Interference signal noise elimination method in ambient air non-dispersive infrared monitoring system

Abstract

The present application discloses a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system comprises the steps of eliminating high-frequency noise in a pre-processing process and eliminating low-frequency noise through a sub-band structure of a filter group in a post-processing process. Eliminating high-frequency noise in the pre-processing process comprises the steps of: converting an optical signal into an initial digital signal, wherein each of the initial digital signals comprises an output signal of a reference cell and an output signal of a measuring cell, and the output signal of the reference cell and the output signal of the measuring cell are a pair of pulse signals; classifying a plurality of the initial digital signals; and eliminating high-frequency noise by performing a moving average on the pulse signals of the classified initial digital signals based on a preset pulse signal baseline.

Description

Technical Field

[0001] The present application relates to the field of gas analysis, and in particular to a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system. Background Art

[0002] With the implementation of national environmental protection policies over the years, the government and the public have placed increasingly high demands on environmental quality. As the concentration of greenhouse gases (such as CO, CO2, CH4, and N2O) in ambient air continues to rise year by year, measurement technology based on non-dispersive infrared (ND) analysis has been widely used and has made significant progress. To enable the measurement of low-concentration pollutants and the simultaneous real-time measurement of multiple gas components, a multi-gas filter correlation wheel (multi GFC wheel) based on ND technology has been developed and applied. During gas analysis using ND technology, signal detection algorithms are required to distinguish between the various measured signals. However, the measured signals are very low-level analog signals, and thus, additive noise can cause numerous problems during signal analysis. Therefore, analog processing units for processing analog signals require stable, low-drift, and low-noise amplifiers. Digital signal processing units for processing digital signals must consider noise cancellation and signal separation tailored to the signal characteristics.

[0003] Among noise cancellation methods, adaptive methods that do not rely on signal or noise characteristics, methods using wavelet transforms, independent component analysis (ICA), and blind methods have all been extensively studied. Adaptive noise cancellation methods are typically most effective when measuring a reference signal and a primary signal using two or more sensors. However, when using only a single sensor, noise cancellation performance degrades significantly. Furthermore, ICA and blind methods require significant computational effort, hindering their practical hardware implementation in the development of multi-pollutant measurement equipment requiring real-time processing. Wavelet transform-based noise cancellation methods utilize finite-interval mother wavelets to effectively decompose and synthesize signals. Compared to Fourier transform-based noise cancellation methods, these methods offer the advantage of multi-resolution signal decomposition, enabling enhanced resolution in both the frequency and time domains. However, the complexity of wavelet functions makes hardware implementation of wavelet-based noise cancellation methods more challenging. Summary of the Invention

[0004] One advantage of the present application is that it provides a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system, wherein the method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system improves the accuracy and precision of noise elimination at the algorithm level, provides algorithm support for gas analysis equipment based on non-dispersive infrared analysis, and has relatively loose hardware requirements, which facilitates the implementation of hardware adapted thereto.

[0005] Another advantage of the present application is that it provides a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system, wherein the method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system specifically eliminates different noises during the pre-processing and post-processing processes, thereby improving the progress and accuracy of non-dispersive infrared measurement.

[0006] Another advantage of the present application is that it provides a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system, wherein the method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system eliminates rapidly changing high-frequency noise in the preprocessing process of time domain characteristics, and can obtain strong resistance to rapidly changing high-frequency noise. In the post-processing process of considering the power ratio through the sub-band structure of the filter group, the sub-band noise elimination technology of the filter group is used to eliminate the slowly changing low-frequency noise, which can improve the resistance to slowly changing low-frequency noise.

[0007] According to one aspect of the present application, a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system is provided, which comprises the steps of:

[0008] Eliminating high-frequency noise in a preprocessing process includes the following steps: converting an optical signal into an initial digital signal, wherein each of the initial digital signals includes an output signal of a reference cell and an output signal of a measuring cell, and the output signal of the reference cell and the output signal of the measuring cell form a pair of pulse signals; classifying a plurality of the initial digital signals; and eliminating high-frequency noise by performing a moving average on the pulse signals of the classified initial digital signals based on a preset pulse signal baseline; and

[0009] In the post-processing process, low-frequency noise is eliminated through the sub-band structure of the filter bank.

[0010] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system described in the present application, the optical signal is converted into an initial digital signal, including the steps of: converting the optical signal into an analog electrical signal through a non-dispersive infrared detector; amplifying the analog electrical signal through a preamplifier to form an amplified analog signal; and converting the amplified analog signal through an analog-to-digital converter to form the initial digital signal.

[0011] In one embodiment of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system described in the present application, multiple initial digital signals are classified, including the steps of: generating a trigger signal based on the initial digital signal; counting the number of pulses in the continuous pulse signal based on the threshold of the trigger signal; wherein, if the number of pulses in the pulse signal of the (nl)th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline, the number of target pulses is recorded as 0; if the number of pulses in the pulse signal of the (nl)th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline and the number of target pulses is recorded as 0, the measurement pool position expression value used to express the position of the measurement pool or the reference pool position expression value used to express the position of the reference pool is increased by 1; l is a set parameter used to select the output value of the number of pulses that is not affected by high-frequency noise and is greater than the threshold of the trigger signal.

[0012] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system according to the present application, the pulse signal after moving average is calculated by the following formula: MA (k+CNT)=avg γ {sig(n)},10≤γ≤20; where sig MA represents the pulse signal after moving average; k is the index of the external trigger; γ represents a constant that depends on fs; avg γ {sig(n)} represents the average value of the first γ% after sig(n) is sorted in descending order.

[0013] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system described in the present application, low-frequency noise is eliminated through the sub-band structure of the filter group during post-processing, including the steps of: estimating the power of each sub-band structure of the filter group; estimating the power of the full frequency band based on the estimated value of the power of each sub-band structure of the filter group; and removing noise based on the estimated value of the power of the full frequency band.

[0014] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system according to the present application, the power of each sub-band structure of the filter bank is estimated by the following formula: i(n) = (1-α) x Di (n-1)+α x Di (n); where x Di (n) = x i (n2 i-1 ); i=1,2,3,…,n.

[0015] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system according to the present application, the power of each sub-band structure of the filter bank is estimated by the following formula: i (n) = (1-α)x i (n-1)+αx i (n); where x i (n) represents the output signal of the i-th frequency band at time n; i = 1, 2, 3, ..., n.

[0016] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system according to the present application, the power of the full frequency band is estimated by the following formula:

[0017] In one embodiment of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system according to the present application, the noise is removed by the following formula: i (n) = βγ i (n)x i (n).

[0018] Further objectives and advantages of the present application will be fully reflected through understanding of the following description and drawings.

[0019] These and other objects, features and advantages of the present application are fully reflected in the following detailed description, drawings and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other purposes, features and advantages of the present application will become more apparent by describing the embodiments of the present application in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or parts.

[0021] Figure 1 The figure shows a flowchart of a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0022] Figure 2The figure shows a flowchart diagram of signal processing for performing multiple pollutant analyses by a non-dispersive infrared analysis method in a method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0023] Figure 3 The figure illustrates a flowchart of a signal processing detection method for distinguishing signals between measurement items in an interference signal noise elimination method in an ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0024] Figure 4 The figure shows a schematic diagram of the structure of a filter bank with a uniform tree structure.

[0025] Figure 5 The figure shows a schematic diagram of the structure of a non-uniform tree-structured filter bank.

[0026] Figure 6A The figure shows a schematic diagram of the non-dispersive detector output and trigger signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0027] Figure 6B The figure shows a schematic diagram of a gas signal measured in a pulse signal during simulation verification of a method for eliminating interference signal noise in a non-dispersive infrared monitoring system for ambient air according to an embodiment of the present application.

[0028] Figure 7A The figure shows a schematic diagram of a state diagram of a method for eliminating interference signal noise in a non-dispersive infrared monitoring system for ambient air according to an embodiment of the present application, in which a baseline level includes noise in simulation verification.

[0029] Figure 7B The figure shows a schematic diagram of a state diagram of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application after the noise is removed by the sub-band structure noise elimination algorithm in the simulation verification.

[0030] Figure 8A The figure shows a schematic diagram of the comparison results of the noisy reference unit signal of NO and the denoised reference unit signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0031] Figure 8B The figure shows a schematic diagram of the comparison results between another NO noisy reference unit signal and a denoised reference unit signal in a simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0032] Figure 9AThe figure shows a schematic diagram of the comparison results of the noisy reference unit signal of CO and the denoised reference unit signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0033] Figure 9B The figure shows a schematic diagram of the comparison results of another CO noisy reference unit signal and the denoised reference unit signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0034] Figure 10A The figure shows a schematic diagram of the comparison results of the noisy reference unit signal of SO2 and the denoised reference unit signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0035] Figure 10B The figure shows a schematic diagram of the comparison results between a noisy reference unit signal of another SO2 and a denoised reference unit signal in a simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0036] Figure 11A The figure shows a schematic diagram of the comparison results of the noisy reference unit signal and the measurement unit signal of NO in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0037] Figure 11B The figure shows a schematic diagram of the comparison results of another NO noisy reference unit signal and the measurement unit signal in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0038] Figure 12A The figure shows a schematic diagram of the comparison results of the noisy reference unit signal and the measurement unit signal of CO in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0039] Figure 12B The figure shows a schematic diagram of the comparison result between the noisy reference unit signal and the measurement unit signal of another CO in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0040] Figure 13A The figure shows a schematic diagram of the comparison results of the noisy reference unit signal and the measurement unit signal of SO2 in the simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application.

[0041] Figure 13B The figure shows a schematic diagram of the comparison result between the noisy reference unit signal and the measurement unit signal of another SO2 in simulation verification of the interference signal noise elimination method in the ambient air non-dispersive infrared monitoring system according to an embodiment of the present application. DETAILED DESCRIPTION

[0042] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0043] It is understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "a" should not be understood as limiting the number. "Multiple" means greater than or equal to two.

[0044] Although ordinal numbers such as "first," "second," and the like will be used to describe various components, these are not intended to limit those components. The terms are used solely to distinguish one component from another. For example, a first component could be referred to as a second component, and similarly, a second component could be referred to as a first component without departing from the teachings of the present disclosure. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0045] The terms used herein are for the purpose of describing various embodiments only and are not intended to be limiting. As used herein, the singular is intended to include the plural, unless the context clearly indicates otherwise. It will also be understood that the terms "including" and / or "having" when used in this specification specify the presence of a stated feature, number, operation, component, element, or combination thereof, and do not preclude the presence or addition of one or more other features, numbers, operations, components, elements, or combinations thereof.

[0046] like Figures 1 to 13BAs shown, the interference signal noise elimination method in the non-dispersive infrared monitoring system of ambient air according to the embodiment of the present application is explained. Taking into account the fact that the gas analysis equipment based on the non-dispersive infrared (ND infrared) analysis method currently has certain limitations in improving the accuracy and precision of noise elimination in terms of hardware due to the constraints of factors such as the overall manufacturing level and the working principle of the device, the present application improves the accuracy and precision of noise elimination at the algorithm level, and provides algorithm support for the gas analysis equipment based on the non-dispersive infrared analysis method. Furthermore, when designing the algorithm, the present application takes into account the hardware feasibility of the gas analysis equipment based on the non-dispersive infrared (ND infrared) analysis method currently, so that the interference signal noise elimination method in the non-dispersive infrared monitoring system of ambient air of the present application has relatively loose requirements on hardware, which is convenient for the implementation of hardware adapted thereto.

[0047] The following describes in detail the principle and implementation of the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system of the present application.

[0048] The interference signal noise elimination method for an ambient air non-dispersive infrared monitoring system specifically eliminates different types of noise during pre-processing and post-processing, thereby improving the progress and accuracy of non-dispersive infrared measurements. Specifically, the interference signal noise elimination method eliminates rapidly changing high-frequency noise during pre-processing of time domain characteristics, thereby achieving strong resistance to rapidly changing high-frequency noise. Furthermore, the sub-band noise elimination technique of the filter bank is used to eliminate slowly changing low-frequency noise during post-processing, which considers power ratios through the sub-band structure of the filter bank, thereby improving resistance to slowly changing low-frequency noise.

[0049] Accordingly, the method for eliminating interference signal noise in the ambient air non-dispersive infrared monitoring system includes step S110, eliminating high-frequency noise in a pre-processing process; and step S120, eliminating low-frequency noise in a post-processing process.

[0050] Specifically, the gas analysis equipment based on the non-dispersive infrared (ND) analysis method includes a non-dispersive infrared detector, wherein the non-dispersive infrared detector includes a multi-path optical absorption chamber (MOA), an optical interference filter and an infrared sensor. The multi-path optical absorption chamber includes a multi gas filter correlation wheel (multiGFC wheel) and a reflector. The multi-gas correlation filter wheel is connected to a reference cell filled with a high concentration of standard gas and a measuring cell filled with nitrogen by alternating switching, thereby offsetting the influence of interfering components, thereby eliminating the influence of other interfering substances during measurement.

[0051] Broad-wavelength infrared light emitted from a light source passes through a rotating multi-gas correlation filter wheel, alternating between a reference cell and a measurement cell. This generates a light signal that forms reference pulses and measurement pulses within the multi-path light absorption cavity. During the measurement pulse, the nitrogen-filled measurement cell absorbs virtually no infrared light, while the multi-path light absorption cavity absorbs infrared light at a level corresponding to the concentration of the measurement gas. Conversely, during the reference pulse, the high-concentration standard gas within the measurement cell absorbs most of the infrared light of a specific wavelength, making the absorption by the measurement gas, which has a lower concentration than the standard gas, negligible.

[0052] Light emitted from the multipath light absorption cavity passes through an optical interference filter, which only allows light of a preset wavelength to enter the infrared sensor. Light entering the infrared sensor is converted into a voltage signal and amplified. The degree of light absorption depends on the concentration of the substance; the relationship between concentration and absorbance follows the Beer-Lambert Law, as follows: A(λ) = ε(λ)LC, where A(λ) represents the absorbance at wavelength λ, ε(λ) represents the absorption coefficient at wavelength λ, L represents the light transmission distance (which is equal to the optical path length within the multipath light absorption cavity), and C represents the gas concentration. According to the Beer-Lambert Law, absorbance is proportional to the length of the multipath light absorption cavity and the gas concentration. Therefore, when the absorption coefficient ε is sufficiently large and the target gas is measured using monochromatic light, the concentration of the target gas can be calculated by measuring the intensity of the transmitted light based on A(λ) = ε(λ)LC without interference from other gases. The concentration of the gas being measured is determined by calculating the differential absorption difference between the reference cell output signal and the measurement cell output signal corresponding to each channel's infrared light. The non-dispersive infrared detector alternately detects light passing through a pair of reference and measurement cells. Therefore, the detector's output signal exhibits a modulated pulse pattern of reference and measurement pulses.

[0053] In step S110, signal processing for various pollutant analyses is performed using the non-dispersive infrared (ND) analysis method. The optical signal passes through the NDIR detector to generate an analog electrical signal. This analog electrical signal has a low level. This analog electrical signal is amplified by a preamplifier to generate an amplified analog signal. This amplified analog signal is then converted by an analog-to-digital converter (ADC) to generate an initial digital signal. In other words, the optical signal is sequentially converted through the NDIR detector, preamplifier, and ADC into an initial digital signal.

[0054] Because the initial digital signals for multiple measurement items are continuous, it is necessary to accurately distinguish each initial digital signal to determine which measurement item each initial digital signal corresponds to, thereby facilitating subsequent analysis. To this end, an external optical trigger sensor located outside the rotating multi-gas correlation filter wheel is used to classify the multiple initial digital signals.

[0055] Specifically, each of the initial digital signals includes an output signal of a reference cell and an output signal of a measuring cell, and is in a form similar to a pulse signal sequence. Corresponding to each measurement item, the output signal of the reference cell and the output signal of the measuring cell are a pair of pulse signals.

[0056] In order to distinguish the initial digital signal of the corresponding measurement item, the trigger signal and threshold value provided by the optical coupler attached to the multi-gas correlation filter wheel are used to accurately find the synchronization of one cycle of the multi-gas correlation filter wheel. Based on the threshold value of the trigger signal, the number of pulses in the continuous pulse signal is calculated to find the output signal of the reference cell and the output signal of the measurement cell of the corresponding measurement item. In the process of calculating the number of pulses in the continuous pulse signal based on the threshold value of the trigger signal to find the output signal of the reference cell and the output signal of the measurement cell of the corresponding measurement item, the baseline level of the pulse signal will be affected by the high-frequency noise caused by various analog components and mechanical rotation, as well as the drifting low-frequency noise caused by the temperature characteristics and thermal noise of the non-dispersive infrared detector. Due to the high-frequency noise added to the pulse signal amplitude, errors may occur when using a fixed threshold value THCNT to calculate the number of pulses, so the following method is used to count the number of pulses in the pulse signal:

[0057]

[0058] Wherein, l is a set parameter used to select the output value of the number of pulses that is not affected by high-frequency noise and is greater than a preset threshold; if the number of pulses in the pulse signal of the (nl)th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline, the number of target pulses is recorded as 0; if the number of pulses in the pulse signal of the (nl)th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline and the number of target pulses is recorded as 0, the measurement pool position expression value used to express the position of the measurement pool or the reference pool position expression value used to express the position of the reference pool is increased by 1.

[0059] When the concentration of the measured item (eg, SO2) is high, the absorption of the infrared light source in the measuring cell increases, resulting in a decrease in the amplitude of the output signal of the measuring cell. As a result, the signal-to-noise ratio decreases, making it difficult to count the number of pulses.

[0060] Furthermore, when extracting valid data from pulse signals corresponding to classified measurement items, the impact of high-frequency noise on the pulse signal peaks must be considered. To extract valid data from pulse signal peaks contaminated by high-frequency noise, a moving average (MA) is applied to the pulse signals of the classified initial digital signal, using a preset pulse signal baseline as a reference to eliminate high-frequency noise.

[0061] sig MA (k+CNT)=avg γ {sig(n)},10≤γ≤20;

[0062] Among them, sig MA represents the pulse signal after moving average; k is the index of the external trigger; γ represents a constant that depends on fs; avg γ {sig(n)} represents the average value of the first γ% after sig(n) is sorted in descending order.

[0063] After the signals for each measurement item are classified, the corresponding reference voltage is predicted. The measured value is determined by taking the average of the first γ% of the data above the predicted reference voltage. This pre-processing noise removal technique can quickly remove rapidly changing high-frequency noise. The value of γ can be selected as needed, for example, 20.

[0064] Accordingly, step S110 includes the following steps: S111, converting the optical signal into an initial digital signal, wherein each of the initial digital signals includes an output signal of a reference cell and an output signal of a measuring cell, and the output signal of the reference cell and the output signal of the measuring cell are a pair of pulse signals; S112, classifying the multiple initial digital signals; and S113, eliminating high-frequency noise by performing a moving average on the pulse signals of the classified initial digital signals based on a preset pulse signal baseline.

[0065] Step S111, converting the optical signal into an initial digital signal, includes the following steps: S1111, converting the optical signal into an analog electrical signal through a non-dispersive infrared detector; S1112, amplifying the analog electrical signal through a preamplifier to form an amplified analog signal; S1113, converting the amplified analog signal through an analog-to-digital converter to form the initial digital signal.

[0066] Step S112: Classify the multiple initial digital signals. Specifically, the multiple initial digital signals are classified using a light-triggered sensor, wherein the light-triggered sensor is located outside the rotating multi-gas correlation filter wheel. Step S112 includes step S1121: generating a trigger signal based on the initial digital signal; and step S1122: counting the number of pulses in the continuous pulse signal based on a threshold of the trigger signal. If the number of pulses in the pulse signal from the (n-1)th to the n-th pulse signal in the initial digital signal is less than a preset pulse signal baseline, the number of target pulses is recorded as 0. If the number of pulses in the pulse signal from the (n-1)th to the n-th pulse signal in the initial digital signal is less than the preset pulse signal baseline and the number of target pulses is recorded as 0, a measurement cell position expression value used to express the position of the measurement cell or a reference cell position expression value used to express the position of the reference cell is incremented by 1. l is a set parameter used to select an output value of the number of pulses that is not affected by high-frequency noise and is greater than the threshold of the trigger signal.

[0067] In step S113, sig MA (k+CNT)=avg γ {sig(n)},10≤γ≤20; where sig MA represents the pulse signal after moving average; k is the index of the external trigger; γ represents a constant that depends on fs; avg γ {sig(n)} represents the average value of the first γ% after sig(n) is sorted in descending order.

[0068] To improve the accuracy of the measured data, the synchronization of the external synchronization signal (trigger signal) with the target signal detection algorithm is continuously performed, and the number of data samples for each measurement item and the estimation of the reference voltage are repeatedly confirmed and executed. As a noise removal technique in this pre-processing process, the rapidly changing high-frequency noise generated by the detection part, including the sensor and pre-processing amplifier, and the mechanical part including the multi-gas correlation filter wheel can be effectively removed. However, the slowly changing low-frequency noise generated by the sensor drive circuit and the stepper motor that drives the multi-gas correlation filter wheel is more difficult to remove. To this end, a sub-band noise removal technique based on a filter bank is adopted as a post-processing method. By passing the noisy signal through each sub-band decomposition filter bank, the noise is removed from the power of the obtained sub-band signal. Generally speaking, unlike noise that is uniformly distributed across the entire frequency range, the signal is distributed within a limited frequency range. Therefore, after passing through the decomposition filter bank, the signal frequency appears in the corresponding frequency interval. When the signal-to-noise ratio (SNR) is high, the signal power is large relative to the power of the added noise, so the sub-band containing the signal has a higher power than other sub-bands. Noise removal is performed by exploiting the frequency characteristics and power relationship between the signal and the noise. Figure 4 The uniform tree-structured filter bank shown here maintains uniform resolution across all frequency bands, ensuring consistent noise removal performance even when the actual signal resides within any frequency band. Furthermore, each subband uses the same decomposition / synthesis filters and the same number of downsampling and upsampling operations. Therefore, hardware implementation eliminates the need to consider delay differences caused by downsampling rates and subband filtering, simplifying system implementation.

[0069] Generally speaking, the actual target signal is distributed in a frequency band lower than the low-frequency noise. Figure 4 Each frequency band in has uniform resolution, Figure 5 The low-frequency band contained in the actual signal is decomposed into a uniform and uneven frequency band structure, thereby improving the resolution of the low-frequency band. Since the frequency components of the actual signal are mostly distributed in the low-frequency band, it will not cause a decrease in noise removal performance, and the hardware cost can be significantly reduced. If the frequency components of the signal are in the high-frequency band, a structure that repeatedly divides the high-frequency band instead of the low-frequency band can be used. However, when the target signal is distributed in both low-frequency and high-frequency areas, the degradation of noise removal performance may be very serious, and in hardware implementation, it is necessary to accurately calculate the downsampling rate of each sub-band and the delay difference caused by sub-band filtering, so it faces certain difficulties. Therefore, it is necessary to appropriately select the sub-band according to the characteristics of the detection signal. Figure 4 or Figure 5 structure.

[0070] The subband power estimation method used to perform noise removal on each subband is as follows:

[0071] P i =σ i 2 ; Where Pi represents the estimated value of the subband power of the filter bank; σ i 2 represents the variance of the signal output from the i-th frequency band.

[0072] Only when the statistical characteristics of the output signals of each sub-band are fully understood can the P i =σ i 2 To achieve sub-band power estimation, it is impossible to directly obtain these characteristics. Instead, we can use the leakage parameter α between the sub-band output signal power at control time n and the past power to perform the following recursive estimation:

[0073] P i (n) = (1-α)x i (n-1)+αx i (n).

[0074] Figure 5 The unevenly decomposed sub-band structures shown in FIG, due to the inconsistent downsampling rates of the signal sequences, it is necessary to consider the number of downsampling times for each frequency band and recursively calculate the sub-band output signals in the following way:

[0075] P i (n) = (1-α) x D i (n-1)+α x Di (n); where x Di (n) = x i (n2 i-1 ), where i = 1, 2, 3, ..., xi(n) represents the output signal of the i-th frequency band at time n.

[0076] When using Figure 4 When the uniform tree structure is used, since there is no need to consider the number of downsampling, the subband output signal is recursively calculated in the following way:

[0077] P i (n) = (1-α)x i (n-1)+αx i (n), i = 1, 2, 3, ...; where x i (n) represents the output signal of the ith frequency band at time n. Using the sub-band power, the power of the full frequency band can be estimated as follows: The power ratio is calculated based on the proportion of the signal and noise power in the corresponding sub-band power as follows:

[0078]

[0079] The power ratio-based noise cancellation method of this application is similar to the soft threshold value selection method in noise cancellation using thresholds. Generally speaking, soft threshold methods have better performance than hard threshold methods, but are more difficult to implement. However, although the power ratio-based noise cancellation method of this application is similar to the soft threshold method, it has the advantage of being simpler to implement.

[0080] By v i (n) = βγ i (n)x i (n) Each subband signal after noise removal is restored by up sampling and synthesis filter. In order for the non-uniform tree structure and the uniform tree structure to obtain the same subband decomposition capability, subband decomposition at the same level is required. In the case of 4 layers, the number of subbands in the uniform tree structure is 16, while the number of subbands in the non-uniform tree structure is 8. Therefore, the uniform tree structure will lead to a significant increase in system cost compared to the non-uniform tree structure. However, when using a non-uniform structure, it is necessary to solve the serious degradation of the noise eliminator performance and the delay mismatch problem caused by the difference in the number of downsampling / upsampling and decomposition / synthesis filters during hardware implementation. Therefore, a noise eliminator based on a uniform tree structure is adopted.

[0081] Accordingly, in step S120, low-frequency noise is eliminated by the subband structure of the filter bank. Step S120 includes the following steps: S121, estimating the power of each subband structure of the filter bank; S122, estimating the power of the entire frequency band based on the estimated power values ​​of each subband structure of the filter bank; and S123, removing noise based on the estimated power value of the entire frequency band.

[0082] In step S121, by P i (n) = (1-α) x Di (n-1)+α x Di (n) or P i (n) = (1-α)x i (n-1)+αx i (n), i = 1, 2, 3, ... Estimate the power of each subband structure of the filter bank: P i (n) = (1-α)x i (n-1)+αx i (n); where x Di (n) = x i (n2 i-1 ), i=1,2,3,…,n; x i(n) represents the output signal of the i-th frequency band at time n.

[0083] In step S122, by Estimate the power of the full frequency band.

[0084] In step S123, by v i (n) = βγ i (n)x i (n) Remove noise.

[0085] In this application, a computer simulation was performed to verify the interference signal noise elimination method in the non-dispersive infrared monitoring system for ambient air, in order to evaluate the application effect of the interference signal noise elimination method in the non-dispersive infrared monitoring system for ambient air. During the computer simulation verification of the interference signal noise elimination method in the non-dispersive infrared monitoring system for ambient air, the standard mixed gas used was as shown in Table 1, and the flow rate was maintained at 1000 [cc / min] by using an orifice.

[0086] Table 1. Standard gas mixtures used in the experiments

[0087]

[0088] Figure 6A and Figure 6B The sequence of measurement items for the ND infrared detector's output signal, external trigger signal, and pulse signal, measured over a 20-second period, is shown. In the multi-signal detection algorithm, the rotational speed of the multiple GFC wheels is K = 0.655 cycles / second, the sampling frequency is fs = 1 kHz, and an upper-layer data processing parameter γ = 20 is used to remove high-frequency noise. The analysis filter bank used for post-processing in the simulation is a length-2 analysis filter bank obtained using a Hadamard transform, employing uniform 4- and 16-subband structures.

[0089] Figure 6A The non-dispersive detector output and trigger signal are shown; Figure 6B The gas signals measured in the pulse signal are shown (a = sample.HCl, b = reference.CO, c = sample.CO, d = reference.NO, e = sample.NO, f = reference.SO2, g = sample.SO2, h = reference.H2O, i = sample.H2O, j = none, k = none, l = reference.HCl). Figure 7A and Figure 7B The image shows the state of the baseline level containing noise and the state after the noise is removed using the noise cancellation algorithm with 4 and 16 subband structures. Changes in the baseline level will affect the measured values ​​of various pollutants in the infrared measurement signal.

[0090] Figure 8A and Figure 8B The comparison results of the noisy baseline unit signal and the denoised baseline unit signal of NO are shown. Figure 9A and Figure 9B The comparison results of the noisy reference unit signal and the denoised reference unit signal of CO are shown. Figure 10A and Figure 10B Comparison results of a noisy reference unit signal and a denoised reference unit signal for SO2 are shown. The subband structure used for denoising is a uniform tree filter bank with 4 and 16 subband structures. It can be seen that the changes in the measured values ​​of each pollutant term depend on the changes in the reference level in Figure 6. Furthermore, noise removal performance improves with an increase in the number of subband decompositions. From these results, it is easy to infer that the increase in resolution with an increase in the number of subbands leads to more accurate noise signal removal.

[0091] Figure 11A and Figure 11B The comparison results of the noisy reference cell signal and the measured cell signal of NO are shown. Figure 12A and Figure 12B The comparison results of the noisy reference cell signal and the measurement cell signal of CO are shown. Figure 13A and Figure 13B The comparison results of the noisy reference cell signal and the measurement cell signal for SO2 are shown. As can be seen from the figure, the reference cell signal and the measurement cell signal for each gas item have similar changing trends. The GFC method cancels out the effects of interfering components by alternating between a reference cell filled with a high-concentration standard gas and a measurement cell filled with nitrogen, thereby removing the influence of other interfering substances during measurement. Even if the reference level changes, the infrared signal measurement values ​​of each gas item will still depend on the change in the reference level, but this mutual cancellation effect can reduce the impact of drift.

[0092] The analysis filters of the subband filter bank used in the experiment consist of coefficients of length 2, obtained through a Hadamard transform. While the cutoff characteristics of each analysis filter are somewhat reduced, achieving higher subband resolution allows for more decomposition without the delay caused by the analysis filter length and the multiplication operations required for convolution. After downsampling at the maximum downsampling rate, the processing of each filter does not overlap, allowing for noise cancellation. Restoration is then performed using a shorter synthesis filter (the length of the synthesis filter should be the same as the analysis filter, so the length of the synthesis filter is 2). The reason why noise cancellation appears as block processing within a certain time interval is that it is performed after downsampling, following subband decomposition. However, the lower sampling rate caused by downsampling reduces the computational complexity and memory requirements of the noise canceller, thus reducing hardware costs and enabling the implementation of low-spec systems.

[0093] In summary, the method for eliminating interference signal noise in the non-dispersive infrared monitoring system for ambient air has been described. This method improves the accuracy and precision of noise elimination at the algorithmic level, provides algorithmic support for gas analysis equipment based on non-dispersive infrared analysis, and has relatively loose hardware requirements, facilitating the implementation of compatible hardware. Specifically, the method for eliminating interference signal noise in the non-dispersive infrared monitoring system for ambient air specifically eliminates different types of noise during pre-processing and post-processing, effectively removing low-frequency and high-frequency noise of varying natures.

[0094] The above description of the present application and its embodiments is non-limiting. The drawings show only one embodiment of the present application, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the inventive purpose of this application, designs a structure and embodiment similar to this technical solution without creatively designing, they shall fall within the scope of protection of this application.

Claims

1. A method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system, characterized in that: Including steps: Eliminating high-frequency noise in a preprocessing process includes the following steps: converting an optical signal into an initial digital signal, wherein each of the initial digital signals includes an output signal of a reference cell and an output signal of a measuring cell, and the output signal of the reference cell and the output signal of the measuring cell form a pair of pulse signals; classifying a plurality of the initial digital signals; and eliminating high-frequency noise by performing a moving average on the pulse signals of the classified initial digital signals based on a preset pulse signal baseline; and In the post-processing process, low-frequency noise is eliminated through the sub-band structure of the filter bank; The classification of the plurality of initial digital signals comprises the steps of: generating a trigger signal based on the initial digital signal; The number of pulses in the continuous pulse signal is counted based on the threshold of the trigger signal; wherein, if the number of pulses in the pulse signal of the n1th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline, the number of target pulses is recorded as 0; if the number of pulses in the pulse signal of the n1th to nth pulse signal in the initial digital signal is less than the preset pulse signal baseline and the number of target pulses is recorded as 0, the measurement pool position expression value used to express the position of the measurement pool or the reference pool position expression value used to express the position of the reference pool is increased by 1; l is a set parameter used to select an output value of the number of pulses that is not affected by high-frequency noise and is greater than the threshold of the trigger signal; In the post-processing process, low-frequency noise is eliminated by using the sub-band structure of the filter bank, including the steps of: Estimating the power of each subband structure of the filter bank; estimating the power of the entire frequency band based on the estimated values ​​of the power of each subband structure of the filter bank; Noise is removed based on the estimated value of power across the entire frequency band.

2. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 1, characterized in that: Converting the optical signal into an initial digital signal includes the following steps: The optical signal is converted into an analog electrical signal through a non-dispersive infrared detector; amplifying the analog electrical signal through a preamplifier to form an amplified analog signal; The amplified analog signal is converted through an analog-to-digital converter to form the initial digital signal.

3. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 2, characterized in that: The pulse signal after moving average is calculated by the following formula: ;in, sig MA Represents the pulse signal after moving average; k The index of the external trigger; γ Depends on fs constant; avg γ { sig ( n )} means sig ( n ) After sorting in descending order, γ % average value.

4. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 1, characterized in that: The power of each subband structure of the filter bank is estimated by the following formula: ; where x Di (n)=x i (n2 i−1 ); i=1,2,3,…,n, α represents the leakage degree parameter.

5. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 1, characterized in that: The power of each subband structure of the filter bank is estimated by the following formula: ; where x i (n) represents the output signal of the i-th frequency band at time n; i = 1, 2, 3, ..., n, α represents the leakage degree parameter.

6. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 1, characterized in that: The power of the full frequency band is estimated by the following formula: .

7. The method for eliminating interference signal noise in an ambient air non-dispersive infrared monitoring system according to claim 6, characterized in that: The noise is removed by the following formula: .

Images