A carbon neutrality control method and system for sewage treatment plants

Abstract

The present invention relates to the technical field of carbon neutrality control scheme design for sewage treatment plants, and specifically to a carbon neutrality control method and system for sewage treatment plants. The method comprises: collecting water quality, energy consumption and gas emission data in real time through multi-source sensors, inputting the data into a spatiotemporal graph convolutional network after cleaning by edge computing nodes to predict carbon emission trends, and dynamically optimizing aeration volume and sludge return ratio based on a fuzzy PID controller. The life cycle assessment method is used to track the carbon footprint of the entire process, and the improved NSGA‑III algorithm is combined to generate a multi-objective compensation strategy, giving priority to scheduling photovoltaic, biogas and green electricity resources. The digital twin platform and the blockchain evidence storage module are innovatively integrated, the former calibrates the model parameters in real time, and the latter generates an unalterable carbon neutrality certificate. The present invention solves the problems of one-sided data, lag in control and lack of credibility of traditional methods, and realizes minute-level perception, prediction and offset closed-loop control of carbon emissions in the sewage treatment process.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon neutrality control scheme design for sewage treatment plants, and specifically to a carbon neutrality control method and system for sewage treatment plants. Background Art

[0002] The current carbon neutrality control technology of sewage treatment plants has the following bottlenecks:

[0003] Data perception and processing flaws: Traditional methods rely on single sensors or offline monitoring, failing to capture dynamic characteristics such as sudden changes in influent load and uneven spatial distribution of dissolved oxygen (DO). Furthermore, they lack real-time data cleaning capabilities at the edge, leading to outliers that interfere with the accuracy of prediction models. For example, aeration control often uses fixed thresholds, which are difficult to adapt to water quality fluctuations, leading to over-aeration (increased energy consumption) or insufficient DO (reduced denitrification efficiency).

[0004] One-sided carbon emission accounting: Existing carbon tracking focuses mostly on direct emissions (CH4, N20), ignoring the full life cycle impact of indirect emissions (such as pharmaceutical production and purchased electricity). In addition, compensation strategies rely on a single energy source (such as photovoltaics only), failing to achieve multi-source coordinated optimization.

[0005] Lag of control strategies: Static models based on historical data cannot predict the coupled impact of external variables such as meteorological and power grid factors on carbon emissions. Process adjustments and compensation decisions often lag for several hours, making it difficult to achieve dynamic carbon neutrality rate tracking at the minute level.

[0006] Lack of credibility and traceability: Carbon offsets rely on manual records or centralized storage, which poses a risk of tampering and cannot meet the carbon trading market's rigid demand for data auditability.

[0007] The above problems make it difficult for existing technologies to support real-time and precise regulation of sewage treatment plants under complex working conditions, which restricts the achievement of carbon neutrality goals.

[0008] Therefore, the existing technology needs to be further developed. Summary of the Invention

[0009] The purpose of the present invention is to overcome the above-mentioned technical deficiencies and provide a carbon neutrality control method and system for sewage treatment plants to solve the problems existing in the prior art.

[0010] To achieve the above technical objectives, according to a first aspect of the present invention, the present invention provides a carbon neutrality control method for a sewage treatment plant, comprising:

[0011] S100, uses multi-source sensors to monitor sewage treatment plant operating parameters in real time, including influent quality, aeration volume, sludge concentration, energy consumption data, and greenhouse gas emissions;

[0012] S200, preprocessing the monitoring data based on the edge computing node, removing outliers and generating a standardized data stream;

[0013] S300: Input standardized data into the artificial intelligence prediction model to predict the energy consumption and carbon emission trends of wastewater treatment in the next 24 hours;

[0014] S400: Dynamically adjust wastewater treatment process parameters based on prediction results, including aeration volume optimization, sludge return ratio control, and output matching of renewable energy power generation equipment;

[0015] S500, through the carbon footprint life cycle tracking module, calculates the real-time carbon neutrality rate after process adjustment;

[0016] S600. If the carbon neutrality rate is lower than the preset threshold, the multi-objective optimization algorithm is triggered to generate a compensation strategy, and the in-plant photovoltaic energy storage system or the external green electricity trading interface is called first.

[0017] Specifically, the preprocessing in step S200 includes:

[0018] The sensor data is denoised using a sliding time window algorithm, and missing values ​​are filled using a long short-term memory network.

[0019] Specifically, the artificial intelligence prediction model is a spatiotemporal graph convolutional network that integrates an attention mechanism, and its input includes historical energy consumption data, weather forecast information, and power grid carbon emission factors.

[0020] Specifically, the dynamic adjustment includes:

[0021] According to the spatiotemporal distribution characteristics of dissolved oxygen concentration in the aeration tank, a fuzzy PID controller is used to achieve precise control of the aeration volume in each zone.

[0022] Specifically, the renewable energy power generation equipment includes:

[0023] Coordinated dispatching system of photovoltaic power generation units, biogas cogeneration units and micro hydro turbines.

[0024] Specifically, the carbon footprint tracking module adopts a life cycle assessment method, covering direct and indirect emissions throughout the entire sewage treatment process, including the implicit carbon emissions from pharmaceutical production and transportation.

[0025] Specifically, the multi-objective optimization algorithm is an improved NSGA-III algorithm, and the objective function simultaneously minimizes operating costs, carbon emissions, and the risk of effluent water quality exceeding standards.

[0026] Specifically, the method further comprises:

[0027] A virtual image of the sewage treatment plant is built through the digital twin platform, the operating deviations between the physical system and the virtual model are compared in real time, and the prediction model parameters are updated based on the deviation values.

[0028] Specifically, the execution effect of the compensation strategy is recorded through the blockchain evidence storage module to generate an unalterable carbon neutrality performance certificate.

[0029] According to a second aspect of the present invention, a carbon neutrality control system for a sewage treatment plant is provided, comprising:

[0030] Acquisition module, including multi-source sensors, for real-time monitoring of sewage plant operating parameters, including influent water quality, aeration volume, sludge concentration, energy consumption data and greenhouse gas emissions;

[0031] The control module is used to pre-process the monitoring data based on the edge computing node, eliminate outliers and generate standardized data streams; it is used to input the standardized data into the artificial intelligence prediction model to predict the energy consumption and carbon emission trends of sewage treatment in the next 24 hours; it is used to dynamically adjust the sewage treatment process parameters according to the prediction results, including aeration volume optimization, sludge return ratio control and output matching of renewable energy power generation equipment; it is used to calculate the real-time carbon neutrality rate after process adjustment through the carbon footprint life cycle tracking module. If the carbon neutrality rate is lower than the preset threshold, the multi-objective optimization algorithm is triggered to generate a compensation strategy, and the in-plant photovoltaic energy storage system or the external green electricity trading interface is called first.

[0032] Beneficial effects:

[0033] This invention significantly improves the intelligence and reliability of carbon neutrality control in sewage treatment plants through systematic innovation:

[0034] Multi-source perception and edge collaboration: A multi-dimensional sensor network captures the dynamic characteristics of water quality, energy consumption, and gas emissions in real time. Combined with efficient preprocessing by edge computing nodes, it eliminates data noise and ensures information integrity, providing high-precision input for subsequent models.

[0035] Spatiotemporal coupling prediction capability: The spatiotemporal graph convolutional network based on the attention mechanism integrates process topology and external variables (such as meteorological and power grid factors) to accurately analyze the spatiotemporal evolution of carbon emissions, breaking through the bottleneck of traditional models in representing complex correlation relationships.

[0036] Adaptive dynamic control: A fuzzy PID controller is used to achieve fine-grained adjustment of the aeration volume in different zones, adapting to water quality fluctuations and equipment status changes. This optimizes energy consumption distribution while ensuring denitrification efficiency, significantly improving process stability.

[0037] Full life cycle carbon tracking: A full-process accounting system covering direct emissions, indirect emissions, and carbon offsets accurately quantifies the synergistic emission reduction effects of process adjustments and renewable energy scheduling to support scientific decision-making.

[0038] Multi-objective collaborative optimization: The improved optimization algorithm balances operating costs, environmental risks and carbon neutrality goals, dynamically generates multi-energy collaborative scheduling strategies, and improves the overall energy efficiency and anti-interference capabilities of the system.

[0039] Trusted traceability and self-calibration mechanism: The digital twin platform uses virtual mirroring to achieve online updates of model parameters, ensuring a high degree of consistency between predictions and physical systems; blockchain technology solidifies carbon offset data and builds a traceable and tamper-proof trust system.

[0040] This invention constructs a closed-loop control link of "perception-prediction-regulation-verification", providing a complete technical framework for sewage treatment plants to achieve dynamic carbon neutrality. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of a carbon neutrality control method for a sewage treatment plant provided in a specific embodiment of the present invention;

[0042] Figure 2 It is a schematic diagram of the system composition of the carbon neutrality control system for sewage treatment plants provided in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is clearly and completely described below in conjunction with the drawings of the present invention. Based on the embodiments in this application, other similar embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of this application. In addition, the directional words mentioned in the following embodiments, such as "up", "down", "left", "right", etc., are only reference to the directions of the drawings. Therefore, the directional words used are used to illustrate rather than limit the invention.

[0044] The present invention will be further described below with reference to the accompanying drawings and preferred embodiments.

[0045] See also Figure 1 The present invention provides a carbon neutrality control method for a sewage treatment plant, comprising:

[0046] S100 uses multi-source sensors to monitor sewage treatment plant operating parameters in real time, including influent water quality, aeration volume, sludge concentration, energy consumption data and greenhouse gas emissions.

[0047] Specifically, the multi-source sensor is composed of Internet of Things sensors deployed in the grid tank, aeration tank, secondary sedimentation tank and outlet, which collect water quality, gas and equipment status data in real time.

[0048] It should be further explained that regarding the arrangement of the multi-source sensors, the design scheme of the present invention is as follows:

[0049] COD, ammonia nitrogen, and pH sensors (accuracy ±0.1mg / L) are deployed at the water inlet.

[0050] Dissolved oxygen (DO) sensors (range 0-15 mg / L) are arranged every 5 m in the aeration tank. It can be understood that based on the typical size of the aeration tank (50-100 m in length) and the experimental data of the oxygen diffusion gradient, the 5 m spacing can ensure that the spatial resolution of the dissolved oxygen concentration field is ≤10% error, covering the oxygen transfer differences at the beginning, middle and end of the tank body.

[0051] An ultrasonic sludge concentration meter (resolution 0.1%) is installed in the sludge return pipeline; the sampling frequency of the sludge concentration meter is preferably 1 time / minute. It is understandable that the response time constant of the sludge return system is 5-10 minutes, and 1Hz sampling can capture transient fluctuations (such as sludge swelling bursts) while avoiding high-frequency noise interference.

[0052] S200: Preprocess the monitoring data based on the edge computing node, remove outliers and generate a standardized data stream.

[0053] Specifically, the preprocessing in step S200 includes:

[0054] The sensor data is denoised using a sliding time window algorithm, and missing values ​​are filled using a long short-term memory network.

[0055] It should be further explained that, regarding step S200, the solution designed by the present invention includes:

[0056] 1. The sliding time window length is set to 60 seconds. When the standard deviation of the data in the window exceeds three times the historical data, it is judged as an outlier. It is understandable that the water flow change cycle of the sewage treatment plant is about 30-120 seconds (such as the impact of pump start and stop). The 60-second window can cover a complete fluctuation cycle. When the standard deviation threshold is set to three times the historical value, the detection sensitivity of sudden water quality shocks (such as industrial wastewater mixing) reaches 95%.

[0057] 2. LSTM (Long Short-Term Memory Network) Missing Value Filling Model:

[0058] The network structure is as follows: input layer (30 time steps), LSTM layer (128 neurons, tanh activation), and output layer;

[0059] Training data: Use the normal operating data of the sewage treatment plant for one year, 80% as the training set and 20% as the validation set;

[0060] Hyperparameters: learning rate 0.001, batch size 32, iteration 200 epochs;

[0061] Filling effect: When the number of missing consecutive data points does not exceed 3 sampling points (15 minutes), the error rate is <2%.

[0062] It is understandable that regarding the number of neurons (128): after grid search verification (64 / 128 / 256 comparison), the 128-layer has the smallest error difference between the training set (RMSE=0.12) and the validation set (RMSE=0.15), indicating that the overfitting risk is controllable.

[0063] It is understandable that regarding the training cycle (200 times): the loss function curve enters a plateau period between 150 and 200 cycles, and continued training causes the validation set error to increase by 0.3%.

[0064] It is understandable that regarding the learning rate (0.001): When using the Adam optimizer, 0.001 is the recommended initial value for LSTM by the NVIDIA GPU computing library, which can balance the convergence speed and stability.

[0065] S300: Input standardized data into the artificial intelligence prediction model to predict the energy consumption and carbon emission trends of sewage treatment in the next 24 hours.

[0066] Specifically, the artificial intelligence prediction model is a spatiotemporal graph convolutional network that integrates an attention mechanism, and its input includes historical energy consumption data, weather forecast information, and power grid carbon emission factors.

[0067] It should be further explained that the present invention designs the following spatiotemporal graph convolutional network (ST-GCN) architecture:

[0068] 1. Input layer:

[0069] Time dimension: historical 24-hour data (sampling interval is 5 minutes);

[0070] Spatial topology: The process units of the sewage treatment plant (screen tank → aeration tank ← secondary sedimentation tank) are defined as directed graph nodes.

[0071] 2. Spatiotemporal convolution formula:

[0072]

[0073] Where:

[0074] : The feature matrix of the lth layer node;

[0075] : K sub-matrices divided by the adjacency matrix (K=3 in this embodiment);

[0076] : The convolution kernel weight of the kth subgraph;

[0077] : GELU activation function.

[0078] The rationale for choosing the time step (24 hours, 5 minute intervals) is understandable:

[0079] The nitrification / denitrification cycle of the sewage treatment process is about 6-24 hours, and the 5-minute interval can analyze the DO control lag effect (for example, the effluent ammonia nitrogen decreases 30-40 minutes after the aeration rate changes).

[0080] It is understandable that the reasons for choosing the adjacency matrix partition (K=3) are:

[0081] Due to the material transfer relationship corresponding to the mainstream process units of the sewage treatment plant (pretreatment → biological treatment → sedimentation), the weight distribution is pretreatment → biological treatment (weight 0.6), biological treatment self-circulation (weight 0.3), and biological treatment → sedimentation (weight 0.1).

[0082] It is understandable that the reasons for choosing the GELU activation function are:

[0083] Compared with ReLU, GELU achieves a better balance between gradient vanishing and nonlinear expression ability. Tests have shown that it can reduce the model prediction error by 2.1%.

[0084] 3. Attention mechanism design:

[0085] Temporal attention weight calculation:

[0086]

[0087] : are the state vectors of the current moment and the historical moment respectively.

[0088] : Feature dimension (this embodiment is set =64). It is understandable that a 64-dimensional vector can encode the interactive relationship of key process states (such as DO, MLSS, and temperature). When the dimension exceeds 128, the computation time increases by 50% while the accuracy only improves by 0.7%.

[0089] Prediction output:

[0090] Prediction error: The total energy consumption prediction error for the next 24 hours is ≤5%, and the carbon emissions error is ≤8%;

[0091] Training data enhancement: ±10% Gaussian noise is added to the water inlet flow rate to improve the robustness of the model. It can be understood that the typical error range of the simulated water inlet flow sensor (±8%-12%) increases the probability of the model's prediction error ≤5% under noise interference to 92%.

[0092] S400: Dynamically adjust sewage treatment process parameters based on prediction results, including aeration volume optimization, sludge return ratio control, and output matching of renewable energy power generation equipment.

[0093] Specifically, the dynamic adjustment includes:

[0094] According to the spatiotemporal distribution characteristics of dissolved oxygen concentration in the aeration tank, a fuzzy PID controller is used to achieve precise control of the aeration volume in each zone.

[0095] Specifically, the renewable energy power generation equipment includes:

[0096] Coordinated dispatching system of photovoltaic power generation units, biogas cogeneration units and micro hydro turbines.

[0097] It should be further explained that the present invention designs the following fuzzy PID controller:

[0098] Input variable error e(t) = DO set value (preset to 2.0 mg / L) - measured value;

[0099] Fuzzy rule table: 49 rules (7×7) are formulated based on e(t) and its rate of change Δe(t);

[0100] Specifically, regarding the fuzzy rule table, the solution designed by the present invention includes:

[0101] 1. Input / output variable definition:

[0102] Input variable 1: Dissolved oxygen error e(t) = DO set value (2.0 mg / L) - measured DO value;

[0103] Fuzzy set partitioning (unit: mg / L): NL (negative large): [-5, -3], NM (negative medium): [-3, -1], NS (negative small): [-1, 0], ZE (zero): [-0.5, 0.5], PS (positive small): [0, 1], PM (positive medium): [1, 3], PL (positive large): [3, 5].

[0104] Input variable 2: Error change rate Δe(t) = e(t) - e(t-1);

[0105] Fuzzy set partitioning (unit: mg / L / min): NL: [-0.5, -0.3], NM: [-0.3, -0.1], NS: [-0.1, 0], ZE: [-0.05, 0.05], PS: [0, 0.1], PM: [0.1, 0.3], PL: [0.3, 0.5].

[0106] Output variable: aeration rate adjustment percentage ΔQ (relative to the baseline aeration rate).

[0107] Fuzzy set partitioning: NL: -20%, NM: -12%, NS: -5%, ZE: 0%, PS: +5%, PM: +12%, PL: +20%.

[0108] 2. The fuzzy rule table designed by the present invention is shown in Table 1:

[0109] Table 1 Fuzzy rules table

[0110]

[0111] 3. Detailed explanation of rule logic:

[0112] 3.1 Rule design principles:

[0113] Diagonal symmetry: When e(t) and Δe(t) are in opposite directions (e.g., e=PL and Δe=NL), the maximum correction (PL or NL) is used to quickly eliminate the deviation.

[0114] Steady-state regulation: When e(t)=ZE, if Δe(t)=ZE, maintain the current aeration volume (output ZE); if Δe(t)≠ZE, suppress overshoot in advance according to the trend.

[0115] Nonlinear compensation: In extreme e(t) regions (such as PL / NL), even if Δe(t) is small, significant adjustments are required to break through process inertia.

[0116] 3.2 Typical rule examples:

[0117] Rule 1: (e=PL, Δe=NL), output PL (+20%);

[0118] Applicable scenarios: The measured DO is much lower than the set value (e=PL) and the deviation continues to expand (Δe=NL, that is, the error change rate is large and negative).

[0119] Action logic: The aeration volume needs to be increased to the maximum extent to quickly increase the DO concentration and avoid the collapse of the nitrification reaction.

[0120] Rule 13: (e=ZE, Δe=PS), output NM(-12%);

[0121] Applicable scenario: DO is close to the set value (e=ZE), but there is an upward trend (Δe=PS, that is, the error change rate is small and positive).

[0122] Action logic: Reduce aeration volume in advance to prevent DO overshoot and energy waste.

[0123] Rule 28: (e=NS, Δe=ZE), output NL (-20%);

[0124] Applicable scenario: DO is slightly higher than the set value (e=NS) and the trend is stable (Δe=ZE).

[0125] Action logic: Although the deviation is small, because it is in the sludge settling sensitive range, the aeration volume needs to be aggressively reduced to avoid sludge overturning in the secondary sedimentation tank.

[0126] 4. Membership function and defuzzification:

[0127] 4.1 Input variable membership function (triangular):

[0128] Membership function parameters of e(t) (unit: mg / L):

[0129] NL: [-5,-3,-1];

[0130] NM: [-3,-1,0];

[0131] NS: [-1, 0, 1];

[0132] ZE: [-0.5, 0, 0.5];

[0133] PS: [0,1,3];

[0134] PM: [1,3,5];

[0135] PL: [3,5,7].

[0136] Note: Interval extension is used to handle out-of-limit values.

[0137] 4.2 Output variable defuzzification uses the center of gravity method (COG) to calculate the exact output value:

[0138]

[0139] The activation degree of the i-th rule (the minimum value of the input membership);

[0140] : The output value of the i-th rule.

[0141] 5. Verification of actual control effect:

[0142] In the aeration tank step response test (DO setpoint increased from 1.5 to 2.0 mg / L):

[0143] Overshoot: ≤3% (traditional PID is 15-20%);

[0144] Adjustment time: 18 minutes (traditional PID takes 40-50 minutes);

[0145] Steady-state error: ±0.1mg / L (meets the process requirement of effluent ammonia nitrogen <1mg / L);

[0146] Rule table parameters are selected based on:

[0147] 7-level classification (not 5 or 9):

[0148] The seven fuzzy sets can cover the nonlinear characteristics of the wastewater treatment process (such as nitrification / denitrification thresholds) while avoiding rule explosion (49 vs 25 or 81 items).

[0149] Output amplitude (±20%):

[0150] Based on the maximum adjustment capability of the variable frequency fan (conventional models allow ±25% flow adjustment), a 5% safety margin is reserved.

[0151] ZE interval (±0.5mg / L):

[0152] Matches the DO sensor accuracy (±0.1mg / L), limiting the impact of measurement noise to the non-operating area.

[0153] Asymmetric regular distribution:

[0154] When e(t) is negative (DO is too high), the rules are more aggressive (such as rule 28 outputs NL) because high DO can easily cause sludge bulking, which needs to be quickly suppressed.

[0155] Example rule:

[0156] Ife(t)=Positive Large and Δe(t)=Negative Medium, the output aeration rate increases by 15%.

[0157] Quantization parameters:

[0158] Proportional coefficient Kp=0.8 (fast response);

[0159] Integration time Ti = 120s (to suppress steady-state error);

[0160] Differential time Td=20s (to suppress overshoot).

[0161] It is understandable that the reasons for the parameter selection of the fuzzy PID controller are:

[0162] DO setting value (2.0 mg / L): Based on the study of nitrifying bacteria activity, the ammonia nitrogen removal rate decreases by 30% when DO < 1.5 mg / L, and the energy consumption increases sharply and may cause sludge disintegration when DO > 2.5 mg / L. 2.0 mg / L is a typical compromise value.

[0163] Proportional coefficient (Kp=0.8): Through step response testing, Kp=0.8 can make the system overshoot ≤5% (traditional PID overshoot is 15%), and the adjustment time is shortened to 20 minutes.

[0164] Integration time (Ti=120s): The integration action is used to eliminate steady-state errors in the aeration system (such as fan efficiency decay). 120s corresponds to 1 / 120 of the sludge age (SRT) to avoid integration saturation.

[0165] Differential time (Td=20s): Suppresses DO oscillation caused by sudden changes in influent load. 20s is twice the response time of the dissolved oxygen sensor (10s), effectively filtering out high-frequency noise.

[0166] Aeration rate optimization formula:

[0167]

[0168] : Area weight of the i-th partition of the aeration pond (Σ =1).

[0169] It should be noted that the preferred weight in the present invention is =0.5, the reason for this selection is that the BOD load at the head end of the aeration tank accounts for 60%-70% of the total, and 50% of the aeration volume needs to be allocated to ensure sufficient carbon oxidation; the end weight =0.2 is used to maintain nitrification reaction.

[0170] Implementation effect: Compared with traditional PID, energy consumption is reduced by 12%-18%.

[0171] S500 calculates the real-time carbon neutrality rate after process adjustment through the carbon footprint life cycle tracking module.

[0172] Specifically, the carbon footprint tracking module adopts a life cycle assessment method, covering direct and indirect emissions throughout the entire sewage treatment process, including the implicit carbon emissions from pharmaceutical production and transportation.

[0173] It should be further explained that regarding carbon footprint tracking and compensation, the solutions designed by the present invention include:

[0174] Design life cycle assessment (LCA) calculation formula:

[0175] ① Design total carbon emission formula:

[0176]

[0177] in:

[0178] : The total carbon emissions of a sewage treatment plant throughout its life cycle (unit: kgCO2), that is, the net carbon emissions after direct emissions, indirect emissions and carbon offset measures.

[0179] : Direct carbon emissions, which come from greenhouse gases (CH4, N20, etc.) released directly during sewage treatment;

[0180] : Indirect carbon emissions, covering the implicit carbon emissions from purchased energy, chemical production and transportation, and other related links;

[0181] : Carbon compensation amount, which is the amount of carbon emissions offset by renewable energy (such as photovoltaic and biogas power generation) or carbon sink projects, and the value is negative (reducing net emissions).

[0182] ② Design direct carbon emission formula:

[0183]

[0184] in:

[0185] : Methane emissions (unit: kg), mainly from the decomposition of organic matter in oxygen-deficient links such as anaerobic digesters and sludge storage tanks;

[0186] : The global warming potential of methane (IPCCAR6 standard is 27.9), which means that the greenhouse effect of 1kgCH4 is equivalent to 27.9kgCO2;

[0187] : Nitrous oxide emissions (unit: kg), mainly from denitrification processes (such as nitrification / denitrification) by-products;

[0188] : The global warming potential value of nitrous oxide (IPCCAR6 standard is 273), reflecting the long life (about 121 years) and strong radiative forcing effect of N20.

[0189] The GWP value adopts the IPCCAR6 standard (CH4=27.9, N2O=273).

[0190] It is understandable that the GWP of CH4 is 27.9 (IPCCAR6): using the latest AR6 assessment report data (instead of AR5's 28-36), reflecting the latest research results on the shortening of methane's residence time in the atmosphere.

[0191] It is understandable that the GWP of N2O is 273: the radiative forcing effect of N2O dominates in sewage treatment emissions, and the use of 273 can avoid underestimating the emissions from the denitrification process (traditional values ​​are 265-298).

[0192] ③ Design indirect carbon emission formula:

[0193]

[0194] in:

[0195] : Power purchased from the grid (unit: kWh), covering the energy consumption of equipment such as water pumps and aeration fans;

[0196] : Grid carbon emission factor (example value 0.55 ), which represents the embodied carbon emissions for every 1kWh of electricity consumed; data source: using regional real-time factors (such as 0.55 for China Southern Power Grid and 0.75 for Northwest Power Grid) to reflect the differences in different energy structures (the proportion of coal-fired power). It is understandable that, based on dynamic design, the present invention adopts time-of-use factors (such as during peak hours when the proportion of coal-fired power is high, EFgrid can reach 0.8; at night when hydropower is mainly used, EFgrid≈0.2) to guide sewage treatment plants to concentrate electricity consumption during low-carbon periods.

[0197] : Chemical dosage (unit: kg), such as polyaluminium chloride (PAC), carbon source (sodium acetate), etc.;

[0198] : Chemical agent emission factor (example value PAC=1.2 ), covering the entire process of pharmaceutical production, transportation, and packaging, specifically including:

[0199] Bauxite mining (0.3 );

[0200] Hydrochloric acid synthesis (0.5 );

[0201] Transport (0.2 );

[0202] Package (0.2 ).

[0203] Data source: Based on the Ecoinvent database, for example:

[0204] PAC (polyaluminium chloride): 1.2 (including energy consumption for bauxite mining and hydrochloric acid synthesis);

[0205] Sodium acetate: 2.8 (Including carbon emissions from fermentation production and distillation purification).

[0206] The grid emission factor EF_grid uses regional real-time data (example value: 0.55 );

[0207] Chemical emission factor EF_chem is based on Ecoinvent database (e.g. PAC=1.2 ).

[0208] The following are some application examples:

[0209] Data of a sewage treatment plant:

[0210] =50kg / d, =5kg / d• =1000kWh / d, =200kg / d• =-300 (Offset by photovoltaic power generation).

[0211] calculate:

[0212] Direct emissions:

[0213]

[0214] Indirect emissions:

[0215]

[0216] Total emissions:

[0217]

[0218] ④Design a calculation formula for real-time carbon neutrality rate:

[0219]

[0220] S600. If the carbon neutrality rate is lower than the preset threshold, the multi-objective optimization algorithm is triggered to generate a compensation strategy, and the in-plant photovoltaic energy storage system or the external green electricity trading interface is called first.

[0221] Specifically, the multi-objective optimization algorithm is an improved NSGA-III algorithm, and the objective function simultaneously minimizes operating costs, carbon emissions, and the risk of effluent water quality exceeding standards.

[0222] Specifically, if the real-time η carbon neutrality is less than 95%, multi-level compensation will be triggered, including:

[0223] Level 1: Calling on the photovoltaic energy storage in the factory (when SOC ≥ 30%, the maximum offset amount = energy storage capacity × 0.85 );

[0224] Level 2: Increase the load of the biogas generator set to 85% (NOx needs to be monitored to be less than 50 ppm);

[0225] Level 3: Purchase green electricity through blockchain API (select a period when EF_grid < 0.1);

[0226] Weight distribution: photovoltaic: biogas: green electricity = 6:3:1 (optimized based on economic efficiency).

[0227] The following is a demonstration using a specific example:

[0228] Scenario: After optimizing the aeration rate at a sewage treatment plant, the real-time carbon neutrality rate needs to be calculated.

[0229] Original data (before adjustments):

[0230] , , ;

[0231] Baseline carbon emissions .

[0232] Data after process adjustment:

[0233] down to (Aeration efficiency improved), (saving 10% electricity);

[0234] (New photovoltaic panels).

[0235] calculate:

[0236] Net emissions ;

[0237] Carbon neutrality rate (excess neutralization);

[0238] Conclusion: Process adjustments have increased the carbon neutrality rate from 75% (150 / 200) to 105%, and green electricity procurement needs to be reduced to optimize costs.

[0239] Specifically, it is necessary to further explain the compensation policy triggering threshold:

[0240] It can be understood that, taking into account the monitoring error (±3%) and carbon offset costs, the 95% threshold can balance environmental benefits and economic efficiency, which is higher than the EU carbon trading market compliance standard (93%). The carbon neutrality rate threshold in the present invention is preferably set at 95% (lower than this, compensation will be triggered).

[0241] Compensation priority:

[0242] Regarding the photovoltaic energy storage system in the factory (discharging when SOC ≥ 30%), it is understandable that based on historical extreme weather data (three consecutive days of rain), 30% SOC can maintain the operation of key equipment (blower, return pump) for 48 hours.

[0243] The load rate of the biogas generator set is increased to 85%-90%. It is understandable that this load range corresponds to the optimal efficiency point of the biogas engine (heat-to-electricity ratio of 1.2-1.5). When the load is greater than 90%, the risk of excessive NOx emissions increases by 50%;

[0244] Purchase external green electricity (select the period when real-time EFgrid ≤ 0.1kgCO2e / kWh).

[0245] Specifically, the method further comprises:

[0246] A virtual image of the sewage treatment plant is built through the digital twin platform, the operating deviations between the physical system and the virtual model are compared in real time, and the prediction model parameters are updated based on the deviation values.

[0247] It should be further explained that the present invention designs the following virtual image construction process:

[0248] 1. Data synchronization layer:

[0249] Hardware: Deploy an edge gateway (NVIDIA Jetson AGX Xavier) and an OPC UA protocol server to synchronize physical system data (DO, MLSS, energy consumption, etc.) every 5 seconds.

[0250] Data preprocessing: Use a sliding window (60 seconds) to filter noise, standardize the data format (JSON Schema), and then transmit it to the cloud twin engine.

[0251] 2. Model Architecture:

[0252] Hybrid Modeling:

[0253] Physical model: Based on the activated sludge process (ASM1), a mass balance equation is constructed to describe the dynamic changes of COD, ammonia nitrogen, and DO.

[0254]

[0255] (X: sludge concentration, μ: specific growth rate, Q_{}: sludge discharge);

[0256] Data-driven model: Superimposes an LSTM network (64 neurons in the hidden layer) to compensate for unmodeled dynamics (such as the impact of sudden changes in water temperature).

[0257] 3. Real-time deviation calculation:

[0258] Indicators: Normalized root mean square error (NRMSE) was used to compare key parameters (DO, effluent ammonia nitrogen):

[0259]

[0260] y_i: measured value of the physical system, _i: output value of the virtual model.

[0261] Threshold setting: NRMSE>15% is considered a significant deviation (needs to trigger model update).

[0262] 4. Parameter update mechanism:

[0263] Online learning: When the NRMSE exceeds the threshold, the Bayesian optimization algorithm is started to adjust the LSTM weights, giving priority to updating neurons related to the deviation (such as the DO prediction module).

[0264] Update frequency: once every 10 minutes at most to avoid high-frequency calculation overload.

[0265] Specifically, the execution effect of the compensation strategy is recorded through the blockchain evidence storage module to generate an unalterable carbon neutrality performance certificate.

[0266] 5. Implementation Cases:

[0267] Scenario: The measured DO values ​​in the aeration tank are consistently lower than the model predictions (NRMSE = 18%).

[0268] action:

[0269] The Bayesian optimizer adjusts the DO prediction branch learning rate of LSTM from 0.001 to 0.002.

[0270] After the update, the NRMSE dropped to 8% and the model was re-aligned.

[0271] It should be further explained that the present invention designs the following virtual image construction process:

[0272] 1. Blockchain architecture:

[0273] Type: Consortium chain (Hyperledger Fabric 2.4), participating nodes include sewage treatment plants, environmental protection bureaus, and third-party auditing agencies.

[0274] 2. Node role:

[0275] Sewage plant node: Submit compensation data (photovoltaic power generation, biogas utilization).

[0276] Audit node: Verify data authenticity (based on sensor digital signature).

[0277] Regulatory node: issues certificates and stores them on the chain.

[0278] 2.1 Data on-chain process:

[0279] Data encapsulation:

[0280] Fields: timestamp, photovoltaic power generation (kWh), biogas CH4 volume (m³), hash value (SHA-256).

[0281] Signature: Use the factory private key (RSA-2048) to sign the data packet to ensure the source is trusted.

[0282] 2.2 Smart Contract Logic:

[0283] Credential generation (chaincode example):

[0284] func generateCarbonCredits(stub shim.ChaincodeStubInterface, args []string) {

[0285] / / Input: photovoltaic power generation solar, biogas

[0286] / / Calculate carbon offset

[0287] offset: = solar * 0.85 + biogas * 0.67 * 27.9

[0288] / / Generate credential ID: timestamp + first 8 digits of hash

[0289] certificateID:= timestamp + SHA256(solar|biogas)[0:8]

[0290] / / Write to blockchain

[0291] stub.PutState(certificateID, offset)

[0292] }

[0293] Verification logic: Check if the sensor ID is in the registration list and the data hash matches the history.

[0294] 2.3 Carbon Neutrality Performance Certificate:

[0295] The credential structure is as follows:

[0296] {

[0297] "certificateID": "20231001-3A5F",

[0298] "timestamp": "2023-10-01T14:30:00Z",

[0299] "offsetAmount": 150.2, / / Unit: kgCO2e

[0300] "source": {"solar": 120, "biogas": 30},

[0301] "signature": "E2C3A5...",

[0302] "auditor": "EPB_Shanghai"

[0303] }

[0304] Query interface: Provides REST API for third-party verification. Enter the certificate ID to return the full amount of evidence data.

[0305] It should be noted here that the preferred parameter values ​​and their basis are shown in Table 2:

[0306] Table 2 Parameter optimization basis

[0307]

[0308] As you can understand, blockchain-stored credentials take less than 10 seconds to be generated and verified, enabling real-time audits by the Environmental Protection Agency and preventing greenwashing. This invention utilizes hybrid modeling and online learning to achieve self-calibration of the virtual model, adapting to water quality fluctuations and equipment aging. The consortium chain architecture designed by this invention balances efficiency and regulatory compliance. The hash chain and digital signature ensure that data cannot be tampered with, meeting the data traceability requirements of the carbon trading market.

[0309] See also Figure 2 The present invention provides another embodiment, which provides a carbon neutrality control system for a sewage treatment plant. The carbon neutrality control system for a sewage treatment plant includes:

[0310] The acquisition module 100 includes multiple source sensors for real-time monitoring of sewage treatment plant operating parameters, including influent water quality, aeration volume, sludge concentration, energy consumption data, and greenhouse gas emissions;

[0311] Control module 200 is used to pre-process monitoring data based on edge computing nodes, eliminate outliers and generate standardized data streams; it is used to input standardized data into an artificial intelligence prediction model to predict the energy consumption and carbon emission trends of sewage treatment in the next 24 hours; it is used to dynamically adjust sewage treatment process parameters according to the prediction results, including aeration volume optimization, sludge return ratio regulation and output matching of renewable energy power generation equipment; it is used to calculate the real-time carbon neutrality rate after process adjustment through the carbon footprint life cycle tracking module. If the carbon neutrality rate is lower than the preset threshold, the multi-objective optimization algorithm is triggered to generate a compensation strategy, and the in-plant photovoltaic energy storage system or the external green electricity trading interface is called first.

[0312] In a preferred embodiment, the present application further provides an electronic device, comprising:

[0313] A memory; and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the carbon-neutral control method for sewage treatment plants is implemented. The computer device can be broadly defined as a server, a terminal, or any other electronic device with the necessary computing and / or processing capabilities. In one embodiment, the computer device may include a processor, a memory, a network interface, a communication interface, etc. connected via a system bus. The processor of the computer device can be used to provide the necessary computing, processing and / or control capabilities. The memory of the computer device may include a non-volatile storage medium and an internal memory. An operating system, a computer program, etc. may be stored in or on the non-volatile storage medium. The internal memory can provide an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface and the communication interface of the computer device can be used to connect and communicate with external devices via a network. When the computer program is executed by the processor, the steps of the method of the present invention are performed.

[0314] The present invention can be implemented as a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, causes the steps of the method of an embodiment of the present invention to be performed. In one embodiment, the computer program is distributed on a plurality of computer devices or processors coupled to a network so that the computer program is stored, accessed, and executed in a distributed manner by one or more computer devices or processors. A single method step / operation, or two or more method steps / operations, can be performed by a single computer device or processor or by two or more computer devices or processors. One or more method steps / operations can be performed by one or more computer devices or processors, and one or more other method steps / operations can be performed by one or more other computer devices or processors. One or more computer devices or processors can perform a single method step / operation, or perform two or more method steps / operations.

[0315] Those skilled in the art will appreciate that the method steps of the present invention can be performed by instructing related hardware, such as a computer device or processor, through a computer program. The computer program can be stored in a non-transitory computer-readable storage medium, and when the computer program is executed, the steps of the present invention are performed. Depending on the circumstances, any reference herein to memory, storage, database, or other media may include non-volatile and / or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state disk, etc. Examples of volatile memory include random access memory (RAM), external cache memory, etc.

[0316] The various technical features described above can be combined arbitrarily. Although not all possible combinations of these technical features are described, any combination of these technical features should be considered to be covered by this specification as long as such combination does not conflict.

[0317] The specific embodiments of the present invention described above do not limit the scope of protection of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A carbon neutrality control method for a sewage treatment plant, characterized in that: The following steps are involved: S100, uses multi-source sensors to monitor sewage treatment plant operating parameters in real time, including influent quality, aeration volume, sludge concentration, energy consumption data, and greenhouse gas emissions; S200, preprocessing the monitoring data based on the edge computing node, removing outliers and generating a standardized data stream; S300: Input standardized data into the artificial intelligence prediction model to predict the energy consumption and carbon emission trends of wastewater treatment in the next 24 hours; S400: Dynamically adjust wastewater treatment process parameters based on prediction results, including aeration volume optimization, sludge return ratio control, and output matching of renewable energy power generation equipment; S500 uses the carbon footprint lifecycle tracking module to calculate the real-time carbon neutrality rate after process adjustments. The carbon neutrality rate is defined as: , is direct carbon emissions, is indirect carbon emissions, is the carbon offset amount; S600: If the carbon neutrality rate is lower than the preset threshold, a multi-objective optimization algorithm is triggered to generate a compensation strategy, with priority given to the in-plant photovoltaic energy storage system or the external green electricity trading interface; The multi-objective optimization algorithm is an improved NSGA-III algorithm, and the objective function simultaneously minimizes operating costs, carbon emissions, and the risk of effluent water quality exceeding standards.

2. The carbon neutrality control method for sewage treatment plants according to claim 1, characterized in that: The pre-processing in step S200 includes: The sensor data is denoised using a sliding time window algorithm, and missing values ​​are filled using a long short-term memory network.

3. The carbon neutrality control method for sewage treatment plants according to claim 1, characterized in that: The artificial intelligence prediction model is a spatiotemporal graph convolutional network that integrates an attention mechanism, and its input also includes historical energy consumption data, weather forecast information and power grid carbon emission factors.

4. The carbon neutrality control method for sewage treatment plants according to claim 1, characterized in that: The dynamic adjustment includes: According to the spatiotemporal distribution characteristics of dissolved oxygen concentration in the aeration tank, a fuzzy PID controller is used to achieve precise control of the aeration volume in each zone.

5. The carbon neutrality control method for sewage treatment plants according to claim 1, characterized in that: The renewable energy power generation equipment includes: Coordinated dispatching system of photovoltaic power generation units, biogas cogeneration units and micro hydro turbines.

6. The carbon neutrality control method for sewage treatment plants according to claim 1, characterized in that: The carbon footprint tracking module adopts the life cycle assessment method, covering the direct and indirect emissions of the entire sewage treatment process, including the implicit carbon emissions of pharmaceutical production and transportation.

7. The carbon neutrality control method for a sewage treatment plant according to any one of claims 1 to 6, characterized in that: The method further comprises: A virtual image of the sewage treatment plant is built through the digital twin platform, the operating deviations between the physical system and the virtual model are compared in real time, and the prediction model parameters are updated based on the deviation values.

8. The carbon neutrality control method for sewage treatment plants according to claim 6, characterized in that: The execution effect of the compensation strategy is recorded through the blockchain evidence module to generate an unalterable carbon neutrality performance certificate.

9. A carbon neutrality control system for sewage treatment plants, characterized in that: The carbon neutrality control method for a sewage treatment plant according to any one of claims 1 to 8 comprises: Acquisition module, including multi-source sensors, for real-time monitoring of sewage plant operating parameters, including influent water quality, aeration volume, sludge concentration, energy consumption data and greenhouse gas emissions; The control module is used to pre-process the monitoring data based on the edge computing node, eliminate outliers and generate standardized data streams; it is used to input the standardized data into the artificial intelligence prediction model to predict the energy consumption and carbon emission trends of sewage treatment in the next 24 hours; it is used to dynamically adjust the sewage treatment process parameters according to the prediction results, including aeration volume optimization, sludge return ratio control and output matching of renewable energy power generation equipment; it is used to calculate the real-time carbon neutrality rate after process adjustment through the carbon footprint life cycle tracking module. If the carbon neutrality rate is lower than the preset threshold, the multi-objective optimization algorithm is triggered to generate a compensation strategy, and the in-plant photovoltaic energy storage system or the external green electricity trading interface is called first.

Images