088] Compared with the prior art, the present embodiment provides an infrared flame detection system based on fractional Fourier transform, which completes flame detection based on single-band infrared signals through infrared sensors, preprocessing modules, fractional Fourier processing modules and flame judgment modules, thereby reducing the number of system sensors, making the detection system more portable and saving hardware costs.

[0089] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0090] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. An infrared flame detection method based on fractional Fourier transform, characterized in that: The steps include: The infrared sensor is used to collect the infrared signal of the main flame detection band to obtain a time series signal; Preprocessing the time series signal; Perform a specified order fractional Fourier transform on the preprocessed signal, and extract the amplitude characteristics of the time-frequency domain signal based on the transform result; Based on the amplitude characteristics and the trained flame judgment model, a flame detection result is obtained.

2. The infrared flame detection method based on fractional Fourier transform according to claim 1 is characterized in that: The specified order is predetermined by the following steps: Generate a target flame signal, an interference signal and a background signal as detection signals respectively; Using infrared sensors to collect the detection signals respectively and perform pre-processing; Calculating fractional Fourier transforms of different orders for each preprocessed detection signal; Based on the transformation results, the correlation coefficients between the target flame signal and the interference signal and the background signal at different orders are calculated; The order corresponding to the minimum correlation coefficient is used as the designated order.

3. The infrared flame detection method based on fractional Fourier transform according to claim 2 is characterized in that: The different orders are obtained by dividing the transformation order interval of the fractional Fourier transform by a step size.

4. The infrared flame detection method based on fractional Fourier transform according to claim 1 is characterized in that: The amplitude characteristics include a support area width characteristic and a positive value ratio characteristic.

5. The infrared flame detection method based on fractional Fourier transform according to claim 4 is characterized in that: The expression of the support area width characteristic is: Wherein, find(A) means counting the number of sequence points that satisfy the condition A greater than 0 among N sampling points; f p (u) represents the p-order fractional Fourier transform result of the preprocessed signal.

6. The infrared flame detection method based on fractional Fourier transform according to claim 4 is characterized in that: The positive value ratio feature is to count the number of points in the transformation result that are greater than a threshold value.

7. The infrared flame detection method based on fractional Fourier transform according to any one of claims 1 to 6, characterized in that: The preprocessing includes performing a first-order difference on the time series signal.

8. The infrared flame detection method based on fractional Fourier transform according to claim 1 is characterized in that: The formula for the fractional Fourier transform is: in, u is the fractional Fourier transform domain; p is the fractional transform order; n is a positive number; σ() represents the unit impulse function.

9. The infrared flame detection method based on fractional Fourier transform according to claim 2 is characterized in that: The flame judgment model is obtained by training based on a support vector machine algorithm.

10. An infrared flame detection system based on fractional Fourier transform, characterized in that: include: Infrared sensor, used to collect infrared signals in the main flame detection band to obtain time series signals; A preprocessing module, used for preprocessing the time series signal; A fractional Fourier processing module is used to perform a fractional Fourier transform of a specified order on the preprocessed signal, and extract the amplitude characteristics of the time-frequency domain signal based on the transform result; The flame judgment module is used to obtain the flame detection result by using the amplitude feature and the trained flame judgment model.

Images