A method of in situ monitoring of harmful gases in a sewage treatment plant structure

Abstract

The application discloses a kind of in-situ harmful gas monitoring methods of sewage treatment plant structure, specifically in-situ harmful gas monitoring field, including in sludge pool or anaerobic tank, the refractive index disturbance data caused by local gas-liquid-gas interface due to bubble release is collected, and the refractive index distribution field that changes with time is formed;Based on the refractive index distribution field, the infrared light propagation path is modeled, the infrared absorption inversion operation is executed, the first concentration result of harmful gas is solved out, and the mapping relationship between the first concentration result and the refractive index distribution field is established.Coupling calculation is carried out between the refractive index distribution field caused by bubble disturbance and infrared absorption inversion process, and a reciprocating iteration and time sequence prediction chain is constructed to correct dynamic optical distortion and stabilize the real release intensity of harmful gas.

Description

Technical Field

[0001] This invention relates to the field of in-situ hazardous gas monitoring technology, and more specifically, to a method for in-situ hazardous gas monitoring of wastewater treatment plant structures. Background Technology

[0002] In the sludge tanks or anaerobic tanks of wastewater treatment plants, microbubbles at the bottom are periodically released, often carrying hazardous gases such as hydrogen sulfide and methane. As the bubbles rise, they form a complex gas-liquid-gas interface layer 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 drastic fluctuations in the bubble volume fraction, resulting in dynamic bending and distortion of the optical path.

[0003] Such effects are transient and highly non-uniform, and existing infrared detection methods that rely on average absorption lines are difficult to capture stably, often resulting in signal saturation or spectral distortion, which in turn masks the true gas release intensity and distribution characteristics.

[0004] Therefore, the core problem with existing technologies is that the dynamic optical distortion caused by bubble eruption destroys the accuracy of infrared monitoring, making in-situ detection unable to effectively characterize the actual release process of harmful gases inside structures at critical moments. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an in-situ method for monitoring harmful gases in wastewater treatment plant structures. This method couples the refractive index distribution field caused by bubble disturbance with the infrared absorption inversion process to construct an iterative and time-series prediction chain, thereby correcting dynamic optical distortion and stably solving the true release intensity of harmful gases, thus addressing the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for in-situ monitoring of harmful gases in sewage treatment plant structures, comprising:

[0007] S1: In the sludge tank or anaerobic tank, collect the refractive index disturbance data of the local gas-liquid-gas interface caused by the release of bubbles to form a refractive index distribution field that changes with time.

[0008] S2: Model the infrared light propagation path based on the refractive index distribution field, perform infrared absorption inversion calculation, solve for the first concentration result of the harmful gas, and establish a mapping relationship between the first concentration result and the refractive index distribution field;

[0009] S3: Calculate the residual between the theoretical spectrum and the actual infrared spectrum corresponding to the first concentration result, back-project the result of the residual onto the refractive index distribution field to form a refractive index update value; correct the infrared absorption inversion operation based on the refractive index update value to generate a concentration update value.

[0010] S4: Construct a reciprocating iterative chain between the refractive index update value and the concentration update value, form a convergent result through multiple rounds of iteration, and output the refractive index update value and the concentration update value simultaneously in each round of iteration for input in the next round of iteration;

[0011] S5: Simultaneously input the concentration update value and refractive index update value output by the reciprocating iterative chain into the time-series modeling algorithm to generate a preset time release prediction result for harmful gases, and feed the prediction result back to the refractive index distribution field to update the input conditions for the next iteration.

[0012] In a preferred embodiment, S1 includes:

[0013] S1-1: Arrange the incident beam and the reference beam along the bubble release area in the sludge tank or anaerobic tank, collect the fringe image of optical interference, and extract the phase difference value sequence from the fringe image;

[0014] S1-2: Expand the phase difference value sequence into a continuous phase curve according to the sampling time sequence, calculate the phase difference components between adjacent sampling points, and convert the phase difference components into a refractive index increment sequence according to the preset refractive index conversion coefficient.

[0015] S1-3: The refractive index increment sequence is matched point by point with the spatial coordinates of the detection area to calculate the refractive index perturbation matrix, and the refractive index perturbation distribution covering the entire gas-liquid-gas interface is formed by matrix interpolation.

[0016] S1-4: Perform an accumulation operation on the refractive index perturbation matrix according to the time index to generate a refractive index distribution field that varies with time.

[0017] In a preferred embodiment, S2 includes:

[0018] 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.

[0019] 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. The absorption spectrum curve serves as the input data for concentration inversion.

[0020] S2-3: Compare the absorption spectrum curve with the pre-established standard absorption database of harmful gases wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitted curve and the actual absorption spectrum curve reaches the preset lower limit, and output the first concentration result of the target harmful gas.

[0021] S2-4: The first concentration result is stored in the grid cell of the refractive index distribution field according to the point-to-point correspondence between the sampling time and the spatial position of the infrared light path. The refractive index data and concentration data are maintained in the distribution field at the same time, thereby establishing a one-to-one mapping relationship between the two.

[0022] In a preferred embodiment, S3 includes:

[0023] S3-1: Calculate the 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 the residual sequence;

[0024] S3-2: The residual sequence is back-projected according to the spatial coordinates of the infrared light path and the detection area to obtain the refractive index update matrix, which is used to characterize the refractive index correction amount of the gas-liquid-gas interface at each spatial position.

[0025] The refractive index update matrix includes refractive index correction values ​​calculated back from the residual signals on each grid cell in the detection area. Each row and column of the refractive index update 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.

[0026] In a preferred embodiment, S3 further includes:

[0027] S3-3: Input the refractive index update value matrix into the infrared absorption inversion calculation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of infrared light in the detection area. Then, the corrected optical path length is substituted into the conversion formula from transmittance to absorption to recalculate the updated absorption spectrum curve. Subsequently, the updated absorption spectrum curve is fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value, and the new concentration update value sequence is formed according to the sampling time order.

[0028] 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 iterative steps.

[0029] In a preferred embodiment, S4 includes:

[0030] S4-1: Use the refractive index update matrix and concentration update sequence as initial inputs, and load the grid coordinate information in the refractive index distribution field to construct the input conditions for the first iteration;

[0031] S4-2: In the first iteration, the optical path length is corrected with the refractive index update matrix, the absorption spectrum curve is corrected with the concentration update sequence, and infrared absorption inversion fitting is performed to output a new refractive index update matrix and a concentration update sequence.

[0032] In a preferred embodiment, S4 further includes:

[0033] S4-3: Re-input the newly output refractive index update matrix and concentration update sequence into the iterative process to form a recurring loop, and calculate the residual amplitude after each loop until the residual amplitude is lower than the preset lower limit;

[0034] S4-4: When the residual amplitude reaches the preset lower limit, the refractive index update matrix and concentration update sequence obtained from this iteration are stored in the refractive index distribution field as convergence results, and used as input conditions for subsequent steps.

[0035] In a preferred embodiment, S5 includes:

[0036] S5-1: In the output of the iterative chain consisting of the refractive index update value matrix and the concentration update value sequence, the concentration update value and the refractive index update value are extracted in the same sampling time order, and the concentration update value and the refractive index update value are matched point by point in the grid coordinates of the refractive index distribution field to form a data set containing the sampling time, spatial position, concentration update value and refractive index update value.

[0037] S5-2: Divide the data set into multiple input sequences according to the continuous sampling time, and perform normalization processing on the concentration update value and refractive index update value in each input sequence to obtain the concentration update value sequence and refractive index update value sequence arranged by time index;

[0038] S5-3: Input the concentration update value sequence and the refractive index update value sequence into the time series modeling algorithm based on the recurrent neural network at the same time. Perform forward calculation at each sampling time in the time series modeling algorithm to obtain the future concentration prediction value sequence covering the preset time.

[0039] S5-4: The future concentration prediction value sequence is written point by point into the grid cell of the refractive index distribution field according to the sampling time index and spatial location. While writing the future concentration prediction value, the refractive index update value remains unchanged, thereby forming a dataset containing the concentration prediction value at the future sampling time in the refractive index distribution field.

[0040] S5-5: Use the predicted future concentration values ​​in the dataset as the concentration input conditions for the next round of iterative chain, and use the updated refractive index values ​​of the refractive index distribution field at the corresponding sampling time as the refractive index input conditions, thereby completing the feedback update of the input conditions for the next round of iterative chain.

[0041] The technical effects and advantages of this invention are as follows:

[0042] 1. By converting the phase change under bubble perturbation 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.

[0043] 2. By establishing a mapping relationship between the refractive index distribution field and infrared absorption inversion, the concentration results can correspond one-to-one with the spatial refractive index changes, thereby maintaining the spatial resolution of the detection results in a non-uniform perturbation field.

[0044] 3. By introducing back projection calculation of the residual sequence, the difference between the theoretical spectrum and the actual spectrum is transformed into the refractive index update value, realizing real-time correction of optical path modeling and avoiding concentration deviation caused by model error;

[0045] 4. By mutually correcting the refractive index update value and the concentration update value in the iterative chain, a convergence mechanism under residual amplitude control is formed, ensuring that the monitoring results obtain a stable and reliable numerical solution after multiple rounds of correction.

[0046] 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 time, the concentration prediction results for a preset duration are obtained, thus realizing the early characterization of future gas release trends.

[0047] 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 are kept consistent under a unified coordinate system, providing complete data support for subsequent continuous iterations and multi-time period predictions. Attached Figure Description

[0048] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Refer to the instruction manual appendix Figure 1 An embodiment of the present invention provides an in-situ method for monitoring harmful gases in wastewater treatment plant structures, comprising:

[0051] S1 Acquisition of refractive index perturbation distribution: In the sludge tank or anaerobic tank, the refractive index perturbation data of the local gas-liquid-gas interface caused by the release of bubbles are collected to form a refractive index distribution field that changes with time.

[0052] S2 Constructing the initial solution of the infrared spectrum: Model the infrared light propagation path based on the refractive index distribution field, perform infrared absorption inversion operation, solve for the first concentration result of the harmful gas, and establish a mapping relationship between the first concentration result and the refractive index distribution field;

[0053] S3 generates a residual correction chain: Calculate the residual between the theoretical spectrum and the actual infrared spectrum corresponding to the first concentration result, back-project the result of the residual onto the refractive index distribution field to form a refractive index update value; Based on the refractive index update value, correct the infrared absorption inversion operation to generate a concentration update value.

[0054] S4 Reciprocating Convergence Iteration: A reciprocating iterative chain is constructed between the refractive index update value and the concentration update value. A convergence result is formed through multiple rounds of iteration, and the refractive index update value and the concentration update value are output simultaneously in each round of iteration for input in the next round of iteration.

[0055] S5 Parallel Construction of Time-Series Prediction: The concentration update value and refractive index update value output by the reciprocating iterative chain are simultaneously input into the time-series modeling algorithm to generate a preset time-based release prediction result for harmful gases, and the prediction result is fed back to the refractive index distribution field to update the input conditions for the next iteration.

[0056] S1 includes:

[0057] S1-1: Arrange the incident beam and the reference beam along the bubble release area in the sludge tank or anaerobic tank, collect the fringe image of optical interference, and extract the phase difference value sequence from the fringe image;

[0058] S1-2: Expand the phase difference sequence into a continuous phase curve according to the sampling time sequence, calculate the phase difference components between adjacent sampling points, and convert the phase difference components into a refractive index increment sequence according to the preset refractive index conversion coefficient; wherein, when expanding the phase difference sequence into a continuous phase curve, it is necessary to perform phase expansion operation on the phase difference values ​​arranged according to the sampling time: when encountering a jump point that crosses positive and 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 disturbance process that changes with time;

[0059] In addition, the phase difference components between adjacent sampling points are 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 the sampling time interval as an index, thereby forming a difference component sequence that reflects the instantaneous phase change rate.

[0060] The preset refractive index conversion factor is a proportional coefficient determined in advance based on the optical path length and the wavelength of the probe light, used to directly convert the phase difference component into the corresponding refractive index change value.

[0061] S1-3: The refractive index increment sequence is mapped point-by-point to the spatial coordinates of the detection area to calculate the 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 performed pixel-coordinate registration according to the calibrated imaging geometry and optical path parameters, the refractive index increment of each sampling point is mapped to the regular grid cell of the detection area, and weighted accumulation and normalization are performed using the effective optical path through the cell as the weight to obtain the refractive index perturbation value of each grid cell; neighborhood interpolation is performed on the grid cells not covered by sampling, and the matrix is ​​the refractive index perturbation matrix by row and column index;

[0062] Matrix interpolation refers to the numerical calculation of the refractive index perturbation value of known grid points according to their spatial position relationship, and filling the missing values ​​with the weighted results of adjacent points, thereby forming a continuously covered refractive index perturbation distribution on the entire gas-liquid-gas interface.

[0063] S1-4: Perform an accumulation operation on the refractive index perturbation matrix according to the time index to generate a refractive index distribution field that varies with time;

[0064] It should be noted that for S1, in practical applications, an optical interferometer (such as a Michelson interferometer or a Mach-Zehnder interferometer) can usually be used. An incident beam that passes through the gas-liquid interface is arranged in the bubble release area of ​​the sludge tank or anaerobic tank, and a reference beam that does not pass through the bubble disturbance area is set up at the same time. The two beams eventually converge on the interferometer detector to compare the phase changes.

[0065] The incident beam is light that passes through the detection area and is affected by bubble disturbance; the reference beam is light that does not pass through the detection area and maintains a stable phase. The two beams are superimposed to form interference fringes, and the phase difference can reflect the change in refractive index.

[0066] The calculation process for obtaining the phase difference sequence from the fringe image of optical interference in S1-1 includes: performing camera dark field correction and flat field correction on the acquired interference fringe image, completing lens distortion correction and inter-frame geometric registration, and 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 within the region of interest, locating the carrier frequency direction and frequency of the main fringe, constructing a single-sideband bandpass filter to extract the main carrier frequency sideband component, performing an inverse transform to form an analytic signal, and solving for the wrapped phase through phase angle calculation; using an unperturbed reference fringe image... Using the wrapped phase as the baseline, temporal and spatial phase unwrapping is performed on the wrapped phase of each frame in the order of sampling time to remove the double circumferential constant jump and global phase drift. The phase difference with the baseline is calculated on the unwrapped phase, and a weighted average is performed along a preset optical path or in the region of interest to form a single-value phase difference. The phase difference value sequence is output in timestamp order. In the above process, a lower limit threshold is set for the fringe contrast and signal-to-noise ratio. Frames that do not meet the threshold are removed, and the phase difference value sequence and corresponding quality mark are recorded for subsequent refractive index conversion and distribution field construction.

[0067] S2 includes:

[0068] S2-1: In the detection area, the infrared transmission spectral intensity after passing through the gas layer is collected, and the reference spectral intensity under gas-free conditions is recorded. The transmittance sequence is calculated by dividing the transmission spectral intensity by the reference spectral intensity and proceeding wavelength by wavelength. S2-1 is explained as follows: In the detection area, an infrared transmission spectral intensity after passing through the gas layer is first collected by a detector, and a reference spectral intensity under the same optical path conditions is also collected. Then, at each same wavelength point, the value of the transmission spectral intensity is divided by the corresponding reference spectral intensity value. The results are arranged sequentially to obtain a transmittance sequence arranged in wavelength order. This transmittance sequence fully characterizes the degree of absorption of infrared light by the gas.

[0069] 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. The absorption spectrum curve serves as the input data for concentration inversion.

[0070] In S2-2, it should be noted that for the transmittance sequence obtained in S2-1, the calculation should be performed point by point. First, take the negative logarithm of the transmittance value at each wavelength point to convert the light intensity attenuation into absorption intensity; then multiply the absorption intensity by the effective optical path length of the detection path to obtain the absorption value corresponding to that wavelength. Repeat the above calculation point by point according to the wavelength order of the transmittance sequence to finally form a complete absorption spectrum curve, which is directly used as the input data for the subsequent concentration inversion step.

[0071] S2-3: Compare the absorption spectrum curve with the pre-established standard absorption database of harmful gases wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitted curve and the actual absorption spectrum curve reaches the preset lower limit, and output the first concentration result of the target harmful gas.

[0072] In section S2-3, it should be noted that the absorption spectrum curves obtained in S2-2 should be compared wavelength by wavelength with a pre-established database of standard absorption curves for harmful gases. Specifically, at each wavelength, the value of the actual absorption spectrum curve is compared with the absorption cross-section of the corresponding gas in the standard database. Then, a least-squares fitting method is used to continuously adjust the concentration coefficients of each gas, gradually reducing the residual between the theoretically fitted curve obtained from the database weighting and the actual absorption spectrum curve. When the residual decreases to a preset lower limit, the fitting process terminates. The concentration coefficients of each gas obtained at this point are the solved concentration results, from which the first concentration result of the target harmful gas is extracted as the output data.

[0073] S2-4: The first concentration result is stored in the grid cell of the refractive index distribution field according to the point-to-point correspondence between the sampling time and the spatial position of the infrared light path. The refractive index data and concentration data are maintained in the distribution field at the same time, thereby establishing a one-to-one mapping relationship between the two.

[0074] In S2-4, the first step is to process the initial concentration result obtained in S2-3. This concentration result must correspond one-to-one with the acquisition time, and simultaneously be placed point-by-point into the grid cells of the refractive index distribution field, taking into account the spatial position of the infrared optical 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, refractive index and concentration are strictly bound within the same coordinate system, forming a one-to-one mapping relationship, providing a unified data foundation with both refractive index and concentration for subsequent iterative calculations.

[0075] S3 includes:

[0076] S3-1: Calculate the 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 the residual sequence. When calculating the theoretical absorption spectrum curve, the first concentration result is required as the core input. The standard absorption cross-section curve corresponding to each target gas is retrieved 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. Then, at each wavelength point, the absorption contributions of all gases are weighted and superimposed according to the optical path length to form a complete theoretical absorption spectrum curve, which is used for wavelength-by-wavelength comparison with the actual infrared spectrum.

[0077] 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 to obtain the residual sequence.

[0078] S3-2: The residual sequence is back-projected according to the spatial coordinates of the infrared light path and the detection area to obtain the refractive index update matrix, which is used to characterize the refractive index correction amount of the gas-liquid-gas interface at each spatial position.

[0079] The refractive index update matrix includes refractive index correction values ​​calculated back from the residual signal on each grid cell in the detection area. Each row and column of the refractive index update 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.

[0080] During backprojection, the residual signal at each wavelength point in the residual sequence needs to be traced back along the corresponding infrared optical path, and the residual is allocated to the grid cells of the detection area traversed by the optical path. Specifically, based on the path length of the light in spatial coordinates, the residual signal is proportionally weighted and superimposed onto each grid position, thus forming a refractive index update matrix across the entire detection area. Since the residual reflects the difference between theoretical absorption and actual absorption, and this difference directly stems from the refractive index shift in the optical path, the refractive index update matrix obtained by backprojection can spatially characterize the refractive index correction at each position of the gas-liquid-gas interface.

[0081] S3 also includes:

[0082] S3-3: Input the refractive index update value matrix into the infrared absorption inversion calculation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of infrared light in the detection area. Then, the corrected optical path length is substituted into the conversion formula from transmittance to absorption to recalculate the updated absorption spectrum curve. Subsequently, the updated absorption spectrum curve is fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value, and the new concentration update value sequence is formed according to the sampling time order.

[0083] The formula for converting absorption is: take the negative logarithm of transmittance at each wavelength and multiply it by the actual optical path length to calculate the absorption value corresponding to that wavelength.

[0084] In addition, when fitting the updated absorption spectrum curve to the standard absorption database wavelength by wavelength, it is necessary to compare the actual absorption value of the 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 from 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.

[0085] 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 iterative steps;

[0086] In S3-4, the refractive index update matrix and the concentration update sequence need to be aligned point by point. This includes: first, according to the same sampling time index, each value in the concentration update sequence is matched with each spatial cell of the refractive index update matrix at the same time index; then, the refractive index update value and the concentration update value at that time are stored simultaneously in the grid cell of the refractive index distribution field, so that the two types of data are stored in the same spatial coordinates; finally, the refractive index distribution field with complete one-to-one correspondence is output, which directly provides the input conditions containing the refractive index correction and concentration update for subsequent iterative steps.

[0087] S4 includes:

[0088] S4-1: Use the refractive index update matrix and concentration update sequence as initial inputs, and load the grid coordinate information in the refractive index distribution field to construct the input conditions for the first iteration;

[0089] S4-2: In the first iteration, the optical path length is corrected with the refractive index update matrix, the absorption spectrum curve is corrected with the concentration update sequence, and infrared absorption inversion fitting is performed to output a new refractive index update matrix and a concentration update sequence.

[0090] S4 also includes:

[0091] S4-3: Re-input the newly output refractive index update matrix and concentration update sequence into the iterative process to form a recurring loop, and calculate the residual amplitude after each loop until the residual amplitude is lower than the preset lower limit;

[0092] S4-4: When the residual amplitude reaches the preset lower limit, the refractive index update matrix and concentration update sequence obtained from this iteration are stored in the refractive index distribution field as convergence results, serving 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, while the residual amplitude is the overall quantification result of the residual sequence.

[0093] In S4-1, it should be noted that the refractive index update matrix and the concentration update sequence are used as initial inputs, and the grid coordinate information in the refractive index distribution field is called to make the refractive index update value and the concentration update value correspond to specific spatial locations point by point, thereby constructing the complete input conditions for the first round of iteration.

[0094] In S4-2, it should be noted that: 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. Infrared absorption inversion fitting operation is performed on the corrected optical path and absorption curve to output a new refractive index update value matrix and concentration update value sequence.

[0095] In S4-3, the new refractive index update matrix and concentration update sequence output from the previous round are re-input into the iterative calculation process to form a recurring loop. After each round of iteration, the residual amplitude is calculated wavelength by 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 continues.

[0096] In S4-4: When the residual amplitude is lower than the preset lower limit, the refractive index update value matrix and concentration update value sequence obtained from this round of iteration are stored in the refractive index distribution field as the convergence result, so that the refractive index distribution field retains the converged refractive index and concentration data in both spatial and temporal dimensions, and serves as the input condition for subsequent steps.

[0097] In S5, it includes:

[0098] S5-1: In the output of the reciprocating iterative chain consisting of the refractive index update value matrix and the concentration update value sequence, the concentration update value and the refractive index update value are extracted in the same sampling time order, and the concentration update value and the refractive index update value are corresponding point by point in the grid coordinates of the refractive index distribution field to form a data set containing the sampling time, spatial position, concentration update value and refractive index update value; since the output of the reciprocating iterative chain itself consists of the concentration update value sequence and the refractive index update value matrix, which respectively carry the sampling time index and the spatial coordinate index of the refractive index distribution field, when they are stored point by point in the same coordinate system, a data set containing the sampling time, spatial position, concentration update value and refractive index update value can be formed;

[0099] S5-2: Divide the data set into multiple input sequences according to the continuous sampling time, and perform normalization processing on the concentration update value and refractive index update value in each input sequence to obtain the concentration update value sequence and refractive index update value sequence arranged by time index;

[0100] S5-3: The concentration update value sequence and the refractive index update value sequence are simultaneously input into a time-series modeling algorithm based on a recurrent neural network. In the time-series modeling algorithm, forward calculation is performed at each sampling time to obtain a future concentration prediction value sequence covering a preset time period. The time-series modeling algorithm based on a recurrent neural network is a neural network method that can use the input and hidden state of the previous time to calculate the output of the current time, and is used to process data sequences with time order.

[0101] The reason why forward computation can be performed step by step in the time-series modeling algorithm is that the concentration update value and refractive index update value at each sampling time are input into the network together with the hidden state at the previous time. The network outputs the prediction result at the current time while maintaining time dependence. As the time steps continue to unfold, the network can gradually infer the future numerical distribution based on the accumulated historical information, thereby obtaining a future concentration prediction value sequence covering a preset time period.

[0102] S5-4: The future concentration prediction value sequence is written point by point into the grid cell of the refractive index distribution field according to the sampling time index and spatial location. While writing the future concentration prediction value, the refractive index update value remains unchanged, thereby forming a dataset containing the concentration prediction value at the future sampling time in the refractive index distribution field.

[0103] In S5-4, it should be noted that: for the sequence of future concentration prediction values ​​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 at the spatial position of each grid cell; during the process of writing the future concentration prediction value, the original refractive index update value remains unchanged, so that each grid cell contains both the refractive index update value and the future concentration prediction value, thereby forming a concentration prediction dataset covering future sampling times in the refractive index distribution field;

[0104] S5-5: Use the predicted future concentration value in the dataset as the concentration input condition for the next round of iterative chain, and use the updated refractive index value of the refractive index distribution field at the corresponding sampling time as the refractive index input condition, thereby completing the feedback update of the input condition for the next round of iterative chain.

[0105] In S5-5, it should be noted that for the dataset formed in the refractive index distribution field, the predicted future concentration values ​​at each sampling time should be extracted from it, and these values ​​should be used as the concentration input conditions for the next round of iterative chain. At the same time, the updated refractive index values ​​at the same sampling time 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 iterative chain. Through this synchronous transmission of concentration input and refractive index input, the iterative process can directly utilize the prediction results and existing refractive index data in the new round of calculation, realizing the closed-loop connection of feedback update.

[0106] 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 propagation path of infrared light is corrected by the refractive index distribution field, the transmittance is converted into an absorption spectrum, and the first concentration result is calculated. Then, the result is written into the refractive index distribution field according to the sampling time and spatial position.

[0107] Subsequently, the first concentration result is compared with the gas absorption curve in the standard database to generate the theoretical absorption spectrum, and the residual sequence is obtained by subtracting the actual infrared spectrum wavelength by wavelength. These residuals are then back-projected along the optical path to form the refractive index update matrix, which is used to correct the optical path and refit the concentration update sequence. The refractive index update matrix and the concentration update sequence are used as inputs to the iterative chain. After each round of calculation, the convergence is judged by the residual amplitude. If it does not converge, the iteration continues until it converges, and then the result is written into the refractive index distribution field.

[0108] Simultaneously, the updated refractive index and concentration values ​​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 time 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 location, and are used as input for the next iteration, forming new initial conditions together with the updated refractive index values. Based on this, this scheme realizes closed-loop operation of refractive index calculation, infrared inversion, residual correction, iterative convergence and time prediction in a unified refractive index distribution field.

[0109] The above description is merely 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 within the protection scope of the present invention.

Claims

1. A method for in-situ monitoring of harmful gases in wastewater treatment plant structures, characterized in that, include: S1: In the sludge tank or anaerobic tank, collect the refractive index disturbance data of the local gas-liquid-gas interface caused by the release of bubbles to form a refractive index distribution field that changes with time. S2: Model the infrared light propagation path based on the refractive index distribution field, perform infrared absorption inversion calculation, solve for the first concentration result of the harmful gas, and establish a mapping relationship between the first concentration result and the refractive index distribution field; S3: Calculate the residual between the theoretical spectrum and the actual infrared spectrum corresponding to the first concentration result, and back-project the result of the residual onto the refractive index distribution field to form the refractive index update value; The infrared absorption inversion calculation is corrected based on the refractive index update value to generate the concentration update value; S4: Construct a reciprocating iterative chain between the refractive index update value and the concentration update value, form a convergent result through multiple rounds of iteration, and output the refractive index update value and the concentration update value simultaneously in each round of iteration for input in the next round of iteration; S5: Simultaneously input the concentration update value and refractive index update value output by the reciprocating iterative chain into the time-series modeling algorithm to generate a preset time release prediction result for harmful gases, and feed the prediction result back to the refractive index distribution field to update the input conditions for the next iteration.

2. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 1, characterized in that: S1 includes: S1-1: Arrange the incident beam and the reference beam along the bubble release area in the sludge tank or anaerobic tank, collect the fringe image of optical interference, and extract the phase difference value sequence from the fringe image; S1-2: Expand the phase difference value sequence into a continuous phase curve according to the sampling time sequence, calculate the phase difference components between adjacent sampling points, and convert the phase difference components into a refractive index increment sequence according to the preset refractive index conversion coefficient. S1-3: The refractive index increment sequence is matched point by point with the spatial coordinates of the detection area to calculate the refractive index perturbation matrix, and the refractive index perturbation distribution covering the entire gas-liquid-gas interface is formed by matrix interpolation. S1-4: Perform an accumulation operation on the refractive index perturbation matrix according to the time index to generate a refractive index distribution field that varies with time.

3. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 2, characterized in that: 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. The absorption spectrum curve serves as the input data for concentration inversion. S2-3: Compare the absorption spectrum curve with the pre-established standard absorption database of harmful gases wavelength by wavelength, adjust the concentration coefficient of each gas using the least squares fitting method, so that the residual between the fitted curve and the actual absorption spectrum curve reaches the preset lower limit, and output the first concentration result of the target harmful gas. S2-4: The first concentration result is stored in the grid cell of the refractive index distribution field according to the point-to-point correspondence between the sampling time and the spatial position of the infrared light path. The refractive index data and concentration data are maintained in the distribution field at the same time, thereby establishing a one-to-one mapping relationship between the two.

4. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 3, characterized in that: S3 includes: S3-1: Calculate the 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 the residual sequence; S3-2: The residual sequence is back-projected according to the spatial coordinates of the infrared light path and the detection area to obtain the refractive index update matrix, which is used to characterize the refractive index correction amount of the gas-liquid-gas interface at each spatial position. The refractive index update matrix includes refractive index correction values ​​calculated back from the residual signals on each grid cell in the detection area. Each row and column of the refractive index update 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.

5. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 4, characterized in that: S3 also includes: S3-3: Input the refractive index update value matrix into the infrared absorption inversion calculation: First, the refractive index update value matrix is ​​used to correct the actual optical path length of infrared light in the detection area. Then, the corrected optical path length is substituted into the conversion formula from transmittance to absorption to recalculate the updated absorption spectrum curve. Subsequently, the updated absorption spectrum curve is fitted to the standard absorption database wavelength by wavelength to calculate the new concentration value, and the new concentration update value sequence is formed according to the sampling time order. 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 iterative steps.

6. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 5, characterized in that: S4 includes: S4-1: Use the refractive index update matrix and concentration update sequence as initial inputs, and load the grid coordinate information in the refractive index distribution field to construct the input conditions for the first iteration; S4-2: In the first iteration, the optical path length is corrected with the refractive index update matrix, the absorption spectrum curve is corrected with the concentration update sequence, and infrared absorption inversion fitting is performed to output a new refractive index update matrix and a concentration update sequence.

7. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 6, characterized in that: S4 also includes: S4-3: Re-input the newly output refractive index update matrix and concentration update sequence into the iterative process to form a recurring loop, and calculate the residual amplitude after each loop until the residual amplitude is lower than the preset lower limit; S4-4: When the residual amplitude reaches the preset lower limit, the refractive index update matrix and concentration update sequence obtained from this round of iteration are stored in the refractive index distribution field as convergence results, and used as input conditions for subsequent steps.

8. The in-situ hazardous gas monitoring method for sewage treatment plant structures according to claim 7, characterized in that: In S5, it includes: S5-1: In the output of the iterative chain consisting of the refractive index update value matrix and the concentration update value sequence, the concentration update value and the refractive index update value are extracted in the same sampling time order, and the concentration update value and the refractive index update value are matched point by point in the grid coordinates of the refractive index distribution field to form a data set containing the sampling time, spatial position, concentration update value and refractive index update value. S5-2: Divide the data set into multiple input sequences according to the continuous sampling time, and perform normalization processing on the concentration update value and refractive index update value in each input sequence to obtain the concentration update value sequence and refractive index update value sequence arranged by time index; S5-3: Input the concentration update value sequence and the refractive index update value sequence into the time-series modeling algorithm based on the recurrent neural network at the same time. Perform forward calculation at each sampling time in the time-series modeling algorithm to obtain the future concentration prediction value sequence covering the preset time. S5-4: The future concentration prediction value sequence is written point by point into the grid cell of the refractive index distribution field according to the sampling time index and spatial location. While writing the future concentration prediction value, the refractive index update value remains unchanged, thereby forming a dataset containing the concentration prediction value at the future sampling time in the refractive index distribution field. S5-5: Use the predicted future concentration values ​​in the dataset as the concentration input conditions for the next round of iterative chain, and use the updated refractive index values ​​of the refractive index distribution field at the corresponding sampling time as the refractive index input conditions, thereby completing the feedback update of the input conditions for the next round of iterative chain.

Images