Electromagnetic interference suppression method for gas detector

Abstract

The invention discloses an electromagnetic interference suppression method for a gas detector. A strategy of combining software suppression and physical suppression is adopted. Wherein the software suppression comprises the steps of collecting AD data for Fourier transform, and extracting multi-dimensional features such as a low-frequency / high-frequency energy ratio, a dominant frequency feature, a waveform slope change rate and a peak factor; the interference intensity is quantified through semantic transformation and weighted scoring; determining an interference type according to the score interval; comprise non-interference, low-frequency impact, high-frequency impact, sine-like interference and superimposed interference, and corresponding targeted processing strategies such as forced zeroing, dynamic sliding filtering or slope verification are executed. According to the physical suppression, noise conduction between circuit boards is blocked by adopting a magnetic coupling isolator, and external radio frequency interference is attenuated by adopting a multi-layer filter screen structure of which the aperture is decreased progressively and which is arranged in a staggered manner. According to the invention, the broadband interference of the interphone can be accurately identified and inhibited, and the monitoring precision and reliability of the gas detector in an interphone coexistence environment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gas detector, and particularly relates to an electromagnetic interference suppression method for a gas detector. BACKGROUND

[0002] With the increasing demand for industrial automation and safety production, embedded systems have been widely used in various key monitoring devices, such as gas detectors, temperature sensors, pressure transmitters, etc. The core function of these systems is to realize real-time monitoring and early warning of target parameters through high-precision signal acquisition, conversion and processing. These systems usually need to operate in complex electromagnetic environments such as industrial plants, chemical parks and underground mines. Electromagnetic interference is one of the core problems that threaten the stable operation of these systems. Electromagnetic interference is caused by electromagnetic field radiation or conduction of various electronic devices. According to the type of interference, it can be divided into power frequency interference, radio frequency interference (such as wireless communication device radiation), pulse interference, etc. Among them, radio frequency interference has a wide frequency range, high radiation intensity and multiple propagation paths, which has a significant impact on the signal acquisition link of embedded systems. When radio frequency interference is coupled to the sensor signal link, it will be superimposed on the weak target signal, causing the digital signal collected by the analog-to-digital conversion (ADC) module to be distorted, and then causing data calculation errors, device malfunctions or even system crashes, which seriously affects the reliability and safety of monitoring devices. For gas detectors, their monitoring objects are mostly flammable, explosive or toxic and harmful gases (such as methane, carbon monoxide, hydrogen sulfide, etc.). The device needs to achieve full-scale signal acquisition accuracy, and has high requirements for false alarm rate. Once false alarms occur due to electromagnetic interference, not only will it reduce the credibility of the device (such as ignoring real alarms after frequent false alarms), but also may cause unnecessary production suspension and investigation, resulting in economic losses. If the interference leads to a missed alarm, it will directly threaten the safety of personnel and the safety of the production environment, so the gas detector has high technical requirements for the suppression of electromagnetic interference.

[0003] In the actual application scenarios of portable gas detectors (such as petrochemical plant areas, municipal pipe corridors, and fire rescue sites), workers usually need to use intercoms (for on-site communication and coordination) and detectors (for gas concentration monitoring) at the same time. The distance between the two is usually less than 1 meter, forming a coupling scenario of close-range radio frequency interference source and sensitive monitoring equipment, in which the radio frequency radiation of the intercom becomes the main interference source leading to false alarms of the detector.

[0004] The intercom as a typical wireless communication equipment, its working frequency band mainly covers two types: 1, public intercom frequency band (409-410MHz, power ≤0.5W); 2, industrial professional intercom frequency band (V section 136-174MHz, U section 400-470MHz, power 1-5W). This kind of equipment will produce continuous radio frequency electromagnetic field radiation in the communication process, and its radiation intensity decays with distance, and the radiation signal contains fundamental signal (such as 450MHz) and harmonic signal (such as 900MHz, 1350MHz), forming a wideband interference source.

[0005] The sensor of gas detector (such as electrochemical sensor, catalytic combustion sensor) needs to contact with the outside environment through the gas permeable net to realize the penetration and detection of gas molecules - the gas permeable net cannot shield the radio frequency signal, and the radio frequency radiation of the intercom will enter the inside of the sensor through the gas permeable net, and be received by the sensitive element (such as electrode, catalytic element) of the sensor and converted into weak electric interference signal; the operational amplifier (U1) on the sensor acquisition conversion board needs to amplify the weak target signal output by the sensor to meet the acquisition requirement of ADC, and the radio frequency interference signal coupled by the intercom will be superimposed with the target signal, and after being amplified by the operational amplifier, the amplitude of the interference signal is amplified synchronously, and is comparable to the amplitude of the normal target signal; the ADC module of the AD acquisition single-chip board will convert the amplified target signal, interference signal mixed signal into digital signal, and since the signal processing algorithm of the existing device cannot distinguish the intercom radio frequency interference from the real gas concentration signal, the interference signal is misjudged as normal concentration signal, resulting in abnormal rise of AD value, and finally the concentration display reaches the upper limit of the range, triggering false alarm.

[0006] The existing hardware filtering mainly adds passive filtering elements (such as π type filter, RC low pass filter) or active filtering chip in the circuit to realize the suppression of specific frequency interference; the existing hardware filtering is mainly designed for fixed frequency, and cannot form effective attenuation to the wide frequency interference of 136-470MHz; in order to improve the filtering effect, the hardware filtering will increase the impedance of the signal link, resulting in attenuation of the weak target signal output by the sensor, affecting the ADC acquisition accuracy.

[0007] The existing software filtering mainly uses sliding filtering and mean filtering to suppress interference by smoothing the continuous AD values. The delay time of the sliding filtering is positively correlated with the window size. However, the intercom interference is bursty (for example, lasting for 10-30 seconds during the conversation), and the delay will cause the device to fail to respond to the real gas concentration change in time, and even miss the early leakage warning. The sliding filtering cannot distinguish the AD value mutation caused by the intercom interference from the rapid rise of the AD value caused by the real gas leakage. If the window is increased to suppress the interference, the real leakage signal will be misjudged as noise smoothing processing, resulting in a missed report. The amplitude of the intercom interference dynamically changes with the distance and the conversation power, and belongs to nonlinear interference. The sliding filtering has limited suppression ability for nonlinear interference, and cannot adapt to the dynamic change of the interference amplitude.

[0008] In summary, in the actual application of the portable gas detector, the wideband radio frequency interference of the intercom as the necessary communication equipment in the industrial field will cause the device AD value to be abnormal, the concentration display to rise sharply, and the false alarm to occur frequently through the paths of sensor coupling, signal link amplification, and ADC misacquisition. The existing hardware filtering, shielding isolation, and software filtering schemes cannot effectively suppress the intercom interference due to the defects of poor frequency adaptability, insufficient shielding integrity, delay, and misjudgment risk. Therefore, how to solve the influence of the intercom radio frequency interference on the signal acquisition of the gas detector and improve the stability and reliability of the device in the intercom coexistence scene has become a key technical problem to be solved in the current gas detector field. Therefore, it is necessary to study an electromagnetic interference suppression method for a gas detector. SUMMARY

[0009] In view of this, the purpose of the present application is to provide an electromagnetic interference suppression method for a gas detector. For the use scenario of the coexistence of the intercom and the gas detector, the existing hardware filtering, shielding isolation, and software filtering schemes cannot effectively suppress the intercom interference due to the defects of poor frequency adaptability, insufficient shielding integrity, delay, and misjudgment risk, resulting in the problems of frequent AD value abnormality, concentration display rising sharply, and frequent false alarms of the gas detector.

[0010] To achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows: an electromagnetic interference suppression method for a gas detector, comprising the following steps: S1. Collecting AD data according to unit collection time duration; and pre-processing the input AD data in response to valid AD data input; S2. Performing Fourier transform on the pre-processed AD data, and extracting a feature set based on the AD data; the feature set includes low-frequency energy ratio, high-frequency energy ratio, main frequency peak value, main frequency position, waveform slope change rate, and peak factor; S3. Performing semantic transformation on the feature set to generate a feature data set; S4, assigning corresponding weights to each sub-feature in the feature data set, and obtaining an interference evaluation score through weighting; S5, obtaining an interference type based on the interval in which the evaluation score is located; S6, based on the interference type, performing a corresponding interference processing strategy on the AD data to obtain the final AD data output; S7, outputting the AD data.

[0011] Further, in S1, the AD data is filtered, when the AD data is greater than the threshold value, the acquisition board is triggered to accept the AD data, and in a unit acquisition time, according to the acquisition interval, a set formed by at least N AD data is taken as effective AD data input.

[0012] Further, the low-frequency energy ratio is the percentage of the energy of the AD data in the low-frequency band to the total frequency band energy; the high-frequency energy ratio is the percentage of the energy of the AD data in the high-frequency band to the total energy; the main frequency peak is the peak value corresponding to the maximum amplitude value in the AD data frequency domain amplitude spectrum; the main frequency position is the actual frequency corresponding to the main frequency peak; the waveform slope change rate is the number of times of the slope sign jump in a unit time; the peak factor is the ratio of the peak value to the effective value of the AD data.

[0013] Further, the semantic transformation is based on the original features of the feature set for standardization processing, and different dimensional features are set to the [0, 1] interval; in the standardization process, according to the provided maximum threshold value and minimum threshold value, the standard deviation value of the maximum threshold value and the minimum threshold value is obtained; based on the original feature value, the actual difference value of the original feature value and the experimental minimum threshold value is obtained; the ratio of the actual difference value to the standard deviation value is input to the feature data set.

[0014] Further, the main frequency peak and the main frequency position form a main frequency feature value based on a fusion strategy, and the fusion strategy is: assigning corresponding frequency values to preset frequency regions, and fusing the main frequency peak and the frequency values to form the main frequency feature value.

[0015] Further, the frequency region includes a first frequency band, a second frequency band, a third frequency band, a fourth frequency band, and a fifth frequency band, and each frequency band is provided with a corresponding frequency value; wherein the first frequency band is 0-50Hz, and the frequency value is K1; the second frequency band is 51-100Hz, and the frequency value is K2; the third frequency band is 101-1000Hz, and the frequency value is K3; the fourth frequency band is 1001-3000Hz, and the frequency value is K4; the fifth frequency band is 3001-5000Hz, and the frequency value is K5.

[0016] Further, in S5, the interval is divided into a first sub-interval, a second sub-interval, a third sub-interval, a fourth sub-interval and a fifth sub-interval in order from small to large; the interference levels corresponding to the first sub-interval, the second sub-interval, the third sub-interval, the fourth sub-interval and the fifth sub-interval are low interference, first medium interference, second medium interference, first high interference and second high interference respectively.

[0017] Further, in S6, If it is a low interference level, it means no interference; If it is a first medium interference, further check the low frequency energy ratio, if it is greater than a preset value, it is determined to be a low frequency impact, otherwise it is no interference; If it is a second medium interference, further check the main frequency characteristic value and the peak factor, if they are within a preset range, it is determined to be a sinusoidal-like interference, otherwise it is a low frequency impact; If it is a first high interference, further check the high frequency energy ratio, if it is greater than a preset value, it is determined to be a high frequency impact, otherwise it is a sinusoidal-like interference; If it is a second high interference, further check whether the high frequency energy ratio and the low frequency energy ratio are within a preset range, if they are, it is determined to be interference superposition, otherwise it is a high frequency impact.

[0018] Further, if it is no interference, the AD data is normally output; If it is a low frequency impact, it is marked as invalid data, and the current AD data is forced to be output as 0, and after a continuous setting time, the sampling process is resumed; If it is a high frequency impact, the size N of the sliding filter window is dynamically adjusted according to the high frequency energy ratio; a sliding filter strategy is adopted to smooth the data by calculating the average value of the data in the window; the sliding filter strategy is to attenuate the high frequency component and retain the low frequency component in the frequency domain; If it is a sinusoidal-like interference, the AD data is converted into a concentration value, the concentration change slope within a certain time is calculated, if it is within a normal slope range, the AD data is directly output, otherwise it is marked as sinusoidal-like interference data and does not participate in concentration calculation; If it is superimposed interference, it is first processed according to the high frequency interference mode, and then the filtered AD data is detected to check whether the low frequency energy ratio is greater than a preset value, if it is, it is forced to be zero, otherwise it is normally output.

[0019] Further, it further includes a physical suppression strategy, the physical suppression strategy includes a magnetic coupling isolator and a filter screen for being placed at the entrance of the detector; the magnetic coupling isolator is arranged between a power supply mainboard and a sensor acquisition board; the magnetic coupling isolator is used to block high frequency noise coupling; the filter screen includes multiple layers of filter screens with different apertures; the multiple layers of filter screens are stacked together according to the principle that the apertures are from large to small and from top to bottom.

[0020] Further, corrugated engagement structures are arranged at intervals where the filter screen is in contact with the detector housing, for increasing the conductive contact; and the aperture direction of each layer of filter screen is arranged in a staggered manner.

[0021] The beneficial effects of the above technical solution are: for the intercom radio frequency interference coupled through the sensor air permeable net, the AD value is abnormally high after the operational amplifier amplification is collected as a normal signal, which leads to abnormal concentration false alarm, the application proposes an electromagnetic interference suppression method taking software suppression as the core and physical suppression as the auxiliary.

[0022] In terms of software suppression, the application extracts multiple features in frequency domain (high and low frequency energy distribution, main frequency characteristics) and time domain (slope change, peak value characteristics) through Fourier transform, fully captures the performance of different interferences, avoids the misjudgment or omission caused by the traditional software relying on single signal amplitude recognition. At the same time, the features are standardized, the influence of different dimensions is eliminated, through the multi-frequency interval feature fusion strategy, the interference evaluation score can accurately quantify the interference strength, replacing the traditional single threshold judgment rough mode.

[0023] For different types of interference, corresponding processing strategies are formulated, and the specific contents are as follows: Low frequency impact: fast zeroing to prevent false alarm: for the low frequency impact of AD value first falling and then rising, the invalid marking and forced zeroing strategy is adopted, the low frequency interference component is cut off from the frequency domain, and the interference residual problem caused by the traditional single data discard is avoided.

[0024] High frequency impact: dynamic filtering to ensure real-time: according to the high frequency interference intensity, the size of the sliding filter window is dynamically adjusted, and the data in the window is allocated weight according to the time distance, which realizes high frequency interference attenuation and avoids the monitoring delay caused by fixed window. Sinusoidal interference: slope verification true or false: after the AD value is converted into concentration, the change slope is calculated, compared with the normal signal slope range, the sinusoidal interference is effectively classified with the real concentration change, and the problem of traditional smooth filtering misfiltering normal signal is avoided.

[0025] Interference superposition: priority processing to stabilize output: for the multi-interference superposition scene, the processing priority is determined according to the interference influence degree, the main interference is suppressed first and then the residual interference is investigated, the output signal is stabilized, and the defects that the traditional single strategy cannot completely eliminate the composite interference are avoided.

[0026] In terms of physical suppression, the application has a magnetic coupling isolator in series between the power mainboard and the sensor acquisition board, which transmits signals through magnetic field coupling to avoid the interference crosstalk problem of traditional conductive / capacitive coupling, effectively blocks the conduction of wideband noise between modules, while ensuring the real-time of signal transmission and not affecting the monitoring efficiency. At the same time, by using a plurality of layers of filter screens with decreasing aperture vertically stacked, and each layer of aperture direction staggered arrangement, the radio frequency signal propagation path is prolonged, and the double attenuation effect is formed by cooperating with the reflection and wave absorbing material on the surface of the filter screen, which effectively weakens the external wideband radio frequency interference.

[0027] Therefore, the application uses physical suppression to block interference conduction from the source and attenuate interference energy from the outside to reduce interference into the signal link in two dimensions, software suppression to accurately identify the type of interference and target matching processing strategy to eliminate the influence of residual interference in two links, which can solve the defects of existing hardware filter wideband adaptation, shielding and isolation difficult to balance ventilation and anti-interference, and software filter delay and misjudgment. At the same time, the application takes into account the miniaturization and low power consumption requirements of portable detectors, and can stably adapt to the safety monitoring needs of industrial scenes such as petrochemical industry and fire rescue. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 The logic block diagram of the method for implementing the application; Figure 2 Feature extraction and processing; Figure 3 Interference type determination logic flowchart; Figure 4 Processing flowchart of the application; Figure 5 Implementation structure diagram of the filter screen; Reference signs: 1, first layer filter screen; 2, second layer filter screen; 3, third layer filter screen; 4, corrugated occlusion structure; 5, conductive rubber gasket. DETAILED DESCRIPTION

[0029] The application will be further described in detail below in combination with the drawings and specific embodiments: The embodiment 1 aims to provide an electromagnetic interference suppression method for a gas detector, relates to the field of embedded system electromagnetic compatibility (EMC), and focuses on the signal acquisition stability of a portable gas detector in a complex electromagnetic environment, in particular to the device false alarm defect caused by the radio frequency interference of a speakerphone, belongs to the anti-interference technology category of industrial safety monitoring equipment, and is applicable to the industrial scene where the portable gas detector and the speakerphone coexist. The wideband radio frequency interference of the speakerphone is coupled through the sensor air permeable net, is amplified by an operational amplifier, is mis-collected as a normal signal by an ADC, causes the AD value to abnormally rise, causes the concentration to be falsely reported, causes the safety management to be chaotic, causes the device availability to be reduced, causes the operation and maintenance cost to be increased, and cannot be effectively suppressed by the existing hardware filtering, shielding isolation and software filtering. Based on this, the embodiment provides an electromagnetic interference suppression method based on improvement of the compatibility of the speakerphone and the gas detector.

[0030] As shown in the specific implementation Figure 1 The electromagnetic interference suppression method for the gas detector is mainly explained by using a software suppression method, and includes the following steps. S1, AD data is collected according to a unit collection time, and in response to valid AD data input, the input AD data is preprocessed.

[0031] During data collection, the AD data is first filtered, when the AD data is greater than a threshold value, the AD data is triggered to be accepted by a collection board, and in the unit collection time, according to a collection interval, a set formed by at least N AD data is taken as valid AD data input. In specific implementation, the unit collection time is set to 1 second, the collection interval is 100 microseconds, when single AD data >-2, the sensor collection board triggers the data to be temporarily stored in a cache area; when ≥100 valid AD data (N=100) are collected in 1 second, it is determined that the valid AD data input is started, and the preprocessing is started; if the valid data is <100 in 1 second, the current cache data is discarded, and the collection is restarted.

[0032] The average value and the standard deviation of the 100 AD data stored in the sensor collection board are calculated by using the RC filter circuit on the sensor collection board, the abnormal data (regarded as extreme interference) is removed, the previous valid data is replaced, and finally 80-100 preprocessed data are reserved for subsequent steps.

[0033] S2, as Figure 2As shown in the Chinese example, this embodiment performs a Fourier transform on the preprocessed AD data and extracts a feature set based on the AD data. The feature set includes low-frequency energy ratio, high-frequency energy ratio, peak value of the dominant frequency, dominant frequency position, waveform slope change rate, and peak factor. On the sensor acquisition board, the filtered AD value is read via a serial port, and the AD value is monitored and analyzed in real time. The AD value is the raw data acquired by the sensor. By analyzing the changing characteristics of the AD value, different types of electromagnetic interference can be identified. From the perspective of frequency domain analysis, the AD value can be regarded as the sampling point of the signal in the time domain, and its frequency domain characteristics reflect the frequency components and distribution of the signal. By performing a Fourier transform on the AD value, it can be converted from the time domain to the frequency domain, thereby allowing for a more intuitive observation of the signal's frequency characteristics and providing a basis for subsequent interference identification and processing.

[0034] In practice, 256 consecutive AD data points after preprocessing are taken (if there are fewer than 256, the last valid data point is used to complete the set). A Fast Fourier Transform is performed using the microcontroller's built-in FFT library to obtain the frequency domain sequence F(k). Only the first 128 frequency points are retained, and the amplitude spectrum is calculated. With power spectrum .

[0035] Low-frequency energy ratio is The low-frequency band is defined as 0~100Hz; the low-frequency energy ratio is the percentage of the energy in the low-frequency band of the AD data relative to the total energy of the frequency band. Calculate the power in this frequency band: The specific time formula for total power is: The formula for calculating the low-frequency energy ratio is: .

[0036] for The high-frequency band is defined as 1kHz~5kHz, and the high-frequency energy ratio is the percentage of the total energy of the AD data in the high-frequency band. Calculate the power in this frequency band: The specific time formula for total power is: The formula for calculating the low-frequency energy ratio is: .

[0037] Peak clock speed The peak value of the main frequency is the peak value corresponding to the maximum amplitude value in the frequency domain amplitude spectrum of the AD data; the specific calculation method is to take the amplitude spectrum. The maximum value; that is .

[0038] Main frequency position The main frequency position is the actual frequency corresponding to the main frequency peak, and the specific calculation method is to find the corresponding frequency point index , and convert the actual frequency . .

[0039] Waveform slope change rate (S): the waveform slope change rate is the number of times of sign change of the slope of AD data per unit time; the specific calculation method is: taking the preprocessed AD data in 1 second, the formula for calculating the slope of adjacent two points is: The number of times of sign change of the slope N is counted, and then (unit: times / second, example: 12 times of change).

[0040] Peak factor (C): the peak factor is the ratio of the peak value of AD data to the effective value, and the specific calculation method is: calculating the peak value of AD data in 1 second ; the effective value ; then .

[0041] S3, performing semantic transformation on the feature set to generate a feature data set; In this embodiment, the semantic transformation is based on the standardized processing of the original features of the feature set, and different dimensional features should be set to the interval [0, 1]; in the standardization process, the standard deviation value of the maximum threshold and the minimum threshold is obtained according to the provided maximum threshold and minimum threshold; based on the original feature value, the actual difference value of the original feature value and the experimental minimum threshold is obtained; the ratio of the actual difference value to the standard deviation value is input to the feature data set.

[0042] S4, each sub-feature in the feature data set is assigned a corresponding weight, and the interference evaluation score is obtained by weighting; in specific implementation, the frequency region includes a first frequency band, a second frequency band, a third frequency band, a fourth frequency band and a fifth frequency band, and each frequency band is provided with a corresponding frequency value; wherein the first frequency band is 0-50Hz, and the frequency value is K1; the second frequency band is 51-100Hz, and the frequency value is K2; the third frequency band is 101-1000Hz, and the frequency value is K3; the fourth frequency band is 1001-3000Hz, and the frequency value is K4; the fifth frequency band is 3001-5000Hz, and the frequency value is K5.

[0043] In practice, the dominant frequency peak and dominant frequency position are used to form a dominant frequency feature value based on a fusion strategy. The fusion strategy involves assigning corresponding frequency values ​​to a preset frequency region and fusing the dominant frequency peak and frequency values ​​to form the dominant frequency feature value. If there is only one valid peak in the frequency domain, the frequency value of the corresponding region is taken. If multiple valid peaks exist, the amplitude proportion weight is calculated; and the corresponding weight is weighted with the corresponding frequency value to obtain the final value. .

[0044] For example, in the first frequency band (0~50Hz), K1=0.2; in the second frequency band (51~100Hz), K1=0.5; in the third frequency band (101~1000Hz), K3=0.8; in the fourth frequency band (1001~3000Hz), K4=1.2; in the fifth frequency band (3001~5000Hz), K5=1.5; if the frequency domain has only one effective peak value (amplitude ≥ 0.7× ); It belongs to the fourth frequency band; therefore .

[0045] If multiple effective peak values ​​exist, such as f1=80Hz, P1=560; f2=1250Hz, P2=800; calculate the amplitude proportion weights: P1 / (P1+P2)=560 / 1360≈0.412; P2 / 1360≈0.588; then This embodiment calculates based on a single effective peak value. = / K5.

[0046] S5, such as Figure 3 As shown, the interference type is obtained based on the interval of the evaluation score; the interval is divided into five partitions in ascending order: first partition, second partition, third partition, fourth partition, and fifth partition; the interference levels corresponding to the first partition, second partition, third partition, fourth partition, and fifth partition are low interference, medium interference, medium interference, high interference, and high interference, respectively.

[0047] In specific real-time terms, such as Defined as the final score, if ; then there is no interference; if This is the first type of interference; if This is the second type of interference; if ; then it is the highest level of interference; if This is the second highest level of interference. In practical implementation, for example... : ; ; .

[0048] S6, based on the interference type, performing corresponding interference processing strategy on the AD data to obtain final AD data output; The specific interference type determination strategy in the implementation of the embodiment is as follows: If it is a low interference level, it means no interference; if it is a first medium interference, further check the low frequency energy ratio, if it is greater than a preset value, it is determined to be a low frequency impact, otherwise it is no interference; if it is a second medium interference, further check the main frequency characteristic value and the peak factor, whether in a preset range, if yes, it is determined to be a sinusoidal interference, otherwise it is a low frequency impact; if it is a first high interference, further check the high frequency energy ratio, if it is greater than a preset value, it is confirmed to be a high frequency impact, otherwise it is determined to be a sinusoidal interference; if it is a second high interference, further check whether the high frequency energy ratio and the low frequency energy ratio are in a preset range, if yes, it is determined to be interference superposition, if not, it is determined to be a high frequency impact.

[0049] S7, as shown in Figure 4 AD data; in specific implementation, the specific interference processing strategy is as follows: If it is no interference, the AD data is normally output; if it is a low frequency impact, it is marked as invalid data, and the current AD data is forced to be output as 0, and after a continuous setting time, the sampling process is resumed.

[0050] This interference type is manifested as that the AD value first decreases from zero point and then rises. In time domain, this interference can be regarded as a low frequency impact signal, and the amplitude and phase thereof change significantly in a short time. The specific performance is as follows: in the decreasing phase, the AD value quickly decreases from zero point to a negative value; in the rising phase, the AD value slowly rises from the negative value to zero point or a positive value.

[0051] From the perspective of frequency domain analysis, this interference type can be manifested as mutation of low frequency component. The power spectral density thereof has a high value in a low frequency band (such as less than 100 Hz), and changes significantly in a short time. Through Fourier transform, the time domain signal of this interference can be converted into a frequency domain signal.

[0052] The mathematical model can be expressed as: Wherein: A is the amplitude of the interference; is a unit step function; T is the duration of the interference.

[0053] Fourier transform is performed on the above time domain signal to obtain a frequency domain representation: This frequency domain representation shows the characteristics of the signal in the low frequency band, especially in Nearby, the amplitude of the signal will increase significantly. This low-frequency impulse signal appears as a low-frequency peak in the frequency domain, whose width is inversely proportional to T.

[0054] By monitoring the low-frequency component in the frequency domain, especially the amplitude change near T, this interference can be identified. If a significant low-frequency peak appears in the frequency domain, and its duration is consistent with T.

[0055] For such interference, it is judged whether it is interference, and if it is interference, the current sampling value is abandoned to avoid false positives. From the perspective of frequency domain analysis, the forced zero strategy is equivalent to truncating the signal in the low-frequency band, thereby eliminating the low-frequency interference component. By setting appropriate threshold and duration T, this type of interference can be effectively identified and processed.

[0056] If it is a high-frequency impulse, the size of the sliding filter window N is dynamically adjusted according to the high-frequency energy ratio; a sliding filter strategy is adopted to smooth the data by calculating the average value of the data in the window; the sliding filter strategy is to attenuate the high-frequency component in the frequency domain, and retain the low-frequency component.

[0057] This type of interference is characterized by irregular fluctuations in AD values, with a very high rate of change. In the time domain, this type of interference can be regarded as a high-frequency noise signal, whose amplitude and frequency change rapidly and has no obvious periodicity. From the perspective of frequency domain analysis, this type of interference may appear as a random distribution of high-frequency components. Its power spectral density has a high value in a wide frequency range, usually concentrated in the high-frequency band (such as greater than 1 kHz). Through Fourier transform, the time domain signal of this type of interference can be converted into a frequency domain signal, whose mathematical model can be expressed as where A is the amplitude of the interference, is a white noise signal whose power spectral density is uniformly distributed at all frequencies.

[0058] The power spectral density of a white noise signal is uniformly distributed at all frequencies, i.e. This indicates that in the frequency domain, the power spectral density of this type of interference is constant at all frequencies, especially in the high-frequency band. By monitoring the high-frequency component in the frequency domain, especially the power spectral density in the high-frequency band, this type of interference can be identified. If uniform distribution of high-frequency components appears in the frequency domain, and its power spectral density is significantly higher than that of normal signals.

[0059] The specific processing strategy adopts a sliding filter strategy to smooth the data by calculating the average value of the data in the window, thereby reducing noise or interference. From the perspective of frequency domain analysis, the sliding filter strategy is equivalent to attenuating the high-frequency component in the frequency domain, and retaining the low-frequency component. By selecting an appropriate sliding window size N, the filtering effect and real-time performance can be balanced, effectively filtering out high-frequency interference.

[0060] If it is a sinusoidal-like interference, the AD data is converted into concentration values, the concentration change slope in a certain time is calculated, if it is within the normal slope range, the AD data is directly output, otherwise it is marked as a sinusoidal-like interference data and does not participate in concentration calculation; If it is a superimposed interference, it is first processed in the manner of high-frequency interference, and then the filtered AD data is detected to check whether the low-frequency energy ratio is greater than a preset value, if yes, it is forced to zero, otherwise the AD data is normally output.

[0061] This type of interference is characterized by AD value changes similar to a sinusoidal function, similar to normal ventilation signals. In the time domain, this interference can be regarded as a sinusoidal signal with similar amplitude and frequency to normal signals, but there may be phase shift or amplitude change; its mathematical model can be expressed as: ; Wherein: A is the amplitude of the interference; is the frequency of the interference; is the phase of the interference.

[0062] The Fourier transform of the above time domain signal is obtained: Wherein, is the frequency of the signal, is the Dirac delta function. This shows that in the frequency domain, this interference is characterized by two discrete frequency components, located at = and =- By monitoring the specific frequency components in the frequency domain, especially the frequency components close to the normal signal frequency, this interference can be identified. If there are significant discrete frequency components in the frequency domain, and their frequencies are close to the normal signal frequency.

[0063] The embodiment aims to solve the problem that the portable gas detector is used in an industrial scene coexisting with a intercom, the radio frequency interference of the intercom is coupled through the sensor air permeable net, amplified by an operational amplifier, and then miscollected as a normal signal by the ADC, resulting in abnormal surge of AD value and concentration false alarm. Therefore, a software suppression based electromagnetic interference suppression method is proposed.

[0064] The specific idea is: first, collect AD data per unit time, screen effective data, and then preprocess by removing outliers; then, perform Fourier transform on the preprocessed data, extract six types of features including low-frequency energy ratio, high-frequency energy ratio, main frequency peak value, main frequency position, waveform slope change rate, and peak factor to analyze interference; then, standardize the features to a fixed interval to eliminate dimensional differences, then preset multiple frequency intervals and corresponding feature values, fuse main frequency related features and assign weights to each feature, and calculate the interference evaluation score; then, divide five subareas according to the score to correspond to different interference levels, and combine the core features to determine the specific interference type (no interference, low-frequency impact, high-frequency impact, sinusoidal interference, and superimposed interference) twice. For different interference types, execute exclusive processing strategies (no interference normal output, low-frequency impact forced zero and recovery sampling, high-frequency impact dynamic adjustment of sliding filter window, sinusoidal interference verification of concentration change slope, and superimposed interference filtering high frequency first and then checking low frequency whether to zero). Finally, output the processed effective AD data to accurately identify and suppress the radio frequency interference of the intercom, and improve the signal acquisition stability of the detector in this scene.

[0065] In embodiment 2, a physical suppression strategy is further adopted.

[0066] The specific physical suppression strategy includes a magnetic coupling isolator and a filter screen placed at the entrance of the detector; the magnetic coupling isolator is arranged between the power mainboard and the sensor acquisition board; and the magnetic coupling isolator is used to block high-frequency noise coupling. In this embodiment, the circuit is divided into a power mainboard (battery powered, low noise current stabilization, and avoiding power line interference) and a sensor acquisition board (responsible for signal acquisition and preprocessing), and the two modules are arranged separately to reduce electromagnetic coupling.

[0067] A double-layer PCB board is adopted, 50% of the left side is a power mainboard area (area A), responsible for acquiring sensor signals and performing preliminary processing, and through physical isolation and signal isolation technology, electromagnetic interference with the power mainboard is reduced, and 50% of the right side is a sensor acquisition board area (area B), responsible for acquiring sensor signals and performing preliminary processing, and through physical isolation and signal isolation technology, electromagnetic interference with the power mainboard is reduced, and a 5mm wide isolation strip (without copper foil and wiring) is reserved between area A and area B to avoid parasitic capacitance coupling of the circuits in the two areas.

[0068] Component positions: the LDO and the battery interface are located in the upper left corner of area A, the power management chip is adjacent to the battery interface; the sensor and the operational amplifier are located in the upper right corner of area B, the ADC and the single-chip microcomputer are located in the lower right corner of area B; the magnetic coupling isolator (ADUM1201) is close to the isolation strip between area A and area B, the primary pins (pins 1 and 2) are welded at the LDO output end of area A, and the secondary pins (pins 3 and 4) are welded at the ADC input end of area B, forming a signal transmission path of the power mainboard, the magnetic coupling, and the sensor acquisition board.

[0069] The battery power supply system has the characteristics of low noise and high stability, can provide stable power supply for the circuit system, and effectively suppress the power supply noise. In the signal transmission between the power supply main board and the sensor acquisition board, a magnetic coupling isolator is used to block high-frequency noise coupling. The magnetic coupling isolator realizes signal transmission by magnetic field coupling, and has the characteristics of high isolation and low transmission delay. Through the application of the magnetic coupling isolator, the stable transmission of signals between modules is ensured, and electromagnetic interference is reduced.

[0070] As shown in Figure 5 The filter screen includes multiple layers of filter screens with different pore sizes; the multiple layers of filter screens are stacked together according to the principle of from large to small and from top to bottom. The corrugated occlusion structure is arranged at the place where the filter screen contacts the detector shell, for increasing the conductive contact.

[0071] In specific implementation, the present embodiment uses three layers of filter screens with different pore sizes, and realizes multi-stage attenuation of electromagnetic waves through staggered arrangement.

[0072] Pore size design: the first layer of filter screen 1 has a pore size of 0.5 mm, for preliminary attenuation of high-frequency electromagnetic waves. The second layer of filter screen 2 has a pore size of 0.3 mm, for further attenuation of medium-frequency electromagnetic waves. The third layer of filter screen 3 has a pore size of 0.1 mm, for deep attenuation of low-frequency electromagnetic waves. Inclination design: the pore size direction of each layer of filter screen is staggered at 30°, increasing the propagation path of electromagnetic waves and improving the attenuation effect.

[0073] A conductive material is arranged on the filter screen, and in specific implementation, a conductive material (such as silver / copper plating) is coated on the surface of the filter screen, to enhance the electromagnetic shielding performance. An electromagnetic wave absorbing material (such as ferrite or carbon-based composite material) is additionally arranged on the surface of the filter screen, to further absorb electromagnetic wave energy.

[0074] Meanwhile, a corrugated occlusion structure 4 is designed on the contact surface between the filter screen and the shell, to increase multi-point conductive contact and ensure good electromagnetic continuous grounding. Conductive glue or conductive rubber gasket 5 is used to fill the gap, to ensure the integrity of electromagnetic shielding.

[0075] Therefore, the present embodiment divides the functional modules into a main board and a sensor acquisition board in a modular isolation manner, ensures the stable transmission of signals between modules through the magnetic coupling isolator, and reduces electromagnetic interference. Meanwhile, the electromagnetic interference is reduced through the multi-layer filter screen structure design of the sensor air-permeable screen, the increase of conductive wave-absorbing materials, and the optimization of the filter screen structure.

Claims

1. A method for suppressing electromagnetic interference in a gas detector, characterized in that: Includes the following steps: S1. Continuously collect AD data according to the unit collection time; in response to valid AD data input, preprocess the input AD data; S2. Perform Fourier transform on the preprocessed AD data and extract the feature set based on the AD data; the feature set includes low-frequency energy ratio, high-frequency energy ratio, peak value of the main frequency, position of the main frequency, rate of change of waveform slope, and peak factor; S3. Perform semantic transformation on the feature set to generate a feature dataset; S4. Assign corresponding weights to each sub-feature in the feature dataset, and obtain the interference evaluation score by weighting. S5. Based on the interval where the evaluation score is located, the interference type is obtained; S6. Based on the type of interference, perform the corresponding interference processing strategy on the AD data to obtain the final AD data output; S7, Output AD data.

2. The electromagnetic interference suppression method for a gas detector according to claim 1, characterized in that: In S1, the AD data is filtered. When the AD data is greater than the threshold, the acquisition board is triggered to accept the AD data. Within the unit acquisition time, according to the acquisition interval, a set with at least N AD data is used as the valid AD data input.

3. The electromagnetic interference suppression method for a gas detector according to claim 1, characterized in that: The semantic transformation is based on the standardization of the original features of the feature set, and features of different dimensions should be set to the [0,1] interval; during the standardization process, the standard deviation between the maximum threshold and the minimum threshold is obtained according to the provided maximum threshold and minimum threshold. Based on the original feature values, the actual difference between the original feature values ​​and the experimental minimum threshold is obtained; Input the ratio of the actual difference to the standard deviation into the feature dataset.

4. The electromagnetic interference suppression method for a gas detector according to claim 3, characterized in that: The main frequency peak and the main frequency position are used to form a main frequency feature value based on a fusion strategy. The fusion strategy is to assign a corresponding frequency value to a preset frequency region and fuse the main frequency peak and the frequency value to form the main frequency feature value.

5. The electromagnetic interference suppression method for a gas detector according to claim 1, characterized in that: The frequency range includes a first frequency band, a second frequency band, a third frequency band, a fourth frequency band, and a fifth frequency band, each with a corresponding frequency.

6. The electromagnetic interference suppression method for a gas detector according to claim 5, characterized in that: In S5, the interval is divided into five partitions in ascending order: the first partition, the second partition, the third partition, the fourth partition, and the fifth partition. The interference levels corresponding to the first partition, the second partition, the third partition, the fourth partition, and the fifth partition are low interference, medium interference, medium interference, high interference, and high interference, respectively.

7. The electromagnetic interference suppression method for a gas detector according to claim 6, characterized in that: In S6, If it is a low interference level, it means there is no interference; If it is the first type of interference, further check the low-frequency energy ratio. If it is greater than the preset value, it is determined to be a low-frequency impact; otherwise, it is no interference. If it is the second type of interference, further check whether the main frequency characteristic value and peak factor are within the preset range. If they are, it is determined to be sinusoidal interference; otherwise, it is low-frequency impact. If it is the highest interference, further check the high-frequency energy ratio. If it is greater than the preset value, it is confirmed as a high-frequency impulse; otherwise, it is determined as a sinusoidal interference. If it is the second highest interference, further check whether the high-frequency energy ratio and low-frequency energy ratio are within the preset range. If they are, it is determined to be interference superposition; if not, it is determined to be high-frequency impulse.

8. The electromagnetic interference suppression method for a gas detector according to claim 7, characterized in that: If there is no interference, the AD data will be output normally; If it is a low-frequency impulse, it is marked as invalid data, and the current AD data is forced to be output as 0. After a set time, the sampling process is resumed. If it is a high-frequency impact, the size N of the sliding filter window is dynamically adjusted according to the high-frequency energy ratio; a sliding filter strategy is adopted to smooth the data by calculating the average value of the data within the window; the sliding filter strategy is to attenuate the high-frequency components in the frequency domain and retain the low-frequency components. If the interference is sinusoidal, the AD data is converted into concentration values, and the slope of the concentration change over a specific time period is calculated. If it is within the normal slope range, the AD data is output directly; otherwise, it is marked as sinusoidal interference data and is not included in the concentration calculation. If it is superimposed interference, it is first processed as high-frequency interference. Then the filtered AD data is detected to check whether the low-frequency energy ratio is greater than the preset value. If it is, it is forced to zero; otherwise, the AD data is output normally.

9. The electromagnetic interference suppression method for a gas detector according to any one of claims 1-8, characterized in that: It also includes physical suppression strategies, which include a magnetic coupler isolator and a filter for placement at the detector inlet; the magnetic coupler isolator is located between the power supply board and the sensor acquisition board; it is used to block high-frequency noise coupling; the filter includes multiple layers of filters with different pore sizes; the multiple filters are stacked together according to the principle of pore size from large to small and from top to bottom.

10. The electromagnetic interference suppression method for a gas detector according to claim 9, characterized in that: A corrugated interlocking structure is provided at intervals where the filter screen contacts the detector housing to increase conductive contact; and the pore sizes of each filter screen are arranged in an alternating pattern.

Images