In-situ harmful gas monitoring method for sewage treatment plant structure

Abstract

The invention discloses an in-situ harmful gas monitoring method for sewage treatment plant structures, and particularly relates to the field of in-situ harmful gas monitoring, and the in-situ harmful gas monitoring method comprises the following steps: collecting refractive index disturbance data of a local gas-liquid-gas interface caused by bubble release in a sludge tank or an anaerobic tank, and forming a refractive index distribution field changing along with time; and modeling an infrared light propagation path based on the refractive index distribution field, executing infrared absorption inversion operation, solving a first concentration result of the harmful gas, and establishing a mapping relationship between the first concentration result and the refractive index distribution field. Through coupling calculation of a refractive index distribution field caused by bubble disturbance and an infrared absorption inversion process, a reciprocating iteration and time sequence prediction chain is constructed so as to correct dynamic optical distortion and stably solve the real release intensity of harmful gas.

Description

Technical Field

[0001] The present invention relates to the technical field of in-situ harmful gas monitoring, and more particularly to an in-situ harmful gas monitoring method for sewage treatment plant structures. Background Art

[0002] In the sludge tanks or anaerobic tanks of sewage treatment plants, microbubbles are periodically released from the bottom, often carrying high-risk gases such as hydrogen sulfide and methane. As the bubbles rise, a complex gas-liquid-gas interface forms near the liquid surface. This interface not only causes instantaneous scattering and strong absorption of infrared light signals, but also causes rapid and irregular changes in the local refractive index due to the dramatic fluctuations in the bubble volume fraction, leading to dynamic bending and distortion of the optical path. This effect is instantaneous and highly non-uniform. Existing infrared detection methods that rely on average absorption spectra are difficult to capture stably, often resulting in signal saturation or spectral distortion, which in turn obscures the true gas release intensity and distribution characteristics. Therefore, the core problem of the existing technology is that the dynamic optical distortion caused by bubble eruption destroys the accuracy of infrared monitoring, making it impossible for in-situ detection to effectively characterize the actual release process of harmful gases inside the structure at the critical moment. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an in-situ harmful gas monitoring method for sewage treatment plant structures. By coupling the refractive index distribution field caused by bubble disturbance with the infrared absorption inversion process, a reciprocating iteration and time series prediction chain is constructed to correct dynamic optical distortion and stably solve the actual release intensity of harmful gases, so as to solve the problems raised in the above-mentioned background technology.

[0004] To achieve the above-mentioned object, the present invention provides the following technical solution: a method for in-situ harmful gas monitoring in sewage treatment plant structures, comprising: S1: In a sludge tank or anaerobic tank, the refractive index disturbance data of the local gas-liquid-gas interface caused by the release of bubbles is collected to form a refractive index distribution field that changes with time; S2: Modeling the infrared light propagation path based on the refractive index distribution field, performing infrared absorption inversion calculation, solving a first concentration result of the harmful gas, and establishing a mapping relationship between the first concentration result and the refractive index distribution field; S3: Calculate the residual between the theoretical spectrum corresponding to the first concentration result and the actual infrared spectrum, back-project the residual result onto the refractive index distribution field to form an updated refractive index value; and modify the infrared absorption inversion operation based on the updated refractive index value to generate an updated concentration value. S4: constructing a reciprocating iterative chain between the refractive index update value and the concentration update value, forming a converged result through multiple rounds of iteration, and simultaneously outputting the refractive index update value and the concentration update value in each round of iteration for input into the next round of iteration; S5: The concentration update value and the refractive index update value output by the reciprocating iterative chain are simultaneously input into the timing modeling algorithm to generate a preset release time prediction result of the harmful gas, and the prediction result is fed back to the refractive index distribution field to update the input conditions of the next round of iteration.

[0005] In a preferred embodiment, S1 includes: S1-1: Arranging an incident light beam and a reference light beam along the bubble release area in a sludge tank or an anaerobic tank, collecting a fringe image of optical interference, and extracting a phase difference value sequence from the fringe image; S1-2: Expanding the phase difference value sequence into a continuous phase curve according to the sampling time sequence, calculating the phase difference components between adjacent sampling points, and converting the phase difference components into a refractive index increment sequence according to a preset refractive index conversion coefficient; S1-3: Corresponding the refractive index increment sequence to the spatial coordinates of the detection area point by point, calculating a refractive index perturbation matrix, and forming a refractive index perturbation distribution covering the entire gas-liquid-gas interface through matrix interpolation; S1-4: performing an accumulation operation on the refractive index disturbance matrix according to the time index to generate a refractive index distribution field that varies with time.

[0006] In a preferred embodiment, S2 includes: S2-1: Collect the infrared transmission spectrum intensity through the gas layer in the detection area, and record the reference spectrum intensity under gas-free conditions. Divide the transmission spectrum intensity by the reference spectrum intensity and calculate the transmittance sequence wavelength by wavelength. S2-2: Take the negative logarithm of the transmittance sequence point by point and multiply it by the effective optical path length to convert the transmittance into the absorption value corresponding to each wavelength, thereby forming an absorption spectrum curve, which serves as the input data for concentration inversion; S2-3: Compare the absorption spectrum curve with a pre-established harmful gas standard absorption database wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitting curve and the actual absorption spectrum curve reaches a preset lower limit, and output the first concentration result of the target harmful gas; S2-4: The first concentration result is mapped point by point to the spatial position of the infrared light path according to the sampling time, and stored in the grid unit of the refractive index distribution field, and the refractive index data and concentration data are simultaneously maintained in the distribution field, thereby establishing a one-to-one mapping relationship between the two.

[0007] In a preferred embodiment, S3 includes: S3-1: Calculate a theoretical absorption spectrum curve based on the first concentration result, and subtract the theoretical absorption spectrum curve from the actual infrared spectrum curve wavelength by wavelength to obtain a residual sequence; S3-2: Back-projecting the residual sequence according to the infrared light path and the spatial coordinates of the detection area to obtain a refractive index update value matrix for representing the refractive index correction amount of the gas-liquid-gas interface at each spatial position; The refractive index update value matrix includes the refractive index correction values ​​obtained by back-calculating the residual signal on each grid cell in the detection area. Each row and each column of the refractive index update value matrix corresponds to a position point on the spatial coordinate axis, which is used to record the refractive index correction distribution of the gas-liquid-gas interface in the entire area.

[0008] In a preferred embodiment, S3 further includes: S3-3: Inputting the refractive index update value matrix into the infrared absorption inversion operation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of the infrared light propagating in the detection area. The corrected optical path length is then substituted into the transmittance to absorption conversion formula to recalculate the updated absorption spectrum curve. The updated absorption spectrum curve is then fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value. The new concentration value is then arranged in the order of sampling time to form a concentration update value sequence; S3-4: Establish a one-to-one correspondence between the refractive index update value matrix and the concentration update value sequence, and store the correspondence in the refractive index distribution field to provide input conditions for subsequent reciprocating iterative steps.

[0009] In a preferred embodiment, S4 includes: S4-1: The refractive index update value matrix and the concentration update value sequence are used as initial inputs, and the grid coordinate information in the refractive index distribution field is loaded to construct the input conditions for the first round of iteration; S4-2: In the first round of iteration, the optical path length is corrected using the refractive index update value matrix, the absorption spectrum curve is corrected using the concentration update value sequence, and infrared absorption inversion fitting is performed to output a new refractive index update value matrix and concentration update value sequence.

[0010] In a preferred embodiment, S4 further includes: S4-3: re-inputting the newly output refractive index update value matrix and concentration update value sequence into the iterative process to form a reciprocating cycle, and calculating the residual amplitude after each cycle until the residual amplitude is lower than the preset lower limit; S4-4: When the residual amplitude reaches a preset lower limit, the refractive index update value matrix and the concentration update value sequence obtained in this round of iteration are stored as convergence results in the refractive index distribution field as input conditions for subsequent steps.

[0011] In a preferred embodiment, in S5, the following steps are included: S5-1: Extracting concentration update values ​​and refractive index update values ​​from the output result of the reciprocating iterative chain consisting of the refractive index update value matrix and the concentration update value sequence according to the same sampling time sequence, and assigning the concentration update values ​​and the refractive index update values ​​to each other point by point in the grid coordinates of the refractive index distribution field, thereby forming a data set including the sampling time, spatial position, concentration update value, and refractive index update value; S5-2: Dividing the data set into multiple input sequences according to consecutive sampling moments, and performing normalization processing on the concentration update values ​​and refractive index update values ​​in each input sequence to obtain concentration update value sequences and refractive index update value sequences arranged by time index; S5-3: Inputting the concentration update value sequence and the refractive index update value sequence simultaneously into a time series modeling algorithm based on a recurrent neural network, performing forward calculations at each sampling moment in the time series modeling algorithm to obtain a future concentration prediction value sequence covering a preset time length; S5-4: writing the future concentration prediction value sequence into the grid cells of the refractive index distribution field point by point according to the sampling time index and spatial position, while keeping the refractive index update value unchanged while writing the future concentration prediction value, thereby forming a data set containing the concentration prediction values ​​at the future sampling times in the refractive index distribution field; S5-5: The future concentration prediction value in the data set is used as the concentration input condition for the next round of reciprocating iterative chain, and the refractive index update value of the refractive index distribution field at the corresponding sampling moment is used as the refractive index input condition, thereby completing the feedback update of the input conditions for the next round of reciprocating iterative chain.

[0012] The technical effects and advantages of the present invention are as follows: 1. By converting the phase change under bubble disturbance into a refractive index distribution field and coupling it with the infrared absorption inversion process, the dynamic optical distortion caused by bubble eruption is successfully overcome, ensuring the in-situ detection accuracy of harmful gases during the instantaneous release process; 2. By establishing a mapping relationship between the refractive index distribution field and infrared absorption inversion, the concentration results can be made to correspond one-to-one with the spatial refractive index changes, thereby maintaining the spatial resolution of the detection results in the non-uniform disturbance field; 3. By introducing the back-projection calculation of the residual sequence, the difference between the theoretical spectrum and the actual spectrum is converted into a refractive index update value, realizing real-time correction of the optical path modeling and avoiding concentration deviation caused by model error; 4. Through the mutual correction of the refractive index update value and the concentration update value in the reciprocating iterative chain, a convergence mechanism under the residual amplitude control is formed, ensuring that the monitoring results obtain stable and reliable numerical solutions after multiple rounds of corrections; 5. By inputting the output data of the iterative chain into the time series modeling algorithm and using the recurrent neural network to perform forward calculations at each sampling moment, the concentration prediction results for the preset time period are obtained, achieving early characterization of future gas release trends; 6. By simultaneously storing the updated refractive index value and the predicted concentration value in the refractive index distribution field, the concentration and refractive index remain consistent in a unified coordinate system, providing complete data support for subsequent continuous iterations and multi-period predictions. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The figure is a flow chart of the method steps of the present invention. DETAILED DESCRIPTION

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0015] Refer to the instruction manual Figure 1 According to one embodiment of the present invention, a method for in-situ harmful gas monitoring of a sewage treatment plant structure includes: S1 obtains the refractive index disturbance distribution: in the sludge tank or anaerobic tank, collects the refractive index disturbance data of the local gas-liquid-gas interface caused by the release of bubbles, and forms a refractive index distribution field that changes with time; S2 constructs a preliminary infrared spectrum solution: models the infrared light propagation path based on the refractive index distribution field, performs infrared absorption inversion calculations, solves for a first concentration result of the harmful gas, and establishes a mapping relationship between the first concentration result and the refractive index distribution field; S3 generates a residual correction chain: calculates the residual between the theoretical spectrum corresponding to the first concentration result and the actual infrared spectrum, back-projects the residual result onto the refractive index distribution field to form a refractive index update value; corrects the infrared absorption inversion operation based on the refractive index update value to generate a concentration update value; S4 reciprocating convergence iteration: constructing a reciprocating iterative chain between the refractive index update value and the concentration update value, forming a convergence result through multiple rounds of iteration, and simultaneously outputting the refractive index update value and the concentration update value in each round of iteration for input into the next round of iteration; S5 constructs a time series prediction in parallel: the concentration update value and the refractive index update value output by the reciprocating iterative chain are simultaneously input into the time series modeling algorithm to generate a preset time release prediction result of the harmful gas, and the prediction result is fed back to the refractive index distribution field to update the input conditions of the next round of iteration.

[0016] Included in S1: S1-1: Arranging an incident light beam and a reference light beam along the bubble release area in a sludge tank or an anaerobic tank, collecting a fringe image of optical interference, and extracting a phase difference value sequence from the fringe image; S1-2: Expanding the phase difference value sequence into a continuous phase curve in the order of sampling moments, calculating the phase difference components between adjacent sampling points, and converting the phase difference components into a refractive index increment sequence based on a preset refractive index conversion coefficient. When expanding the phase difference value sequence into a continuous phase curve, it is necessary to perform a phase unwrapping operation on the phase difference values ​​arranged according to the sampling moments. When encountering a jump point spanning positive or negative π, the discontinuity is eliminated by adding or subtracting 2π, thereby obtaining a smooth and continuous phase curve that fully reflects the refractive index perturbation process that changes with time. In addition, the phase difference component between adjacent sampling points is obtained by performing a first-order difference operation on the continuous phase curve, that is, subtracting the phase value of the previous sampling point from the phase value of the next sampling point, and outputting it with the sampling time interval as the index, thereby forming a difference component sequence reflecting the instantaneous phase change rate; The preset refractive index conversion coefficient refers to a proportional coefficient determined in advance according to the optical path length and the wavelength of the detection light, and is used to directly convert the phase difference component into the corresponding refractive index change value; S1-3: The refractive index increment sequence is matched point by point with the spatial coordinates of the detection area to calculate a refractive index perturbation matrix, and a refractive index perturbation distribution covering the entire gas-liquid-gas interface is formed by matrix interpolation; wherein the refractive index increment sequence is pixel-coordinate aligned according to the calibrated imaging geometry and optical path parameters, the refractive index increment of each sampling point is mapped to a regular grid cell of the detection area, and weighted accumulation and normalization are performed using the effective optical path through the cell as a weight to obtain the refractive index perturbation value of each grid cell; neighborhood interpolation is performed on the grid cells not covered by the sampling, and the matrix is ​​the refractive index perturbation matrix according to the row and column indexes; Matrix interpolation refers to the numerical calculation of the refractive index perturbation values ​​of known grid points according to the spatial position relationship, and the values ​​of the missing positions are filled with the weighted results of the adjacent points, thereby forming a continuous coverage of the refractive index perturbation distribution on the entire gas-liquid-gas interface; S1-4: performing an accumulation operation on the refractive index perturbation matrix according to a time index to generate a refractive index distribution field that varies with time; Regarding S1, it should be noted that in practical applications, an optical interferometer (such as a Michelson interferometer or a Mach-Zehnder interferometer) can be usually used. An incident light beam passing through the gas-liquid interface is arranged in the bubble release area of ​​the sludge tank or anaerobic tank, and a reference light beam that does not pass through the bubble disturbance area is set at the same time. The two beams of light are finally combined on the interferometer detector for comparing phase changes. The incident beam is the light that passes through the detection area and is affected by the bubble disturbance; the reference beam is the light that does not pass through the detection area and maintains a stable phase. The two are superimposed to form interference fringes, and the phase difference can reflect the change in refractive index. In S1-1, the calculation process of obtaining the phase difference value sequence from the fringe image of optical interference includes: performing camera dark field correction and flat field correction on the collected interference fringe image, completing lens distortion correction and inter-frame geometric alignment, setting the detection area crossing the gas-liquid-gas interface as the region of interest; performing a two-dimensional fast Fourier transform on the fringe intensity field in the region of interest, locating the direction and frequency of the main fringe carrier frequency, constructing a single-sideband bandpass filter to extract the main carrier frequency sideband component, inversely transforming to form an analytical signal and solving the wrapped phase through the amplitude and angle operation; using the undisturbed reference fringe image as the reference frequency, the phase difference value sequence is obtained by the phase difference correction and the phase difference correction. The wrapped phase is used as the baseline, and time domain phase unwrapping and spatial domain phase unwrapping are performed on each frame of the wrapped phase in the order of sampling time to remove the double circular constant jump and global phase drift; and the phase difference with the baseline is calculated on the unwrapped phase, and a weighted average is performed along the preset optical path or in the region of interest to form a single-value phase difference, and a phase difference value sequence sorted by timestamp is output; in the above process, a lower limit threshold is set for the fringe contrast and the signal-to-noise ratio, and frames that do not meet the threshold are eliminated, and the phase difference value sequence and the corresponding quality mark are recorded for subsequent refractive index conversion and distribution field construction.

[0017] Included in S2: S2-1: The infrared transmission spectrum intensity after passing through the gas layer is collected within the detection area, while the reference spectrum intensity under gas-free conditions is recorded. The transmission spectrum intensity is divided by the reference spectrum intensity to calculate the transmittance sequence wavelength by wavelength. S2-1 is explained as follows: within the detection area, the infrared transmission spectrum intensity after passing through the gas layer is first collected by the detector, while the reference spectrum intensity under the same optical path conditions is collected in the absence of gas. Subsequently, at each identical wavelength point, the transmission spectrum intensity value is divided by the corresponding reference spectrum intensity value, and the results are arranged in sequence to obtain a transmittance sequence arranged in wavelength order. This transmittance sequence fully characterizes the degree of infrared light absorption by the gas. S2-2: Take the negative logarithm of the transmittance sequence point by point and multiply it by the effective optical path length to convert the transmittance into the absorption value corresponding to each wavelength, thereby forming an absorption spectrum curve, which serves as the input data for concentration inversion; It should be noted in S2-2 that the transmittance sequence obtained in S2-1 should be calculated point by point. First, the negative logarithm of the transmittance value at each wavelength is taken to convert the light intensity attenuation into absorption intensity. Then, this absorption intensity is multiplied by the effective optical path length of the detection path to obtain the absorption value corresponding to that wavelength. The above calculations are repeated point by point according to the wavelength order of the transmittance sequence, and finally a complete absorption spectrum curve is formed. This absorption spectrum curve is directly used as the input data for the subsequent concentration inversion step. S2-3: Compare the absorption spectrum curve with a pre-established harmful gas standard absorption database wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitting curve and the actual absorption spectrum curve reaches a preset lower limit, and output the first concentration result of the target harmful gas; It should be noted in S2-3 that the absorption spectrum curve obtained in S2-2 should be compared with the pre-established harmful gas standard absorption database wavelength by wavelength. The specific process is: at each wavelength point, the value of the actual absorption spectrum curve is compared with the absorption cross section of the corresponding gas in the standard database; then the least squares fitting method is used to continuously adjust the concentration coefficient of each gas, so that the residual between the theoretical fitting curve obtained by the database weighting and the actual absorption spectrum curve is gradually reduced. When the residual is reduced to the preset lower limit, the fitting process is terminated. At this time, the concentration coefficient of each gas obtained is the solved concentration result, from which the first concentration result of the target harmful gas is extracted as the output data; S2-4: Corresponding the first concentration result to the spatial position of the infrared light path point by point according to the sampling time, and storing it in the grid unit of the refractive index distribution field, while simultaneously maintaining the refractive index data and the concentration data in the distribution field, thereby establishing a one-to-one mapping relationship between the two; In S2-4, the first thing to process is the first concentration result obtained in S2-3. This concentration result must correspond one-to-one with the acquisition time, and at the same time, it must be placed point by point in the grid cell of the refractive index distribution field in combination with the spatial position of the infrared light path. Through this point-by-point correspondence, each grid cell in the refractive index distribution field not only contains the refractive index value, but also stores the concentration value at the corresponding time and spatial position. In this way, the refractive index and concentration are strictly bound within the same coordinate system, forming a one-to-one mapping relationship, providing a unified data foundation for both refractive index and concentration for subsequent iterative calculations.

[0018] Included in S3: S3-1: Calculate a theoretical absorption spectrum curve based on the first concentration result, and subtract the theoretical absorption spectrum curve from the actual infrared spectrum curve wavelength by wavelength to obtain a residual sequence. When calculating the theoretical absorption spectrum curve, the first concentration result is used as the core input. The standard absorption cross-section curve corresponding to each target gas is called from the database, and the absorption cross-section is multiplied by the first concentration result wavelength by wavelength to obtain the theoretical absorption contribution of each gas. Subsequently, the absorption contributions of all gases are weighted and superimposed according to the optical path length at each wavelength point to form a complete theoretical absorption spectrum curve, which is used for wavelength-by-wavelength comparison with the actual infrared spectrum. In addition, when subtracting the theoretical absorption spectrum curve from the actual infrared spectrum curve wavelength by wavelength, it is necessary to take the values ​​of the two curves at the same wavelength point, and subtract the absorption of the theoretical spectrum from the absorption of the actual spectrum, and calculate the residual sequence in sequence; S3-2: Back-projecting the residual sequence according to the infrared light path and the spatial coordinates of the detection area to obtain a refractive index update value matrix for representing the refractive index correction amount of the gas-liquid-gas interface at each spatial position; The refractive index update value matrix includes refractive index correction values ​​calculated by back-calculating the residual signal at each grid cell in the detection area, and each row and column of the refractive index update value matrix corresponds to a position point on the spatial coordinate axis, and is used to record the refractive index correction distribution of the gas-liquid-gas interface in the entire area; When performing back-projection, the residual signal at each wavelength point in the residual sequence needs to be traced back along the corresponding infrared light path propagation path, and the residual is distributed to the grid cells of the detection area that the light path passes through. Specifically, the residual signal is weighted and superimposed proportionally to each grid position passed through based on the path length of the light in spatial coordinates, thereby forming a refractive index update value matrix for the entire detection area. Because the residual reflects the difference between theoretical absorption and actual absorption, and this difference is directly due to the offset of the refractive index in the optical path, the refractive index update value matrix obtained by back-projection can represent the refractive index correction amount at each position of the gas-liquid-gas interface in terms of spatial distribution.

[0019] Also included in S3: S3-3: Inputting the refractive index update value matrix into the infrared absorption inversion operation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of the infrared light propagating in the detection area. The corrected optical path length is then substituted into the transmittance to absorption conversion formula to recalculate the updated absorption spectrum curve. The updated absorption spectrum curve is then fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value. The new concentration value is then arranged in the order of sampling time to form a concentration update value sequence; The conversion formula for absorption is: take the negative logarithm of the transmittance at each wavelength and multiply it by the actual optical path length to calculate the absorption value corresponding to the wavelength; In addition, when fitting the updated absorption spectrum curve to the standard absorption database wavelength by wavelength, it is necessary to compare the absorption value of the actual curve with the standard absorption cross-section value of the corresponding gas in the database at each wavelength point. Then, by adjusting the concentration coefficient of each gas, the theoretical curve calculated by the concentration coefficient and the standard absorption cross-section is made to fit the actual curve as closely as possible, thereby minimizing the difference between the two at all wavelength points. S3-4: establishing a one-to-one correspondence between the refractive index update value matrix and the concentration update value sequence, and storing the correspondence in the refractive index distribution field to provide input conditions for subsequent reciprocating iterative steps; In S3-4, the refractive index update value matrix and the concentration update value sequence need to be aligned point by point, including: first, according to the same sampling time index, each value in the concentration update value sequence is matched with each spatial unit of the refractive index update value matrix under the same time index; then, the refractive index update value and the concentration update value at that moment are simultaneously stored in the grid unit of the refractive index distribution field, so that the two types of data are stored consistently under the same spatial coordinates; finally, the refractive index distribution field with one-to-one correspondence is output to directly provide the input conditions including the refractive index correction amount and the concentration update amount for the subsequent reciprocating iterative steps.

[0020] Included in S4: S4-1: The refractive index update value matrix and the concentration update value sequence are used as initial inputs, and the grid coordinate information in the refractive index distribution field is loaded to construct the input conditions for the first round of iteration; S4-2: In the first round of iteration, the optical path length is corrected using the refractive index update value matrix, the absorption spectrum curve is corrected using the concentration update value sequence, and infrared absorption inversion fitting is performed to output a new refractive index update value matrix and concentration update value sequence.

[0021] Also included in S4: S4-3: re-inputting the newly output refractive index update value matrix and concentration update value sequence into the iterative process to form a reciprocating cycle, and calculating the residual amplitude after each cycle until the residual amplitude is lower than the preset lower limit; S4-4: When the residual amplitude reaches a preset lower limit, the refractive index update value matrix and concentration update value sequence obtained from this round of iteration are stored as convergence results in the refractive index distribution field, which serves as input conditions for subsequent steps. In this scheme, the residual amplitude refers to the overall value obtained after performing statistical operations (such as taking the absolute value average or root mean square) on the wavelength-by-wavelength residual sequence, which is used to measure the overall size of the residual. The residual is the point-to-point difference value for each wavelength, and the residual amplitude is the overall quantitative result of the residual sequence. It should be noted that in S4-1, the refractive index update value matrix and the concentration update value sequence are used as the initial input, and the grid coordinate information in the refractive index distribution field is called at the same time, so that the refractive index update value and the concentration update value can be mapped to specific spatial positions point by point, thereby constructing the complete input conditions for the first round of iteration; It should be noted that in S4-2, the optical path length is corrected point by point using the refractive index update value matrix, and the absorption spectrum curve is corrected point by point using the concentration update value sequence, an infrared absorption inversion fitting operation is performed on the corrected optical path and absorption curve, and a new refractive index update value matrix and concentration update value sequence are output; In S4-3, the new refractive index update value matrix and concentration update value sequence output in the previous round are re-input into the iterative calculation process to form a reciprocating loop. After each round of iteration, the residual amplitude is calculated for each wavelength to determine whether the residual amplitude is lower than the preset lower limit. If the condition is not met, the next round of loop is continued. In S4-4: when the residual amplitude is lower than the preset lower limit, the refractive index update value matrix and the concentration update value sequence obtained in this round of iteration are stored as convergence results in the refractive index distribution field, so that the refractive index distribution field retains the converged refractive index and concentration data in both spatial and temporal dimensions, and serves as input conditions for subsequent steps.

[0022] In S5, including: S5-1: From the output of the iterative chain consisting of the refractive index update value matrix and the concentration update value sequence, extract the concentration update values ​​and the refractive index update values ​​in the same sampling time sequence, and correspond the concentration update values ​​and the refractive index update values ​​point by point in the grid coordinates of the refractive index distribution field to form a data set including the sampling time, spatial position, concentration update value, and refractive index update value; wherein, because the output of the iterative chain itself consists of the concentration update value sequence and the refractive index update value matrix, both of which have sampling time indexes and spatial coordinate indexes of the refractive index distribution field, respectively, when stored point by point in the same coordinate system, a data set including the sampling time, spatial position, concentration update value, and refractive index update value can be formed; S5-2: Dividing the data set into multiple input sequences according to consecutive sampling moments, and performing normalization processing on the concentration update values ​​and refractive index update values ​​in each input sequence to obtain concentration update value sequences and refractive index update value sequences arranged by time index; S5-3: Inputting the concentration update value sequence and the refractive index update value sequence simultaneously into a time series modeling algorithm based on a recurrent neural network, performing forward calculations at each sampling moment in the time series modeling algorithm to obtain a sequence of future concentration prediction values ​​covering a preset time period; wherein the time series modeling algorithm of the recurrent neural network is a neural network method that can use the input and hidden state of the previous moment to calculate the output of the current moment, and is used to process a data sequence with a time sequence; The reason why forward calculations can be performed at each sampling moment in the time series modeling algorithm is that the updated concentration and refractive index values ​​at each sampling moment are input into the network together with the hidden state at the previous moment. The network outputs the prediction result at the current moment while maintaining time dependence. As the time steps continue to unfold, the network can gradually infer the future value distribution based on the accumulated historical information, thereby obtaining a sequence of future concentration prediction values ​​covering a preset time period. S5-4: writing the future concentration prediction value sequence into the grid cells of the refractive index distribution field point by point according to the sampling time index and spatial position, while keeping the refractive index update value unchanged while writing the future concentration prediction value, thereby forming a data set containing the concentration prediction values ​​at the future sampling times in the refractive index distribution field; It should be noted in S5-4 that: for the future concentration prediction value sequence obtained in the time series modeling algorithm, the grid cells with the same time index in the refractive index distribution field should be matched one by one according to the sampling time index, and the corresponding future concentration prediction value should be written into the spatial position of each grid cell; in the process of writing the future concentration prediction value, the original refractive index update value is kept unchanged, so that each grid cell contains both the refractive index update value and the future concentration prediction value, thereby forming a concentration prediction data set covering the future sampling moments in the refractive index distribution field; S5-5: Using the future concentration prediction value in the data set as the concentration input condition for the next round of reciprocating iterative chain, and simultaneously using the refractive index update value of the refractive index distribution field at the corresponding sampling moment as the refractive index input condition, thereby completing the feedback update of the input condition for the next round of reciprocating iterative chain; It should be noted in S5-5 that: for the data set formed in the refractive index distribution field, the future concentration prediction values ​​at each sampling moment should be extracted therefrom, and these values ​​should be used in turn as the concentration input conditions for the next round of reciprocating iterative chain; at the same time, the refractive index update values ​​at the same sampling moment should be read from the refractive index distribution field, and these values ​​should be used as the refractive index input conditions for the next round of reciprocating iterative chain; through this synchronous transmission of concentration input and refractive index input, the iterative process can directly use the prediction results and existing refractive index data in the new round of calculations, realizing a closed-loop connection of feedback updates.

[0023] Working Principle: By using the phase processing of interference fringes, the refractive index change under bubble disturbance is converted into a refractive index distribution field that is continuous in space and time. On this basis, the refractive index distribution field is used to correct the propagation path of infrared light, convert the transmittance into an absorption spectrum, and calculate the first concentration result, which is then written into the refractive index distribution field according to the sampling time and spatial position. Subsequently, the first concentration result is compared with the gas absorption curve in the standard database to generate a theoretical absorption spectrum, which is then subtracted wavelength by wavelength from the actual infrared spectrum to obtain a residual sequence. These residuals are then back-projected along the optical path to form a refractive index update value matrix, which is used to correct the optical path and refit the concentration update value sequence. The refractive index update value matrix and the concentration update value sequence are used as inputs to a reciprocating iterative chain. After each round of calculation, the residual amplitude is used to determine whether convergence has occurred. If not, the iteration continues until convergence is achieved, and the result is written into the refractive index distribution field. At the same time, the refractive index update value and the concentration update value are input into a time series modeling algorithm such as a recurrent neural network according to a time window, and forward calculation is performed at each sampling moment to obtain a sequence of future concentration prediction values; these prediction values ​​are written into the refractive index distribution field point by point according to time and spatial position, and serve as the input for the next round of iteration, together with the refractive index update value to form a new initial condition; based on this, this scheme realizes the closed-loop operation of refractive index calculation, infrared inversion, residual correction, iterative convergence and time prediction in a unified refractive index distribution field.

[0024] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for in-situ harmful gas monitoring of sewage treatment plant structures, characterized in that: include: S1: In a sludge tank or anaerobic tank, the refractive index disturbance data of the local gas-liquid-gas interface caused by the release of bubbles is collected to form a refractive index distribution field that changes with time; S2: Modeling the infrared light propagation path based on the refractive index distribution field, performing infrared absorption inversion calculation, solving a first concentration result of the harmful gas, and establishing a mapping relationship between the first concentration result and the refractive index distribution field; S3: Calculate the residual between the theoretical spectrum corresponding to the first concentration result and the actual infrared spectrum, and back-project the residual result into the refractive index distribution field to form an updated refractive index value; Correcting the infrared absorption inversion operation based on the refractive index update value to generate a concentration update value; S4: constructing a reciprocating iterative chain between the refractive index update value and the concentration update value, forming a converged result through multiple rounds of iteration, and simultaneously outputting the refractive index update value and the concentration update value in each round of iteration for input into the next round of iteration; S5: The concentration update value and the refractive index update value output by the reciprocating iterative chain are simultaneously input into the timing modeling algorithm to generate a preset release time prediction result of the harmful gas, and the prediction result is fed back to the refractive index distribution field to update the input conditions of the next round of iteration.

2. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 1, characterized in that: Included in S1: S1-1: Arranging an incident light beam and a reference light beam along the bubble release area in a sludge tank or an anaerobic tank, collecting a fringe image of optical interference, and extracting a phase difference value sequence from the fringe image; S1-2: Expanding the phase difference value sequence into a continuous phase curve according to the sampling time sequence, calculating the phase difference components between adjacent sampling points, and converting the phase difference components into a refractive index increment sequence according to a preset refractive index conversion coefficient; S1-3: Corresponding the refractive index increment sequence to the spatial coordinates of the detection area point by point, calculating a refractive index perturbation matrix, and forming a refractive index perturbation distribution covering the entire gas-liquid-gas interface through matrix interpolation; S1-4: performing an accumulation operation on the refractive index disturbance matrix according to the time index to generate a refractive index distribution field that varies with time.

3. The in-situ harmful gas monitoring method for a sewage treatment plant structure according to claim 2, characterized in that: Included in S2: S2-1: Collect the infrared transmission spectrum intensity through the gas layer in the detection area, and record the reference spectrum intensity under gas-free conditions. Divide the transmission spectrum intensity by the reference spectrum intensity and calculate the transmittance sequence wavelength by wavelength. S2-2: Take the negative logarithm of the transmittance sequence point by point and multiply it by the effective optical path length to convert the transmittance into the absorption value corresponding to each wavelength, thereby forming an absorption spectrum curve, which serves as the input data for concentration inversion; S2-3: Compare the absorption spectrum curve with a pre-established harmful gas standard absorption database wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitting curve and the actual absorption spectrum curve reaches a preset lower limit, and output the first concentration result of the target harmful gas; S2-4: The first concentration result is mapped point by point to the spatial position of the infrared light path according to the sampling time, and stored in the grid unit of the refractive index distribution field, and the refractive index data and concentration data are simultaneously maintained in the distribution field, thereby establishing a one-to-one mapping relationship between the two.

4. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 3, characterized in that: Included in S3: S3-1: Calculate a theoretical absorption spectrum curve based on the first concentration result, and subtract the theoretical absorption spectrum curve from the actual infrared spectrum curve wavelength by wavelength to obtain a residual sequence; S3-2: Back-projecting the residual sequence according to the infrared light path and the spatial coordinates of the detection area to obtain a refractive index update value matrix for representing the refractive index correction amount of the gas-liquid-gas interface at each spatial position; The refractive index update value matrix includes the refractive index correction values ​​obtained by back-calculating the residual signal on each grid cell in the detection area. Each row and each column of the refractive index update value matrix corresponds to a position point on the spatial coordinate axis, which is used to record the refractive index correction distribution of the gas-liquid-gas interface in the entire area.

5. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 4, characterized in that: Also included in S3: S3-3: Inputting the refractive index update value matrix into the infrared absorption inversion operation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of the infrared light propagating in the detection area. The corrected optical path length is then substituted into the transmittance to absorption conversion formula to recalculate the updated absorption spectrum curve. The updated absorption spectrum curve is then fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value. The new concentration value is then arranged in the order of sampling time to form a concentration update value sequence; S3-4: Establish a one-to-one correspondence between the refractive index update value matrix and the concentration update value sequence, and store the correspondence in the refractive index distribution field to provide input conditions for subsequent reciprocating iterative steps.

6. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 5, characterized in that: Included in S4: S4-1: The refractive index update value matrix and the concentration update value sequence are used as initial inputs, and the grid coordinate information in the refractive index distribution field is loaded to construct the input conditions for the first round of iteration; S4-2: In the first round of iteration, the optical path length is corrected using the refractive index update value matrix, the absorption spectrum curve is corrected using the concentration update value sequence, and infrared absorption inversion fitting is performed to output a new refractive index update value matrix and concentration update value sequence.

7. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 6, characterized in that: Also included in S4: S4-3: re-inputting the newly output refractive index update value matrix and concentration update value sequence into the iterative process to form a reciprocating cycle, and calculating the residual amplitude after each cycle until the residual amplitude is lower than the preset lower limit; S4-4: When the residual amplitude reaches a preset lower limit, the refractive index update value matrix and the concentration update value sequence obtained in this round of iteration are stored as convergence results in the refractive index distribution field as input conditions for subsequent steps.

8. The method for in-situ harmful gas monitoring of a sewage treatment plant structure according to claim 7, characterized in that: In S5, including: S5-1: Extracting concentration update values ​​and refractive index update values ​​from the output result of the reciprocating iterative chain consisting of the refractive index update value matrix and the concentration update value sequence according to the same sampling time sequence, and assigning the concentration update values ​​and the refractive index update values ​​to each other point by point in the grid coordinates of the refractive index distribution field, thereby forming a data set including the sampling time, spatial position, concentration update value, and refractive index update value; S5-2: Dividing the data set into multiple input sequences according to consecutive sampling moments, and performing normalization processing on the concentration update values ​​and refractive index update values ​​in each input sequence to obtain concentration update value sequences and refractive index update value sequences arranged by time index; S5-3: Inputting the concentration update value sequence and the refractive index update value sequence simultaneously into a time series modeling algorithm based on a recurrent neural network, performing forward calculations at each sampling moment in the time series modeling algorithm to obtain a future concentration prediction value sequence covering a preset time length; S5-4: writing the future concentration prediction value sequence into the grid cells of the refractive index distribution field point by point according to the sampling time index and spatial position, while keeping the refractive index update value unchanged while writing the future concentration prediction value, thereby forming a data set containing the concentration prediction values ​​at the future sampling times in the refractive index distribution field; S5-5: The future concentration prediction value in the data set is used as the concentration input condition for the next round of reciprocating iterative chain, and the refractive index update value of the refractive index distribution field at the corresponding sampling moment is used as the refractive index input condition, thereby completing the feedback update of the input conditions for the next round of reciprocating iterative chain.

Images