A rapid start-up method for anaerobic ammonium oxidation based on dynamic adjustment of nitrogen oxide concentration.

By dynamically adjusting the combined effects of nitrogen oxide concentration and nitrate, and utilizing a sequencing batch reactor and control algorithm, the problem of long start-up time in the anaerobic ammonia oxidation process was solved, enabling rapid start-up and stable operation, and avoiding chemical pollution and resource constraints.

CN123062074APending Publication Date: 2026-11-10CHONGQING UNIV OF ARTS & SCI
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611589930.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-14
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing anaerobic ammonia oxidation processes have long start-up times. Existing technologies, such as adding exogenous promoters, adding biological carriers, or inoculating mature sludge, have problems such as high cost, chemical pollution, or difficulty in obtaining them. Moreover, existing methods have not effectively solved the problems of endogenous organic matter release, insufficient nitrite supply, and fixed and singular influent nitrogen oxide ratio in the early stage of start-up.

Method used

By dynamically adjusting the nitrogen oxide concentration and employing a sequencing batch reactor, the dual role of nitrate is utilized: denitrification generates nitrite to supply anaerobic ammonia-oxidizing bacteria and consumes endogenous COD. Combined with extended Kalman filtering and fuzzy fusion to determine the timing of nitrate-to-nitrite replacement, rapid start-up is achieved.

Benefits of technology

It shortens the start-up time of the anaerobic ammonia oxidation reactor, avoids chemical pollution and resource limitations, enables rapid reproduction and stable operation of anaerobic ammonia oxidizing bacteria, and breaks through the bottleneck of the inability of the fixed nitrogen oxide ratio to match the microbial enrichment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123062074A_ABST
    Figure CN123062074A_ABST
Patent Text Reader

Abstract

This invention provides a rapid start-up method for an anaerobic ammonia oxidation process based on dynamically adjusting nitrogen oxide concentration. Utilizing a sequencing batch reactor (SBR), the specific steps include: inoculation and establishment of anaerobic operating conditions – pre-start-up – replacement timing determination – replacement of nitrate with nitrite – completion of start-up. This method accelerates the start-up of the anaerobic ammonia oxidation reactor through the dual effects of nitrate, effectively solving problems in existing anaerobic ammonia oxidation start-up processes such as the release of endogenous organic matter in the early start-up phase, insufficient nitrite supply in the later phase, and the inability of the fixed influent nitrogen oxide composition to adapt to the start-up process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of biological nitrogen removal technology for wastewater, specifically to a rapid start-up method for anaerobic ammonia oxidation based on dynamically adjusting the concentration of nitrogen oxides. Background Technology

[0002] Anammox is considered a core technology for next-generation biological nitrogen removal from wastewater due to its advantages such as no need for aeration, no need for external organic carbon sources, and low sludge production. However, anammox bacteria (AnAOB) grow extremely slowly, with a doubling time of several days to more than ten days, resulting in a long reactor start-up cycle (usually more than 180 days), which seriously restricts the engineering application of this technology.

[0003] Currently, existing technologies for shortening the start-up time of anammox mainly include adding exogenous promoters, adding biological carriers, and inoculating mature anammox sludge. Adding exogenous promoters, such as Fe... 2+ Chemicals such as reduced graphene oxide and FeS can be used to stimulate the metabolic activity of anaerobic ammonia oxidizing bacteria, but adding exogenous promoters has problems such as high cost and the chemical substances can easily cause secondary pollution. Adding biological carriers, such as polyurethane sponge, volcanic rock, and granular activated carbon, can provide a surface for anaerobic ammonia oxidizing bacteria to attach and grow, but the addition of carriers can only shorten the start-up time to a limited extent (by 17% to 28%) and the effect is not significant. Inoculating mature anaerobic ammonia oxidation sludge can effectively shorten the start-up time, but the number of wastewater treatment plants with mature anaerobic ammonia oxidation sludge worldwide is limited (mainly concentrated in Europe), making it difficult to obtain a large amount of mature seed sludge.

[0004] Furthermore, Chinese patent document CN117285152A discloses an anaerobic ammonia oxidation device and its start-up method, which provides an anaerobic ammonia oxidation device that can be started up quickly, has high denitrification efficiency, good denitrification effect, and can operate stably for a long time. However, this patent focuses on using packing material to attach a biofilm to reduce biomass outflow, rather than solving the problem from the perspective of the enrichment nature and mechanism of anaerobic ammonia oxidizing bacteria. Chinese patent document CN104016478A discloses a method for rapidly starting up an anaerobic ammonia oxidation reactor, which accelerates the start-up of the anaerobic ammonia oxidation reactor by adjusting the concentration of ferrous iron to increase the activity of key enzymes related to the metabolism of anaerobic ammonia oxidizing bacteria. However, this method does not take into account the problem that ferrous iron is easily oxidized to ferric iron in the presence of dissolved oxygen, thereby inhibiting the absorption of iron by anaerobic ammonia oxidizing bacteria. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention aims to provide a rapid start-up method for anaerobic ammonia oxidation based on dynamically adjusting the concentration of nitrogen oxides. This method utilizes nitrate nitrogen (NO3) -The nitrate nitrogen (NO3-N) has a dual function: it consumes the organic matter (COD) released during the anaerobic adaptation period of activated sludge to eliminate its inhibitory effect on anaerobic ammonia-oxidizing bacteria, and it removes nitrate nitrogen (NO3-N) through partial denitrification. - -N) is converted into nitrite nitrogen (NO2) - The role of nitrogen oxides (NOx) in anaerobic ammonia oxidation (ANAO) reactors is to accelerate the start-up process, effectively addressing the challenges of insufficient nitrite nitrogen supply, the release of endogenous organic matter, and the fixed and singular proportion of influent nitrogen oxides (NOx) during the initial start-up phase of existing ANAO processes. x - This can lead to problems such as inability to adapt to the startup process.

[0006] The objective of this invention is achieved through the following technical solution: A rapid start-up method for anaerobic ammonia oxidation based on dynamically adjusted nitrogen oxide concentration, using a sequencing batch reactor, includes the following steps: Step S1, Inoculation and Establishment of Anaerobic Operating Conditions: Inoculate the sludge into the sequencing batch reactor and simultaneously introduce wastewater, and initialize the model; Step S2, Initial Start-up (Nitrate Dominance: The influent in the initial startup phase contains both ammonia nitrogen and nitrogen oxides, with nitrate being the dominant nitrogen oxide): Perform three-nitrogen soft measurement, extended Kalman filter (EKF) state estimation, and cumulative nitrogen balance verification in sequence. In the initial stage of startup, under anaerobic conditions, the inoculated sludge decays and hydrolysis releases endogenous COD; denitrifying bacteria utilize endogenous COD to reduce some of the nitrates in the influent to nitrites, and the nitrites produced are used by anaerobic ammonia oxidizing bacteria to produce nitrogen gas; Step S3, Replacement Timing Determination: Based on three criteria—significant ammonia nitrogen removal, increased nitrate residue, and no nitrite accumulation—as well as circumstantial evidence of anammox bacterial growth, the replacement timing is determined through fuzzy fusion, i.e., the replacement trigger flag is output. Step S4: Replace nitrate with nitrite: After the replacement trigger flag is reached, the proportion of candidate nitrites is screened through outer Gaussian process-Bayes optimization, and the concentration of nitrite replacing nitrate is automatically adjusted through inner model prediction control. Step S5, Startup Complete: The startup is determined to be complete, and the PLC switches to the normal Anammox operating mode; and increases the nitrogen volumetric loading rate (NLR) or shortens the hydraulic retention time (HRT) according to the target wastewater quality gradient. Each step is still controlled by model prediction for safe tracking.

[0007] Based on further optimization of the above scheme, the sequencing batch reactor is equipped with a gas collection port, a water inlet, a water outlet, and an aeration port (in the early stage of start-up, nitrogen is purged to ensure an anaerobic environment). One end of the gas collection port is connected to a gas collection bag, one end of the water inlet is connected to a first peristaltic pump, and the other end of the first peristaltic pump is connected to an external water inlet tank. The timed opening and closing of the first peristaltic pump is controlled by a timer. The water outlet is connected to an external water outlet tank through a second peristaltic pump (the timed opening and closing of the second peristaltic pump is controlled by a timer). The sequencing batch reactor is placed on a shaking table. A UV-Vis full-spectrum probe is installed in the middle of the inner sidewall of the batch reactor, and the UV-Vis full-spectrum probe is located at 1 / 2 underwater level. The pH electrode, ORP electrode, DO electrode, temperature electrode, and EC conductivity electrode are all in the same protective sleeve as the UV-Vis full-spectrum probe. An inlet flow meter is installed on the inlet pipe between the first peristaltic pump and the batch reactor, and an outlet flow meter is installed between the second peristaltic pump and the outlet tank. A level gauge is installed on the top of the batch reactor, and a wet gas flow meter is installed on the pipe between the gas collection port and the gas collection bag.

[0008] Based on further optimization of the above scheme, in step S1, the inoculated sludge is a mixture of two or more of the following: ordinary activated sludge, anaerobic granular sludge, denitrifying sludge, and municipal activated sludge; preferably, a mixture of anaerobic granular sludge and municipal activated sludge, with a mass ratio of 1:1, and an inoculation concentration of volatile suspended solids (VSS) of 2-4 g / L.

[0009] Based on further optimization of the above scheme, in step S1, the ammonia nitrogen concentration of the low ammonia nitrogen wastewater... S NH4,in The nitrogen oxide concentration is 15–25 mg N / L. S NO3,in The concentration is 20–30 mg N / L; simultaneously, the proportion of nitrite nitrogen to total nitrogen oxides in the influent during the initial startup phase is... , S NO2,in This indicates the concentration of nitrite in the influent.

[0010] Based on further optimization of the above scheme, in step S1, after the sludge is fed into the sequencing batch reactor, it is carried out at 32±1℃ and shaking speed of 80~120 rpm; each hydraulic retention cycle is 15 min for water inlet → 2~4 h for reaction → 30 min for settling → 10 min for drainage, with a water exchange ratio of 50%.

[0011] Based on further optimization of the above scheme, in step S1, the model initialization specifically involves: the partial regression coefficients of soft-sensor partial least squares regression (PLSR). process quantity regression coefficients , , j Traversing pH, ORP, T, DO, EC, Q in , Q out , L y ; Initial state values ​​for the Extended Kalman Filter (EKF): ,in, This indicates the initial concentration of anammox bacteria. This indicates the initial concentration of some denitrifying bacteria. This indicates the initial endogenous biodegradable COD concentration in the reactor. These represent the initial ammonia nitrogen concentration, initial nitrite nitrogen concentration, and initial nitrate nitrogen concentration of the reactor, respectively. EKF initial state covariance matrix Pix 0.

[0012] Based on further optimization of the above scheme, in step S2, the trinitrogen soft measurement specifically involves: Vector sampled every 5 minutes: ; In the formula: This represents the absorbance at wavelength λ. Soft measurement consists of two layers, specifically: The first layer is the Partial Least Squares Regression (PLSR) layer: ; ; In the formula: This represents the intercept of the i-th nitrogen regression equation. Represents the process quantum vector, express The j-th component; The second layer is for Long Short-Term Memory (LSTM) residual correction, with a 12-step input window: ; Final soft measurement output: .

[0013] Based on further optimization of the above scheme, in step S2, the extended Kalman filter (EKF) state estimation is specifically as follows: State vector The state transition is discretized using a continuous equation: ; In the formula: This indicates the maximum specific growth rate of anammox. This represents the anammox half-saturation constant for nitrite. This represents the anammox ammonia nitrogen half-saturation constant. This represents the nitrite inhibition constant. Indicates the rate of endogenous death in anammox bacteria. This indicates the maximum specific growth rate of denitrifying bacteria. This represents the half-saturation constant of denitrified nitrates. This represents the denitrification COD half-saturation constant. Indicates the rate of denitrifying bacteria death. Indicates the COD-to-consumption rate. Indicates the carbon release rate during sludge decay. The concentration of dead / decaying sludge is indicated by Q, and Q and V represent the average flow rate and effective volume of the reactor, respectively. This represents the anammox cell yield coefficient. Indicates the denitrifying bacteria yield coefficient; This represents the anammox temperature correction factor. This represents the temperature correction factor for denitrifying bacteria, where R represents the gas constant and T represents the reactor temperature. E A This indicates the anammox activation energy. E D Indicates the activation energy of heterotrophic denitrification; Observation vector: , These represent the soft-measured values ​​of ammonia nitrogen, nitrite nitrogen, and nitrate nitrogen in the effluent, respectively. These represent the measured redox potential and pH value of the reactor, respectively; measured redox potential The regression relationship with trinitrogen is as follows: , Indicates the regression intercept. These represent the regression coefficients for the nitrate and nitrite terms, respectively. EKF outputs the state estimate at step t every 5 minutes. and the state covariance matrix Pix t .

[0014] Based on further optimization of the above scheme, in step S2, the cumulative nitrogen balance verification specifically involves: performing periodic water balance and nitrogen balance checks. ; In the formula: This represents the average gas production volume flow rate during the reactor cycle. Represents the standard molar volume. Indicates the conversion factor; like The continuous positive value and the increase in gas production measured by the wet gas flow meter prove that anammox has been activated, and this is written into step S3 fuzzy fusion. Output a set every 5 minutes After running continuously for 2 to 5 hydraulic residence cycles, proceed to step S3.

[0015] Based on further optimization of the above scheme, step S3, based on the net ammonia nitrogen removal significance (CUSUM), specifically includes: Taking the first three hydraulic residence cycles as the controlled baseline, calculate the baseline standard deviation: ; In the formula: std() represents the standard deviation function. Indicates the net removal rate of ammonia nitrogen; The single-sided CUSUM is (initial value) (0): ; In the formula: Indicates the baseline mean. Indicates the CUSUM tolerance parameter; The membership degree of the first criterion is finally obtained: ; In the formula: This represents the CUSUM decision threshold.

[0016] Based on further optimization of the above scheme, in step S3, the increase in nitrate residue (BOCPD) specifically refers to: The residual ratio is: ; The residual ratio sequence was analyzed to detect the upward trend of nitrate residue, and the predicted distribution was Student-t. The hazard constant was 1 / 20, and the posterior length of each maintenance operation step was... (Initial value) (1), posterior probability at the change point: ; Use the segment after the point change to calculate the least squares slope To obtain the membership degree of the second criterion: .

[0017] Based on further optimization of the above scheme, in step S3, the non-accumulation of nitrite (FNA constraint) specifically involves obtaining a third criterion based on the free nitrite concentration FNA(t): ; ; In the formula: This indicates that nitrous acid at temperature T The acid dissociation constant under the given conditions.

[0018] Based on further optimization of the above scheme, in step S3, the supporting evidence for the growth of anammox bacteria specifically includes: ; Require This means that EKF estimates the amount of anammox bacteria to have increased by more than 20% compared to the time of inoculation.

[0019] Based on further optimization of the above scheme, in step S3, fuzzy fusion specifically involves: First of all, nearly seven days , , The extreme values ​​of the three indicators were normalized to... Then normalize by column. ;No. j Term entropy value: ; Weighting coefficients: ; In the formula: N This represents the total number of samples collected over seven days. p ij Indicates the first i The sample at the th j The normalized value of the indicator; the three terms of k correspond to respectively , , ; Overall criterion score: ; In the formula: These represent the normalized values ​​of the corresponding membership degrees; Preset replacement trigger threshold D ,like D(t) > D And after 2 to 5 consecutive hydraulic residence cycles, the trigger flag is replaced. Flag swap =1, proceed to step S4 and begin the replacement; If D(t) <D If the condition persists for more than seven days, the system will issue an alarm and recommend manual verification of the inoculated sludge or a reduction in the nitrogen volumetric loading rate (NLR).

[0020] Based on further optimization of the above scheme, in step S4, the specific proportion of candidate nitrites screened by the outer Gaussian process-Bayesian optimization is as follows: by For input, This indicates the proportion of nitrite nitrogen to total nitrogen oxides in the influent. This indicates the number of days the reactor has been in operation since inoculation was completed; in... For the output, construct a Matérn 5 / 2+ linear kernel Gaussian process (GP): ; In the formula: This represents the variance of the Mattérn kernel signal. Represents linear kernel variance. Represents the length scale of the d-th dimension ( All are hyperparameters, and marginal likelihood is used (retrained every 24 hours). dis represents Mahalanobis distance. Simultaneously, for each of the three safety constraints—nitrite effluent constraint g1, FNA inhibition constraint g2, and anammox bacteria non-retreat constraint g3—an independent Gaussian process classifier is constructed to output the corresponding constraint safety probability. : ; In the formula: Indicates the safe upper limit for nitrite nitrogen in the effluent; exist Improve point selection by applying constraints on the mesh: ; In the formula: Indicates the desired improvement in constraints; , Representing the standard normal CDF and PDF respectively; , Let represent the posterior mean and standard deviation of the Gaussian process, respectively. This represents the current best observation value; Final output of candidate nitrite percentage .

[0021] Based on further optimization of the above scheme, in step S4, the inner model predictive control (MPC) performs tracking as follows: Will As the MPC reference trajectory, and at the end of each hydraulic residence cycle, a rolling optimization of Np=3 steps is solved: ; In the formula: This indicates a reference value for ammonia nitrogen removal rate. This represents the proportional change at step k+i (i.e. ); These represent the corresponding weight coefficients; constraint: ; The prediction model uses the biochemical equation of the EKF state estimation in step S2 to linearize at the current operating point, solves it using quadratic programming, and recalculates it every 5 minutes.

[0022] Based on further optimization of the above scheme, in step S4, the automatic adjustment of the concentration of nitrite replacing nitrate specifically involves: PLC is obtained according to MPC. Adjust the speed of the two inlet pumps for both water and medicine: ; In the formula: This indicates the target concentration of total nitrogen oxides in the influent (which remains constant). These represent the concentrations of the two stock solutions, respectively. New observations will be made after each hydraulic residence cycle. Add to the training set for near-Gaussian processes; If at the end of two consecutive hydraulic residence cycles If it rises or FNA > 0.08, then Back to , ; like Decrease in two consecutive hydraulic residence cycles or If the decrease is greater than 10%, then Back to And called the police. ; If the Gaussian process predicts 95% confidence half-width The model predictive control is performed with a conservative step size of 10%. A set is issued for each hydraulic residence cycle. The training set for Gaussian processes continues to grow; when When step S1 is achieved and the nitrogen content in the effluent is continuously and stably stable, proceed to step S5.

[0023] Based on further optimization of the above scheme, in step S5, the nitrogen volumetric loading rate (NLR) is increased or the hydraulic retention time (HRT) is shortened in a gradient of 10% to 20% according to the target wastewater quality, and each step is still controlled by model prediction for safe tracking. Finally, the entire startup trajectory was packaged and archived as a priori for the Gaussian process of the next batch of reactors, forming a self-evolving closed loop.

[0024] The following are the technical effects of the present invention: This invention utilizes sludge to start up a low-ammonia nitrogen anaerobic ammonium oxidation reactor (ANAO) without relying on mature ANAO seed sludge. In the initial start-up phase, nitrate-dominant nitrogen oxides are used as influent, without the addition of external organic carbon sources. The "dual role" of nitrate is employed to address the low-ammonia nitrogen start-up: Initial nitrate addition serves two purposes: firstly, it is reduced to nitrite by denitrifying bacteria using endogenous COD, addressing the later NO2 supply shortage; secondly, it consumes the endogenous COD released by sludge hydrolysis, eliminating the inhibition of ANAO bacteria by organic matter. Simultaneously, a fuzzy fusion dynamic determination of the nitrate-nitrite replacement timing is achieved through three criteria and one supporting evidence. Compared to existing fixed-time, fixed-concentration switching methods, the net ammonia nitrogen removal significance (CUSUM) demonstrates that AnAO… The activation of B and the increase in nitrate residue (BOCPD turning point) prove that denitrification is saturated (nitrate surplus). FNA constraint proves that nitrite is consumed by AnAOB once it is generated. Coupled with the circumstantial evidence of AnAOB biomass growth, this avoids the problems of insufficient nitrite due to switching too early and waste of nitrate and depletion of nitrite due to switching too late. The switching time is precisely controlled, so that the influent becomes nitrite-dominant or contains only nitrite, which is directly supplied to anaerobic ammonia oxidizing bacteria. Through the synergy of rolling tracking between the outer GP-BO and the inner MPC, not only is safe exploration achieved, but the biochemical model can also be linearized according to the current working point, breaking through the bottleneck of "fixed influent nitrogen oxide composition cannot match the microbial enrichment process".

[0025] Compared with a fixed ratio of nitrite and nitrate, this invention arranges the early and late effects of nitrate at different start-up stages, taking into account both the start-up speed and the later anaerobic ammonia oxidation activity. This avoids insufficient nitrite supply caused by the reduction of nitrate's partial reduction capacity and is conducive to the continued proliferation of anaerobic ammonia oxidizing bacteria. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of a sequencing batch reactor in an embodiment of the present invention.

[0027] Figure 2 This is a flowchart of the rapid start-up method for anaerobic ammonia oxidation in an embodiment of the present invention.

[0028] Figure 3 This is a schematic diagram of the rapid start-up method for anaerobic ammonia oxidation in an embodiment of the present invention.

[0029] The components include: 1. Sequencing batch reactor; 2. Gas collection port; 3. Water inlet; 4. Water outlet; 5. Aeration port; 6. Gas collection bag; 7. First peristaltic pump; 8. Water inlet tank; 9. Second peristaltic pump; 10. Water outlet tank; and 11. Shaking table. Detailed Implementation

[0030] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0031] Example 1: A rapid start-up method for anaerobic ammonia oxidation based on dynamically adjusted nitrogen oxide concentration, using a sequencing batch reactor, includes the following steps: Step S1, Inoculation and Establishment of Anaerobic Operating Conditions: Inoculate the sludge into the sequencing batch reactor and simultaneously introduce wastewater, and initialize the model; Reference Figure 1 As shown: The sequencing batch reactor 1 is equipped with a gas collection port 2, a water inlet 3, a water outlet 4, and an aeration port 5 (in the early stage of start-up, nitrogen is purged to ensure an anaerobic environment). One end of the gas collection port 2 is connected to a gas collection bag 6. One end of the water inlet 3 is connected to a first peristaltic pump 7, and the other end of the first peristaltic pump 7 is connected to an external water inlet tank 8. The timed opening and closing of the first peristaltic pump 7 is controlled by a timer (at the same time, the water inlet 3 can be connected to a nitrite storage tank and a nitrate storage tank. The peristaltic pumps of the nitrite storage tank and the nitrate storage tank are independently adjusted by a PLC according to the MPC output signal, thereby dynamically adjusting the nitrite ratio). The water outlet 4 is connected to an external water outlet tank 10 through a second peristaltic pump 9 (the timed opening and closing of the second peristaltic pump 9 is controlled by a timer). The sequencing batch reactor 1 is placed on a shaking table 11. A UV-Vis full-spectrum probe (200~720) is installed in the middle of the inner side wall of the sequencing batch reactor 1. The UV-Vis full-spectrum probe is located at half the water level (full absorption spectral vector, soft measurement inversion of three nitrogens). The pH electrode, ORP electrode, DO electrode, temperature electrode, and EC conductivity electrode are all in the same protective sleeve (i.e., all located at half the water level, used to test pH, ORP, DO, temperature T, and EC conductivity, respectively). An inlet flow meter is installed on the inlet pipe between the first peristaltic pump 7 and the sequencing batch reactor 1 (used to test the inlet flow rate). Q in A flow meter is installed between the second peristaltic pump 9 and the outlet tank 10 (for testing the outlet flow rate). Q out A level gauge (200mm above the liquid surface) is installed at the top of the sequencing batch reactor 1 to test the real-time liquid level. L yTo determine the stages of water inlet, reaction, sedimentation and drainage of the sequencing batch reactor, a wet gas flow meter (used to test the volumetric flow rate of the produced gas) is installed on the pipeline between the gas collecting port 2 and the gas collecting bag 6.

[0032] The inoculated sludge is a mixture of anaerobic granular sludge and municipal activated sludge in a mass ratio of 1:1, and the inoculation concentration of volatile suspended solids (VSS) is 2-4 g / L.

[0033] ammonia nitrogen concentration in low ammonia nitrogen wastewater S NH4,in The nitrogen oxide concentration is 15–25 mg N / L. S NO3,in The concentration is 20–30 mgN / L; simultaneously, the proportion of nitrite nitrogen to total nitrogen oxides in the initial influent during startup is... , S NO2,in This indicates the concentration of nitrite in the influent.

[0034] After the sludge is fed into the sequencing batch reactor, it is carried out at 32±1℃ and shaking speed of 80~120 rpm; each hydraulic retention cycle is 15 min for influent → 2~4 h for reaction → 30 min for settling → 10 min for effluent, with a water exchange ratio of 50%.

[0035] The model initialization specifically involves the partial regression coefficients of soft-sensor partial least squares regression (PLSR). (Corresponding wavelength λ, i-th type of nitrogen), regression coefficients of process quantities (Corresponding to the j-th process quantity and the i-th type of nitrogen). , j Traversing pH, ORP, T, DO, EC, Q in , Q out , L y (Initial values ​​for offline training, obtained from historical data using partial least squares regression); Initial state values ​​for the Extended Kalman Filter (EKF): ,in, This indicates the initial concentration of anammox bacteria (usually 0.02 g VSS / L). This indicates the initial concentration of denitrifying bacteria (typically 0.15 g VSS / L). This indicates the initial endogenous biodegradable COD concentration in the reactor (typically 150 mg COD / L). These represent the initial ammonia nitrogen concentration, initial nitrite nitrogen concentration, and initial nitrate nitrogen concentration of the reactor, respectively. EKF initial state covariance matrix Pix 0 (Diagonal elements take 25% of the initial value of the corresponding state, i.e., standard deviation = 50% of the initial value).

[0036] Step S2, Early Start-up (Nitrate Dominance: The influent in the early start-up period contains both ammonia nitrogen and nitrogen oxides, with nitrate nitrogen being the dominant nitrogen oxide): Perform three-nitrogen soft measurement, extended Kalman filter (EKF) state estimation, and cumulative nitrogen balance verification in sequence. In the initial stage of startup, under anaerobic conditions, the inoculated sludge decays and hydrolysis releases endogenous COD; denitrifying bacteria utilize endogenous COD to reduce some of the nitrates in the influent to nitrites, and the nitrites produced are used by anaerobic ammonia oxidizing bacteria to produce nitrogen gas; The specific details of the trinitrogen soft measurement are as follows: Vector sampled every 5 minutes: ; In the formula: This represents the absorbance at wavelength λ (obtained from a UV-Vis full-spectrum probe, including eight characteristic wavelengths: 220nm, 227nm, 254nm, 275nm, 350nm, 400nm, 500nm, and 700nm). Soft measurement consists of two layers, specifically: The first layer is the Partial Least Squares Regression (PLSR) layer: ; ; In the formula: The intercept of the i-th nitrogen regression equation is represented (obtained through PLSR calibration). Represents the process quantum vector, express The j-th component; The second layer is for Long Short-Term Memory (LSTM) residual correction, with a 12-step input window: ; Final soft measurement output: .

[0037] The state estimation of the Extended Kalman Filter (EKF) is as follows: State vector The state transition is discretized using continuous equations (step size 5 min): ; In the formula: This indicates the maximum specific growth rate of anammox (typically 0.08–0.12 d). -1 ), This represents the anammox nitrite half-saturation constant (typically 0.1–0.5 mg N / L). This represents the anammox ammonia nitrogen half-saturation constant (typically 0.05–0.2 mg N / L). This represents the nitrite inhibition constant (typically 0.5 mg HNO2-N / L). This indicates the rate of endogenous death of anammox bacteria (typically 0.005–0.01 d). -1 ), This indicates the maximum specific growth rate of denitrifying bacteria (generally 0.2–0.5 d). -1 ), This represents the half-saturation constant of denitrified nitrate (typically 1–5 mg N / L). This represents the denitrification COD half-saturation constant (typically 10–30 mg COD / L). This indicates the rate of denitrifying bacteria death (generally 0.02–0.05 days). -1 ), This indicates the COD specific consumption rate (typically 0.5–1 g COD / g VSS·d). This indicates the carbon release rate during sludge decay (typically 0.05–0.1 days). -1 ), Indicates the concentration of dead / decaying sludge (based on the total volatile suspended solids concentration of the inoculated sludge). X seed - X A - X D (Obtained), Q and V represent the average flow rate and effective volume of the reactor, respectively. This represents the anammox cell yield coefficient (typically 0.066–0.1 g VSS / g N). This represents the denitrifying bacteria yield coefficient (typically 0.4 g VSS / g COD). This represents the anammox temperature correction factor. The value represents the temperature correction factor for denitrifying bacteria, R represents the gas constant (typically 8.314 J / (mol·K)), and T represents the reactor temperature. E A This indicates the anammox activation energy (typically 70 kJ / mol). ED This indicates the activation energy of heterotrophic denitrification (generally 40–60 kJ / mol). Observation vector: , These represent the soft-measurement values ​​of ammonia nitrogen, nitrite nitrogen, and nitrate nitrogen in the effluent, respectively (output via PLSR+LSTM soft measurement). These represent the measured redox potential and pH value of the reactor, respectively; measured redox potential The regression relationship with trinitrogen is as follows: , This represents the regression intercept (offline calibration, typically -150 to -250 mV). These represent the regression coefficients for nitrate and nitrite terms, respectively (offline calibration, generally). ); EKF outputs the state estimate at step t every 5 minutes. and the state covariance matrix Pix t .

[0038] The specific evidence for cumulative nitrogen balance is as follows: Periodic checks of water and nitrogen balances are performed. ; In the formula: This represents the average gas production volume flow rate during the reactor cycle (measured by a wet gas flow meter). This indicates the standard molar volume (typically 24.5 L / mol). This represents the conversion factor (typically 14000 mg N / mol). like The continuous positive value and the increase in gas production measured by the wet gas flow meter prove that anammox has been activated, and this is written into step S3 fuzzy fusion. Output a set every 5 minutes After running continuously for 2 to 5 hydraulic residence cycles, proceed to step S3.

[0039] Step S3, Replacement Timing Determination: Based on three criteria—significant ammonia nitrogen removal, increased nitrate residue, and no nitrite accumulation—as well as circumstantial evidence of anammox bacterial growth, the replacement timing is determined through fuzzy fusion, i.e., the replacement trigger flag is output. The significance of net ammonia nitrogen removal (CUSUM) is specifically as follows: Taking the first three hydraulic residence cycles as the controlled baseline, calculate the baseline standard deviation: ; In the formula: std() represents the standard deviation function. Indicates the net removal rate of ammonia nitrogen; The single-sided CUSUM is (initial value) (0): ; In the formula: This represents the baseline mean (usually 0). This represents the CUSUM tolerance parameter (generally...). ); The membership degree of the first criterion is finally obtained: ; In the formula: This represents the CUSUM decision threshold (generally 1). ).

[0040] The specific manifestations of increased nitrate residue (BOCPD) are: The residual ratio is: ; The residual ratio sequence was analyzed to detect the upward trend of nitrate residue, and the predicted distribution was Student-t. The hazard constant was 1 / 20, and the posterior length of each maintenance operation step was... (Initial value) (1), posterior probability at the change point: ; Use the segment after the point change to calculate the least squares slope To obtain the membership degree of the second criterion: .

[0041] The non-accumulation of nitrite (FNA constraint) is specifically defined as follows: a third criterion is obtained based on the free nitrite concentration FNA(t). ; ; In the formula: This indicates that nitrous acid at temperature T The acid dissociation constant under the given conditions.

[0042] Based on further optimization of the above scheme, in step S3, the supporting evidence for the growth of anammox bacteria specifically includes: ; Require This means that EKF estimates the amount of anammox bacteria to have increased by more than 20% compared to the time of inoculation.

[0043] Fuzzy fusion specifically refers to: First of all, nearly seven days , , The extreme values ​​of the three indicators were normalized to... Then normalize by column. ;No. j Term entropy value: ; Weighting coefficients: ; In the formula: N This represents the total number of samples collected over seven days. p ij Indicates the first i The sample at the th j The normalized value of the indicator; the three terms of k correspond to respectively , , ; Overall criterion score: ; In the formula: These represent the normalized values ​​of the corresponding membership degrees (nearly seven days). The membership degrees of the three columns are normalized to [0,1]); Preset replacement trigger threshold D (Usually 0.6), if D(t) > D And for 2 to 5 consecutive hydraulic residence cycles (the hydraulic residence cycle is determined by...) D(t) The limit is determined, for example: D(t) >0.8, using 2 hydraulic residence cycles, D(t) If the value is exactly greater than 0.6, then use 5 hydraulic residence cycles to replace the trigger flag. Flag swap =1, proceed to step S4 and begin the replacement; If D(t) < D If the condition persists for more than seven days, the system will issue an alarm and recommend manual verification of the inoculated sludge or a reduction in the nitrogen volumetric loading rate (NLR).

[0044] Step S4: Replace nitrate with nitrite: After the replacement trigger flag is reached, the proportion of candidate nitrites is screened through outer Gaussian process-Bayes optimization, and the concentration of nitrite replacing nitrate is automatically adjusted through inner model prediction control. The specific percentage of candidate nitrites selected by the outer Gaussian process-Bayesian optimization screening is as follows: by For input, This indicates the proportion of nitrite nitrogen to total nitrogen oxides in the influent. This indicates the number of days the reactor has been in operation since inoculation was completed; in... For the output, construct a Matérn 5 / 2+ linear kernel Gaussian process (GP): ; In the formula: This represents the variance of the Mattérn kernel signal. Represents linear kernel variance. Represents the length scale of the d-th dimension ( All are hyperparameters, and marginal likelihood is used (retrained every 24 hours). dis represents Mahalanobis distance. Simultaneously, for each of the three safety constraints—nitrite effluent constraint g1, FNA inhibition constraint g2, and anammox bacteria non-retreat constraint g3—an independent Gaussian process classifier is constructed to output the corresponding constraint safety probability. : ; In the formula: Indicates the safe upper limit for nitrite nitrogen in the effluent; exist Improve point selection by applying constraints on the mesh: ; In the formula: Indicates the desired improvement in constraints; , Representing the standard normal CDF and PDF respectively; , Let represent the posterior mean and standard deviation of the Gaussian process, respectively. This represents the current best observed value (i.e., maxy, the highest historical ammonia nitrogen removal rate). Final output of candidate nitrite percentage .

[0045] The inner model predictive control (MPC) performs the following tracking: Will As the MPC reference trajectory, and at the end of each hydraulic residence cycle, a rolling optimization of Np=3 steps is solved: ; In the formula: This represents the reference value for ammonia nitrogen removal rate (which increases from 0.3 to 0.85 during the start-up phase). This represents the proportional change at step k+i (i.e. ); These represent the corresponding weighting coefficients (generally) ); constraint: ; The prediction model uses the biochemical equations derived from the EKF state estimation in step S2 to linearize at the current operating point, solves by quadratic programming, and is recalculated every 5 minutes.

[0046] The automatic adjustment of the nitrite-to-nitrate concentration is as follows: PLC is obtained according to MPC. Adjust the speed of the two inlet pumps for both water and medicine: ; In the formula: This indicates the target concentration of total nitrogen oxides in the influent (which remains constant). These represent the concentrations of the two stock solutions, respectively. New observations will be made after each hydraulic residence cycle. Added to the training set of Gaussian processes; If at the end of two consecutive hydraulic residence cycles If it rises or FNA > 0.08, then Back to , ; like Decrease in two consecutive hydraulic residence cycles or If the decrease is greater than 10%, then Back to And called the police. ; If the Gaussian process predicts 95% posterior standard deviation The model predictive control is performed with a conservative step size of 10%. A set is issued for each hydraulic residence cycle. The training set for Gaussian processes continues to grow; when When step S1 is achieved and the nitrogen content in the effluent is continuously and stably stable, proceed to step S5.

[0047] Step S5, Startup Complete: The startup is determined to be complete. The PLC switches to the normal Anammox operating mode. The nitrogen volumetric loading rate (NLR) is increased or the hydraulic retention time (HRT) is shortened in a gradient of 10% to 20% according to the target wastewater quality. Each step is still controlled by model prediction for safe tracking. Finally, the entire startup trajectory was packaged and archived as a priori for the Gaussian process of the next batch of reactors, forming a self-evolving closed loop.

[0048] Example 2: As another preferred embodiment of the present invention, based on the scheme of Example 1, a sequencing batch reactor with an effective volume of 10 L is used to inoculate the residual activated sludge and anaerobic granular sludge from the municipal wastewater treatment plant in a mass ratio of 1:1; during the initial start-up phase, nitrate nitrogen accounts for 70% to 100% of the nitrogen oxides in the influent, and sodium acetate, glucose or other external organic carbon are not added.

[0049] Start-up is completed when nitrite nitrogen accounts for more than 80% of the nitrogen oxides in the influent, and ammonia nitrogen and nitrite are stably removed (ammonia nitrogen removal rate is not less than 80%, NLR 0.05→0.3).

Claims

1. A rapid start-up method for anaerobic ammonia oxidation based on dynamically adjusting nitrogen oxide concentration, characterized in that: The sequencing batch reactor (SBR) is used, and the specific steps include: Step S1, Inoculation and Establishment of Anaerobic Operating Conditions: Inoculate the sludge into the sequencing batch reactor and simultaneously introduce low ammonia nitrogen wastewater, and initialize the model; Step S2, Initial Start-up: Perform three-nitrogen soft measurement, extended Kalman filter state estimation, and cumulative nitrogen balance verification in sequence; Step S3, Replacement Timing Determination: Based on three criteria—significant ammonia nitrogen removal, increased nitrate residue, and no nitrite accumulation—as well as circumstantial evidence of anammox bacterial growth, the replacement timing is determined through fuzzy fusion, i.e., the replacement trigger flag is output. Step S4: Replace nitrate with nitrite: After the replacement trigger flag is reached, the proportion of candidate nitrites is screened through outer Gaussian process-Bayes optimization, and the concentration of nitrite replacing nitrate is automatically adjusted through inner model prediction control. Step S5, Startup Complete: The startup is determined to be complete, and the PLC switches to the normal Anammox operating mode; and the nitrogen volumetric loading rate is increased or the hydraulic retention time is shortened according to the target wastewater quality gradient. Each step is still controlled by model prediction for safe tracking.

2. The rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 1, characterized in that: The sequencing batch reactor (SBR) is equipped with an air collection port, a water inlet, a water outlet, and an aeration port. One end of the air collection port is connected to an air collection bag, and one end of the water inlet is connected to a first peristaltic pump, the other end of which is connected to an external water inlet tank. The timed opening and closing of the first peristaltic pump is controlled by a timer. The water outlet is connected to an external water outlet tank via a second peristaltic pump. The SBR is placed on a shaking table. A UV-Vis full-spectrum probe is installed in the middle of the inner side wall of the SBR, and the UV-Vis full-spectrum probe is located at 1 / 2 underwater level. The pH electrode, ORP electrode, DO electrode, temperature electrode, and EC conductivity electrode are all in the same protective sleeve as the UV-Vis full-spectrum probe. An inlet flow meter is installed on the water inlet pipe between the first peristaltic pump and the SBR, an outlet flow meter is installed between the second peristaltic pump and the outlet tank, a level gauge is installed on the top of the SBR, and a wet gas flow meter is installed on the pipe between the air collection port and the air collection bag.

3. A rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 1 or 2, characterized in that: In step S1, the inoculated sludge is a mixture of two or more of the following: ordinary activated sludge, anaerobic granular sludge, denitrifying sludge, and municipal activated sludge.

4. A rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 2 or 3, characterized in that: In step S1, the ammonia nitrogen concentration of the low ammonia nitrogen wastewater... S NH4,in The nitrogen oxide concentration is 15–25 mg N / L. S NO3,in The concentration is 20–30 mg N / L; simultaneously, the proportion of nitrite nitrogen to total nitrogen oxides in the influent during the initial startup phase is... , S NO2,in This indicates the concentration of nitrite in the influent.

5. A rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 2 or 4, characterized in that: In step S1, after the sludge is fed into the sequencing batch reactor, it is carried out at 32±1℃ and shaking speed of 80~120 rpm; each hydraulic retention cycle is 15 min for water inlet → 2~4 h for reaction → 30 min for settling → 10 min for drainage, with a water exchange ratio of 50%.

6. The rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 5, characterized in that: In step S2, the trinitrogen soft measurement specifically involves: Vector sampled every 5 minutes: ; In the formula: This represents the absorbance at wavelength λ. Soft measurement consists of two layers, specifically: The first layer is a partial least squares regression layer: ; ; In the formula: This represents the intercept of the i-th nitrogen regression equation. Represents the process quantum vector, express The j-th component; The second layer is for Long Short-Term Memory network residual correction, with a 12-step input window: ; Final soft measurement output: 。 7. The rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 6, characterized in that: In step S2, the extended Kalman filter state estimation specifically involves: State vector The state transition is discretized using a continuous equation: ; In the formula: This indicates the maximum specific growth rate of anammox. This represents the anammox half-saturation constant for nitrite. This represents the anammox ammonia nitrogen half-saturation constant. This represents the nitrite inhibition constant. Indicates the rate of endogenous death in anammox bacteria. This indicates the maximum specific growth rate of denitrifying bacteria. This represents the half-saturation constant of denitrified nitrates. This represents the denitrification COD half-saturation constant. Indicates the rate of denitrifying bacteria death. Indicates the COD-to-consumption rate. Indicates the carbon release rate during sludge decay. The concentration of dead / decaying sludge is indicated by Q, and Q and V represent the average flow rate and effective volume of the reactor, respectively. This represents the anammox cell yield coefficient. Indicates the denitrifying bacteria yield coefficient; This represents the anammox temperature correction factor. This represents the temperature correction factor for denitrifying bacteria, where R represents the gas constant and T represents the reactor temperature. E A This indicates the anammox activation energy. E D Indicates the activation energy of heterotrophic denitrification; Observation vector: , These represent the soft-measured values ​​of ammonia nitrogen, nitrite nitrogen, and nitrate nitrogen in the effluent, respectively. These represent the measured redox potential and pH value of the reactor, respectively; measured redox potential The regression relationship with trinitrogen is as follows: , Indicates the regression intercept. These represent the regression coefficients for the nitrate and nitrite terms, respectively. EKF outputs the state estimate at step t every 5 minutes. and the state covariance matrix Pix t .

8. The rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 7, characterized in that: In step S2, the cumulative nitrogen balance verification specifically involves: performing periodic water balance and nitrogen balance checks. ; In the formula: This represents the average gas production volume flow rate during the reactor cycle. Represents the standard molar volume. Indicates the conversion factor; like The continuous positive value and the increase in gas production measured by the wet gas flow meter prove that anammox has been activated, and this is written into step S3 fuzzy fusion. Output a set every 5 minutes After running continuously for 2 to 5 hydraulic residence cycles, proceed to step S3.

9. The rapid start-up method for low-ammonia nitrogen anaerobic ammonia oxidation based on dynamic adjustment of nitrogen oxide concentration according to claim 8, characterized in that: In step S5, the nitrogen volumetric loading rate is increased or the hydraulic retention time is shortened in a gradient of 10% to 20% according to the target wastewater quality. Each step is still controlled by model prediction for safe tracking. Finally, the entire startup trajectory was packaged and archived as a priori for the Gaussian process of the next batch of reactors, forming a self-evolving closed loop.

Citation Information

Patent Citations

  • Method for quickly starting anaerobic ammoxidation reactor

    CN104016478A

  • Anaerobic ammonia oxidation device and starting method thereof

    CN117285152A