Wet garbage anaerobic digestion biogas liquid whole process real-time prediction intelligent regulation method and system

Abstract

This invention discloses a real-time predictive intelligent control method and system for the entire process of anaerobic digestion of wet waste biogas slurry, relating to the field of waste treatment technology. The system includes a biogas slurry storage unit, a nitrogen stripping unit, an electrochemical reaction unit, a membrane separation unit, a sediment collection unit, and a central control unit, all connected sequentially by pipelines. This invention solves the problem of periodic failure of the prediction model caused by the dynamic passivation-activation cycle of electrodes. This improvement makes the prediction model not only data-driven but also guided by physical laws, thus significantly improving the accuracy and robustness of predicting ammonia nitrogen removal rate, total phosphorus removal rate, and electrode efficiency trends even in non-stationary data environments, laying a reliable foundation for subsequent precise control.

Description

Technical Field

[0001] This invention relates to the field of waste treatment technology, and in particular to a real-time predictive intelligent control system for the entire process of anaerobic digestion of wet waste. Background Technology

[0002] In recent years, with the increasing demand for waste sorting and resource recovery, anaerobic digestion technology for wet waste (such as kitchen waste) has been widely used due to its ability to generate biogas energy. However, the large amount of biogas slurry produced in this process has become a prominent problem for the industry—biogas slurry has a complex composition, containing high concentrations of ammonia nitrogen, phosphorus, and organic pollutants. If not treated properly, it can easily cause eutrophication and secondary pollution of water bodies. To achieve the goal of "carbon neutrality," traditional biogas slurry treatment methods, such as direct discharge or simple aerobic treatment, are no longer sufficient to meet the dual requirements of resource recovery and efficient pollutant removal.

[0003] Currently, the industry is actively developing biogas slurry resource recovery systems based on material flow direction. These systems typically integrate units such as nitrogen stripping, electrochemical precipitation, and membrane separation, aiming to recover nitrogen, phosphorus, and carbon sources from the biogas slurry separately. For example, existing technologies include schemes that use a combination of stripping denitrification and membrane processes to recover ammonium salts and reclaimed water from liquid digestate of organic waste; simultaneously, promoting phosphorus crystallization recovery through electrochemical methods is also a known technological approach. These systems achieve, to a certain extent, targeted nutrient recovery.

[0004] However, those skilled in the art have discovered in practice that existing electrochemical resource recovery units have technical problems: during long-term continuous operation, the electrode surfaces within the electrochemical reaction unit undergo a dynamic "passivation-activation cycle." This periodic fluctuation in electrode state, caused by microscopic physicochemical changes at the reaction interface, such as the formation and peeling of the oxide layer, leads to non-stationarity in system operating data. When such data is input into predictive models trained based on historical conditions, such as machine learning algorithms used for intelligent control, prediction inaccuracies may occur in specific periods, for example, the inflection points of ammonia nitrogen and total phosphorus removal rates cannot be accurately predicted.

[0005] This technical problem has had a significant negative impact on practical engineering. The periodic failure of the predictive model causes the instructions issued by the central control system, such as adjusting the current density and stirring rate, to deviate from the actual optimal operating conditions. This not only makes the phosphorus crystallization efficiency unstable and the nitrogen removal incomplete, but may also exacerbate electrode wear and increase system energy consumption due to mis-control, ultimately restricting the long-term stable operation of the entire resource recovery system and the quality of the recovered products.

[0006] To address this, we propose a real-time predictive intelligent control system for the entire process of anaerobic digestion of wet waste and biogas slurry. Summary of the Invention

[0007] The purpose of this invention is to provide a real-time predictive intelligent control system for the entire process of wet waste anaerobic digestion slurry by embedding physical constraints of electrochemical reaction kinetics into a neural network model to solve the prediction failure problem caused by electrode passivation-activation cycle, thereby achieving accurate prediction and intelligent control of the treatment process and stable and efficient recovery of carbon, nitrogen, and phosphorus resources.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] Compared with the prior art, the present invention has the following beneficial effects:

[0010] (1) The real-time prediction and intelligent control system for the whole process of anaerobic digestion of wet waste proposes an improved prediction and control method for wet waste treatment system. By explicitly embedding the Butler-Volmer electrochemical reaction kinetic equation as a physical constraint in the training loss function of the neural network, and introducing the electrode impedance phase angle as an input feature to characterize the state of the electrode surface, the problem of periodic failure of the prediction model caused by the dynamic passivation-activation cycle of the electrode is solved. This improvement makes the prediction model not only dependent on data-driven but also guided by physical laws, thereby significantly improving the accuracy and robustness of the prediction of ammonia nitrogen removal rate, total phosphorus removal rate and electrode efficiency trend in non-stationary data environment, laying a reliable foundation for subsequent precise control.

[0011] (2) The real-time prediction and intelligent control system for the entire process of anaerobic digestion of wet waste proposes a novel online update mechanism for the physical constraint adaptive learning model. Its core technical means is to set a trigger condition based on the continuous exceeding of the threshold of prediction error, and to use a sliding window with a fixed capacity to store the latest operating data, and to dynamically update the neural network weights in an incremental learning manner. This method enables the model to quickly adapt to the changes in system characteristics caused by the changes in the state of the electrode surface without forgetting historical knowledge, and realizes the self-evolution and self-optimization of the prediction model in long-term operation, ensuring the continuous effectiveness of the intelligent control system throughout the entire life cycle.

[0012] (3) The real-time prediction and intelligent control system for the whole process of anaerobic digestion of wet waste slurry proposes a collaborative optimization control system for the resource utilization of wet waste slurry. Through the prediction output of the central control unit based on the physical constraint model, it synchronously coordinates and controls the process parameters of the nitrogen stripping unit, the power supply and stirring rate of the electrochemical reaction unit, and the membrane flux of the membrane separation unit, thereby realizing the collaborative optimization of the carbon, nitrogen and phosphorus recovery process. This technology breaks through the limitation of independent control of each unit and forms a closed-loop intelligent control system from the front-end pretreatment to the end product collection. Ultimately, it ensures that the system operates stably and efficiently with ammonia nitrogen removal rate ≥93%, total phosphorus removal rate ≥95%, effluent biodegradability BOD / COD ≥0.46, and sediment recovery rate ≥85% within the range of slurry treatment volume of 0.5-5L / h.

[0013] (4) The real-time prediction and intelligent control system for the whole process of wet waste anaerobic digestion biogas slurry proposes an overall intelligent control scheme for wet waste biogas slurry treatment system based on "data-physical model fusion drive". The overall idea is to build a central control unit that integrates a predictive analysis module with physical constraint adaptive learning capability, and to deeply integrate and control it with a series of physical units that have been specifically designed and optimized. The overall scheme embeds knowledge in fields such as electrochemistry and mass transfer process into the data-driven model in a computable form, and then dynamically offsets the complex time-varying disturbances inside the system through real-time prediction and execution control. Thus, the overall wet waste anaerobic digestion biogas slurry treatment process has achieved a leap from experience operation to precise intelligent control, and achieved the dual goals of stable treatment efficiency and maximum resource recovery. Attached Figure Description

[0014] Figure 1 A schematic diagram of the structure of the real-time prediction and intelligent control system for the entire process of anaerobic digestion of wet waste biogas slurry provided in an embodiment of the present invention;

[0015] Figure 2 A diagram illustrating the training and workflow of a physical constraint neural network in a real-time predictive intelligent control system for the entire process of anaerobic digestion of wet waste biogas slurry, provided as an embodiment of the present invention.

[0016] Figure 3 A schematic diagram illustrating the application of the real-time prediction and intelligent control method for the entire process of wet waste anaerobic digestion slurry, provided in an embodiment of the present invention, to the monitoring of parameters and the high-value output of the entire process. Detailed Implementation

[0017] 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.

[0018] like Figures 1 to 3 As shown, this embodiment provides a real-time predictive intelligent control system for the entire process of wet waste anaerobic digestion biogas slurry.

[0019] Example: Real-time predictive intelligent control system for the entire process of anaerobic digestion of wet waste biogas slurry, such as... Figures 1-3 As shown, it includes a biogas slurry storage unit, a nitrogen stripping unit, an electrochemical reaction unit, a membrane separation unit, a precipitate collection unit, and a central control unit, which are connected in sequence by pipelines.

[0020] The biogas slurry storage unit is used to buffer the biogas slurry produced by the anaerobic digestion of wet waste, and its output end is connected to the input end of the nitrogen stripping unit through a transfer pump;

[0021] The nitrogen stripping unit receives the biogas slurry output from the biogas slurry storage unit, removes ammonia nitrogen and adjusts the N / P ratio and C / N ratio through aeration, pH adjustment and temperature control, and outputs the pretreated biogas slurry to the electrochemical reaction unit;

[0022] The electrochemical reaction unit is used to receive the pretreated biogas slurry output from the nitrogen stripping unit. It achieves phosphorus crystallization recovery and deep nitrogen removal through electrode reaction, while retaining organic matter in the biogas slurry. The treated tailwater is output to the membrane separation unit, and the precipitate is output to the precipitate collection unit.

[0023] The membrane separation unit is used to receive the treated wastewater output from the electrochemical reaction unit, and purify it through membrane separation to output treated water that meets the carbon source standards for leachate.

[0024] The central control unit is electrically connected to the sensors and actuators of the biogas slurry storage unit, nitrogen stripping unit, electrochemical reaction unit, membrane separation unit, and sediment collection unit, respectively.

[0025] The central control unit includes a data acquisition module, a predictive analysis module, and an execution control module;

[0026] Among them, the prediction and analysis module integrates an adaptive learning model that incorporates physical constraints. This model embeds the electrochemical reaction kinetic equation as a physical constraint into the neural network algorithm to simulate and predict the dynamic passivation-activation cycle state of the electrode surface in the electrochemical reaction unit.

[0027] The predictive analysis module dynamically predicts the changing trends of ammonia nitrogen removal rate, total phosphorus removal rate, and electrode efficiency based on real-time collected data and the output of an adaptive learning model that integrates physical constraints, and generates control instructions.

[0028] The execution control module adaptively adjusts the power parameters and stirring rate of the electrochemical reaction unit according to the control instructions to counteract efficiency fluctuations caused by changes in the electrode surface state; at the same time, it coordinates and controls the operating parameters of the nitrogen stripping unit and the membrane separation unit; through real-time prediction and intelligent control of the central control unit, the system overcomes the influence of the non-stationary data generation mechanism caused by the dynamic passivation-activation cycle of the electrodes on the prediction model, ensuring stable processing efficiency and maximized resource recovery throughout the process.

[0029] Furthermore, in this embodiment, the adaptive learning model in the prediction and analysis module that integrates physical constraints adopts a feedforward neural network structure. This network includes an input layer, three hidden layers, and an output layer. The input layer nodes receive real-time feature data, including current, voltage, pH, temperature, ammonia nitrogen concentration, total phosphorus concentration, and electrode impedance phase angle derived from the electrochemical impedance spectroscopy sensor. The hidden layers use the ReLU activation function to handle nonlinear relationships, and the output layer provides prediction targets, including ammonia nitrogen removal rate, total phosphorus removal rate, and electrode state index. The neural network training process is based on a historical running dataset, which contains input features and corresponding measured output targets. During training, a loss function is minimized, defined as the sum of the data error term and the physical constraint term, specifically in the form of…

[0030]

[0031] in It is the sample size. These are measured values. These are neural network predictions, where t represents the real-time running time / time-series sampling time of the electrochemical reaction unit during the anaerobic digestion of wet waste. Hyperparameters are used to balance data fitting and physical consistency. The value was determined to be 0.05 using grid search and cross-validation. The optimization algorithm used was Adam, with a learning rate of 0.001 and 1000 training iterations. This represents a physical model of electrochemical reaction rates based on the Butler-Volmer equation, which describes the relationship between current density and overpotential, as shown in the following formula:

[0032]

[0033] in It is the current density. It is the exchange current density. and It is the transmission coefficient. It is Faraday's constant. It is an overpotential. It is the gas constant. It's temperature, exchange current density. and transmission coefficient , The values ​​were obtained through calibration in a standard electrolyte using linear sweep voltammetry, and the specific values ​​were determined for the magnesium-nickel composite electrode as follows: =0.01 A / cm 2 , =0.5, =0.5; Physical constraint embedding is achieved by incorporating the gradient difference between the physical model output and the neural network prediction into the loss function, ensuring that the prediction results conform to the basic principles of electrochemistry;

[0034] After training, the neural network performs forward propagation calculations in its working mechanism to generate predicted outputs based on real-time input features. At the same time, the model integrates online learning functions. When the prediction error continuously exceeds a preset threshold, incremental training is automatically triggered to update the network weights using the latest data within the sliding window, so as to dynamically adapt to changes in the state of the electrode surface.

[0035] Furthermore, in this embodiment, the electrochemical reaction unit includes an electrolytic cell, an electrode assembly, a power supply module, a pH adjustment assembly, a stirring device, and a time control module;

[0036] In this embodiment, the electrode assembly is determined to be a magnesium-nickel composite electrode system, wherein the anode is a three-dimensional porous foam nickel-magnesium composite electrode and the cathode is a graphite rod; the electrode plate area is fixed at 2cm*2cm and the thickness is fixed at 0.3cm; the power supply module is a constant current power supply, providing a current density of 15mA per square centimeter; the pH adjustment component is used to adjust the pH of the biogas slurry entering the electrolyzer to 9.5; the stirring device adopts a mechanical stirrer, and the stirring rate is adjusted in real time according to the electrode reaction rate to maintain uniform mixing of the biogas slurry in the electrolyzer; the time control module precisely controls the electrolysis time to 6 hours; the operating parameters of the electrochemical reaction unit are controlled in a closed loop by the central control unit based on the output of the predictive analysis module to ensure that the removal rate of ammonia nitrogen is not less than 40%, the removal rate of total phosphorus is not less than 78%, and the retention rate of biogas slurry COD is not less than 69%; the dynamic passivation-activation cycle of the electrode surface is monitored by integrated electrochemical impedance spectroscopy, and the data is fed back to the predictive analysis module for model optimization.

[0037] The nitrogen stripping unit includes a stripping reaction tank, an aeration device, a pH adjustment module, a temperature control module, and a sampling and detection module. The aeration device is located at the bottom of the stripping reaction tank, and the aeration rate is adjustable up to 4 liters per minute. The pH adjustment module is connected to the stripping reaction tank and uses a dosing pump to add sodium hydroxide solution to adjust the pH of the biogas slurry to 10. The temperature control module includes an electric heating device and a platinum resistance temperature sensor to stabilize the biogas slurry temperature at 55 degrees Celsius, with a temperature control accuracy of ±0.5 degrees Celsius. The heating power is adjusted by a PID controller, and the sensor accuracy is ±0.1 degrees Celsius. The sampling and detection module integrates an online ammonia nitrogen analyzer and a spectrometer to monitor ammonia nitrogen concentration, N / P ratio, and C / N ratio in real time. When the ammonia nitrogen removal rate reaches 90% and the N / P ratio drops to 1.35, the biogas slurry is automatically sent to the electrochemical reaction unit. The output parameters of the nitrogen stripping unit are optimized in real time by the central control unit to ensure that the pretreated biogas slurry meets the input requirements of the electrochemical reaction unit, including adjusting the C / N ratio to a suitable range to promote the efficiency of subsequent electrode reactions.

[0038] The membrane separation unit includes an organic membrane module, a biodegradability detection module, and a flow control module. The organic membrane module uses an ultrafiltration membrane with a molecular weight cutoff of 1000 Daltons to remove unreacted suspended particles and large molecular impurities from the effluent. The biodegradability detection module integrates BOD and COD sensors to monitor the BOD / COD ratio of the effluent in real time. When this ratio reaches 0.46, the effluent is deemed to meet the carbon source usage standards for leachate. The flow control module uses a variable frequency pump to regulate the flow rate of the effluent through the organic membrane, matching the receiving capacity of the subsequent leachate treatment system. The operation of the membrane separation unit is monitored and controlled by a central control unit, which dynamically adjusts the flow rate based on the output of the predictive analysis module to ensure stable treated water quality and maximize resource recovery efficiency.

[0039] The data acquisition module includes a multi-channel data interface and signal conditioning circuitry, used to acquire sensor data from each unit in real time. Acquired parameters include biogas slurry pH, temperature, ammonia nitrogen concentration, total phosphorus concentration, COD, BOD / COD ratio, N / P ratio, C / N ratio, electrolysis current, aeration rate, membrane separation flow rate, and electrode impedance phase angle. The data sampling frequency is 1 Hz, and the acquired data is transmitted to the predictive analysis module via industrial Ethernet for model input and system status assessment.

[0040] The execution control module includes a PLC controller and drive circuits, which receive control commands generated by the predictive analysis module and drive the actuators of each unit. Control actions include adjusting the aeration rate and pH dosage of the nitrogen stripping unit, adjusting the power output current and stirring rate of the electrochemical reaction unit, and controlling the flow pump speed of the membrane separation unit. The execution control module uses a PID control algorithm to ensure parameter adjustment accuracy. When parameters deviate from the set range, an alarm is automatically triggered and a correction operation is performed. For example, if the pH fluctuation exceeds 0.3 or the temperature fluctuation exceeds 3 degrees Celsius, an emergency adjustment program is initiated. The execution control module works in conjunction with other modules of the central control unit to achieve fully automated operation of the entire process.

[0041] The biogas slurry storage unit includes a storage tank made of corrosion-resistant material, a level sensor, and a transfer pump. The storage tank has a capacity of 1,000 liters and is used to buffer the biogas slurry produced by the anaerobic digestion of wet waste. The level sensor monitors the biogas slurry level in real time and automatically triggers an alarm when the level is below a set threshold. The transfer pump is a peristaltic pump with a flow rate range of 0.5 to 5 liters per hour to ensure a stable delivery of biogas slurry to the nitrogen stripping unit. The biogas slurry storage unit also integrates a stirrer to prevent solid sedimentation and ensure the homogeneity of the biogas slurry.

[0042] The sediment collection unit includes a sedimentation tank, a screw conveyor, and a drying device. The sedimentation tank receives the sediment discharged from the electrochemical reaction unit, the main component of which is magnesium ammonium phosphate crystals. The screw conveyor transports the sediment to the drying device, which uses hot air drying technology to reduce the moisture content of the sediment to below 10%. The collected dried sediment is used as a slow-release fertilizer. The sediment collection unit also integrates a weighing sensor to monitor the recovery rate in real time, ensuring that it is not less than 85%.

[0043] The overall operating parameters of the system are managed uniformly by the central control unit; the biogas slurry treatment rate is controlled between 0.5 and 5 liters per hour; the total ammonia nitrogen removal rate is not less than 93%; the total phosphorus removal rate is not less than 95%; the effluent BOD / COD ratio is not less than 0.46; the sediment recovery rate is not less than 85%; the system is equipped with a safety alarm mechanism. When any unit parameter deviates from the set range for more than 5 minutes, the central control unit will automatically trigger an audible and visual alarm and record the event log, while adjusting the relevant actuators to restore stable operation.

[0044] Combining the above parameters with the real-time prediction and intelligent control method for the entire process of wet waste anaerobic digestion slurry of the present invention, an application to a wet waste anaerobic digestion treatment system is also provided, with the following steps:

[0045] Step S1: The biogas slurry produced by the anaerobic digestion of wet waste is buffered in the biogas slurry storage unit and then transported to the nitrogen stripping unit by a transfer pump;

[0046] The biogas slurry storage unit has a tank capacity of 1,000 liters to ensure stable buffering of the generated biogas slurry.

[0047] The delivery pump is a peristaltic pump with a flow rate ranging from 0.5 to 5 liters per hour, ensuring that the biogas slurry is delivered evenly to the nitrogen stripping unit;

[0048] An integrated agitator is installed inside the storage tank to prevent solid sedimentation and ensure the homogeneity of the biogas slurry;

[0049] The liquid level sensor monitors the biogas slurry level in real time and triggers an alarm when the level falls below a set threshold.

[0050] Step S2: In the nitrogen stripping unit, ammonia nitrogen is removed by aeration, pH adjustment and temperature control, and the N / P ratio and C / N ratio in the biogas slurry are adjusted, and the pretreated biogas slurry is output to the electrochemical reaction unit.

[0051] Among them, the aeration rate of the aeration device in the nitrogen stripping unit can be adjusted to 4 liters per minute to ensure effective removal of ammonia nitrogen;

[0052] The pH adjustment module adjusts the pH of the biogas slurry to 10 by adding sodium hydroxide solution through a dosing pump.

[0053] The temperature control module stabilizes the biogas slurry temperature at 55°C with an accuracy of ±0.5°C, ensuring that the temperature is suitable for ammonia nitrogen removal.

[0054] The ammonia nitrogen removal rate can reach 90%, and the N / P ratio can be adjusted to 1.35, while the C / N ratio range is adapted to subsequent electrode reactions.

[0055] Step S3: In the electrochemical reaction unit, phosphorus is crystallized and recovered and nitrogen is deeply removed through electrode reactions, while organic matter in the biogas slurry is retained. The treated wastewater is output to the membrane separation unit, and the precipitate is output to the precipitate collection unit.

[0056] The electrochemical reaction unit adopts a magnesium-nickel composite electrode system, with the anode being a three-dimensional porous foam nickel-magnesium composite electrode, the cathode being a graphite rod, and the current density being 15 mA per square centimeter.

[0057] The pH adjustment component adjusts the pH of the biogas slurry entering the electrolyzer to 9.5;

[0058] The stirring rate of the biogas slurry in the electrolyzer is adjusted in real time according to the electrode reaction rate.

[0059] Electrochemical impedance spectroscopy is used to monitor the electrode surface to ensure that the ammonia nitrogen removal rate is not less than 40%, the total phosphorus removal rate is not less than 78%, and the COD retention rate is not less than 69%.

[0060] The electrode state during the reaction process is monitored through a dynamic passivation-activation cycle to ensure electrode efficiency.

[0061] Step S4: In the membrane separation unit, the effluent is purified and treated using membrane separation technology to output treated water that meets the carbon source standards for leachate;

[0062] Among them, the organic membrane module uses an ultrafiltration membrane with a molecular weight cutoff of 1000 Daltons to remove suspended particles and macromolecular impurities in the effluent.

[0063] The biodegradability detection module integrates BOD and COD sensors to monitor the BOD / COD ratio of the effluent in real time. When the ratio reaches 0.46, the effluent is determined to meet the carbon source standard for leachate.

[0064] The flow control module uses a variable frequency pump to regulate the flow rate of the tailwater through the organic membrane, ensuring the amount received by the subsequent leachate treatment system.

[0065] Step S5: Monitor the sensor data of each unit through the central control unit, and use the data acquisition module, predictive analysis module and execution control module to control the system operation in real time;

[0066] The data acquisition module collects data such as pH value, temperature value, ammonia nitrogen concentration value, total phosphorus concentration value, COD value, BOD / COD ratio, N / P ratio, C / N ratio, and electrolysis current value through a multi-channel data interface;

[0067] The data sampling frequency is 1 Hz, and the collected data is transmitted to the predictive analysis module via industrial Ethernet for model input and system status assessment.

[0068] Step S6: The predictive analysis module integrates an adaptive learning model that incorporates physical constraints to dynamically predict the changing trends of ammonia nitrogen removal rate, total phosphorus removal rate, and electrode efficiency, and generates control instructions.

[0069] The predictive analysis module adopts a feedforward neural network structure, including an input layer, three hidden layers and an output layer. The input data includes current value, voltage value, pH value, temperature value, ammonia nitrogen concentration value, total phosphorus concentration value, etc.

[0070] The neural network was trained using a historical dataset and a composite loss function that minimized the loss function of data error and unconstrained terms. The training iterations were 1000, the learning rate was 0.001, and the Adam optimization algorithm was used.

[0071] The loss function of a neural network consists of the sum of a data error term and a physical constraint term. The specific form of the loss function is:

[0072]

[0073] in It is the sample size. These are measured values. These are neural network predictions, where t represents the real-time running time / time-series sampling time of the electrochemical reaction unit during the anaerobic digestion of wet waste. Hyperparameters are used to balance data fitting and physical consistency. The learning rate was determined to be 0.05 using grid search and cross-validation. The optimization algorithm used was Adam, with a learning rate of 0.001 and 1000 training iterations.

[0074] This represents a physical model of electrochemical reaction rates based on the Butler-Volmer equation, which describes the relationship between current density and overpotential, as shown in the following formula:

[0075]

[0076] in It is the current density. It is the exchange current density. and It is the transmission coefficient. It is Faraday's constant. It is an overpotential. It is the gas constant. It's temperature, exchange current density. and transmission coefficient , The values ​​were obtained through calibration in a standard electrolyte using linear sweep voltammetry, and the specific values ​​were determined for the magnesium-nickel composite electrode as follows: =0.01 A / cm 2 , =0.5, =0.5; Physical constraint embedding is achieved by incorporating the gradient difference between the physical model output and the neural network prediction into the loss function, ensuring that the prediction results conform to the basic principles of electrochemistry;

[0077] During training, the network weights are updated through backpropagation, and the trained neural network is finally obtained, which can achieve high-precision prediction of ammonia nitrogen / total phosphorus removal rate.

[0078] In its working mechanism, the neural network performs forward propagation calculations, generates predicted outputs based on real-time sensor data input, and further generates optimized control commands. Simultaneously, the model integrates online learning capabilities; when the prediction error continuously exceeds a preset threshold, incremental training is automatically triggered, updating the network weights using the latest data within a sliding window to dynamically adapt to changes in the electrode surface state.

[0079] Step S7: The execution control module adaptively adjusts the power parameters and stirring rate of the electrochemical reaction unit according to the control instructions, and coordinates the operation parameters of the nitrogen stripping unit and the membrane separation unit to ensure stable processing efficiency;

[0080] The execution control module adopts a PID control algorithm to adjust parameters such as aeration rate, pH dosage, power output current, stirring rate, and flow pump speed in real time.

[0081] When pH fluctuations exceed 0.3 or temperature fluctuations exceed 3°C, an emergency adjustment procedure is initiated to restore stable system operation.

[0082] Step S8: By using real-time data and predictive analysis results, overcome the impact of the dynamic passivation-activation cycle of electrodes to ensure stable processing efficiency and maximized resource recovery throughout the entire process;

[0083] The system ensures that key monitoring indicators such as ammonia nitrogen removal rate is not less than 93%, total phosphorus removal rate is not less than 95%, effluent BOD / COD ≥ 0.46, and sediment recovery rate is not less than 85% through real-time prediction and intelligent control by the central control unit. At the same time, it ensures that the treated water meets the standards and serves as a carbon source for leachate, and that the dried sediment serves as a slow-release fertilizer, thereby realizing the high value of the output.

[0084] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0085] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0086] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A wet garbage anaerobic digestion biogas liquid full-process real-time prediction intelligent regulation method, characterized in that, The method comprises the following steps: Step S1: storing biogas slurry generated by anaerobic digestion of wet garbage through a biogas slurry storage unit, and delivering the biogas slurry to a nitrogen stripping unit through a delivery pump; Step S2: in the nitrogen stripping unit, ammonia nitrogen is removed through aeration, pH adjustment and temperature control, and the N / P ratio and C / N ratio in the biogas slurry are adjusted, and pretreated biogas slurry is output to an electrochemical reaction unit; Step S3: in the electrochemical reaction unit, crystallization recovery of phosphorus and deep removal of nitrogen are realized through electrode reaction, while organic matter in the biogas slurry is retained, treated tail water is output to a membrane separation unit, and precipitates are output to a precipitate collection unit; Step S4: in the membrane separation unit, treated tail water is purified by using membrane separation technology, and treated water meeting the leachate carbon source standard is output; Step S5: sensor data of each unit are monitored through a central control unit, and data acquisition module, prediction analysis module and execution control module are used to control system operation in real time; Step S6: in the prediction analysis module, an adaptive learning model integrating physical constraints is used to dynamically predict the change trend of ammonia nitrogen removal rate, total phosphorus removal rate and electrode efficiency, and generate control instructions; The adaptive learning model is a feedforward neural network structure, and the input features thereof include current value, voltage value, pH value, temperature value, ammonia nitrogen concentration value, total phosphorus concentration value and electrode impedance phase angle value; the training process of the feedforward neural network structure is based on a historical operation data set, and a loss function is minimized during training, and the loss function is defined as: , where N is the number of samples, is the measured value, is the neural network predicted value, t is the real-time running time / time sequence sampling time of the electrochemical reaction unit in the wet garbage anaerobic digestion biogas slurry treatment process, is the hyperparameter; represents the electrochemical reaction rate physical model based on the Butler-Volmer equation; Step S7: the execution control module adaptively adjusts power supply parameters and stirring rate of the electrochemical reaction unit according to the control instructions, and coordinates and controls operation parameters of the nitrogen stripping unit and the membrane separation unit, to ensure stable treatment efficiency; Step S8: through real-time data and prediction analysis results, the influence of electrode dynamic passivation-activation cycle is overcome, to ensure stable treatment efficiency and maximize resource recovery.

2. A system for real-time prediction and intelligent control of the whole process of wet garbage anaerobic digestion effluent based on the method of claim 1, characterized in that, The system comprises units and a central control unit; The units comprise a biogas slurry storage unit, a nitrogen stripping unit, an electrochemical reaction unit, a membrane separation unit and a precipitate collection unit, which are sequentially connected through pipelines; An output end of the biogas slurry storage unit is connected to an input end of the nitrogen stripping unit through a delivery pump, for storing biogas slurry generated by anaerobic digestion of wet garbage; An output end of the nitrogen stripping unit is connected to an input end of the electrochemical reaction unit, for removing ammonia nitrogen and adjusting N / P ratio and C / N ratio through aeration, pH adjustment and temperature control, and outputting pretreated biogas slurry to the electrochemical reaction unit; One output end of the electrochemical reaction unit is connected to the precipitate collection unit, and the other output end of the electrochemical reaction unit is connected to the precipitate collection unit, for realizing crystallization recovery of phosphorus and deep removal of nitrogen through electrode reaction, retaining organic matter in the biogas slurry, outputting treated tail water to the membrane separation unit, and outputting precipitates to the precipitate collection unit; The membrane separation unit purifies treated water meeting the leachate carbon source standard through membrane separation; The central control unit is electrically connected to sensors and actuators of each unit, and is linked with external data acquisition module, prediction analysis module and execution control module, for acquiring data in real time, performing prediction analysis and generating control instructions.

3. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The prediction analysis module is integrated with an adaptive learning model with physical constraints, which embeds electrochemical reaction kinetics equations as physical constraints into a neural network algorithm to simulate and predict the dynamic passivation-activation cycle state of the electrode surface in the electrochemical reaction unit; The prediction analysis module dynamically predicts the trends of ammonia nitrogen removal rate, total phosphorus removal rate and electrode efficiency based on real-time collected data and the output of the adaptive learning model, and generates control instructions; The execution control module adaptively adjusts the power supply parameters and stirring rate of the electrochemical reaction unit according to the control instructions, and coordinates and controls the operating parameters of the nitrogen stripping unit and the membrane separation unit.

4. The system of claim 2, wherein the prediction analysis module is integrated with an adaptive learning model with physical constraints, which embeds electrochemical reaction kinetics equations as physical constraints into a neural network algorithm to simulate and predict the dynamic passivation-activation cycle state of the electrode surface in the electrochemical reaction unit; The adaptive learning model with physical constraints in the prediction analysis module adopts a feedforward neural network structure, which includes an input layer, three hidden layers and an output layer; the input layer has 7 nodes corresponding to 7 input features; the three hidden layers have 64, 32 and 16 nodes respectively, and use ReLU activation function to process nonlinear relationships; the output layer has 3 nodes corresponding to three prediction targets of ammonia nitrogen removal rate, total phosphorus removal rate and electrode state index; the hyperparameter λ in the loss function is 0.05; and the Butler-Volmer equation is specific to: ， wherein is the current density, is the exchange current density, and is the transfer coefficient, is the Faraday constant, is the overpotential, is the gas constant, is the temperature, the exchange current density and the transfer coefficient , are obtained by calibration in standard electrolyte by linear sweep voltammetry, the specific values being determined for a magnesium-nickel composite electrode as = 0.01 A / cm 2 , = 0.5, = 0.

5.

5. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The electrochemical reaction unit includes an electrolytic tank, an electrode assembly, a power supply module, a pH adjusting assembly, a stirring device and a time control module; The electrode assembly adopts a magnesium-nickel composite electrode system, with a three-dimensional porous nickel-magnesium composite electrode as the anode and a graphite rod as the cathode; The electrode plate size is 2cm x 2cm x 0.3cm, and the magnesium loading is 10% of the mass of the nickel foam; the power module is a constant current power supply, providing a current density of 15mA / cm 2 The pH adjusting component adjusts the pH of the biogas slurry entering the electrolytic tank to 9.5; the stirring device uses a mechanical stirrer, with a rotation speed range of 100-500rpm; and the time control module controls the electrolysis time to be 6 hours.

6. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The nitrogen stripping unit includes a stripping reaction tank, an aeration device, a pH adjusting module, a temperature control module and a sampling and detection module; the aeration device uses a microporous aeration head with an aeration rate of 4 L / min; the pH adjusting module adjusts the pH of the biogas slurry to 10 by adding 20% sodium hydroxide solution through a dosing pump; the temperature control module includes an electric heating device and a platinum resistance temperature sensor to stabilize the biogas slurry temperature at 55±0.5℃; and the sampling and detection module integrates an online ammonia nitrogen analyzer and a spectrometer to monitor the ammonia nitrogen concentration, N / P ratio and C / N ratio in real time.

7. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The membrane separation unit comprises an organic membrane assembly, a biodegradability detection module and a flow control module; the organic membrane assembly adopts a polyvinylidene fluoride ultrafiltration membrane with a molecular weight cut-off of 1000 Da, a membrane area of 0.1 m 2 2, and an operating pressure of 0.1-0.5 MPa; the biodegradability detection module integrates a BOD sensor and a COD sensor, and the measurement frequency is once every 10 minutes; and the flow control module adopts a variable frequency pump, and the flow rate control precision is ±0.01 L / min.

8. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The data acquisition module includes a multi-channel data interface and a signal conditioning circuit; It supports 4-20mA current signal and 0-5V voltage signal acquisition with a sampling accuracy of 16 bits; the acquisition parameters include pH value, temperature value, ammonia nitrogen concentration value, total phosphorus concentration value, COD value, BOD / COD ratio value, N / P ratio, C / N ratio, electrolytic current value, aeration rate value, membrane separation flow rate value and electrode impedance phase angle value, with a data sampling frequency of 1 Hz.

9. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The execution control module includes a PLC controller and a driving circuit; the PLC controller controls actions including adjusting aeration rate, pH dosing amount, power output current, stirring rate and flow pump speed; and a PID control algorithm is used to trigger an alarm when the detected pH fluctuation exceeds 0.3 or the temperature fluctuation exceeds 3℃.

10. The system for real-time prediction and intelligent regulation of the whole process of wet garbage anaerobic digestion effluent according to claim 2, characterized in that: The biogas slurry storage unit comprises a polyethylene storage tank, an ultrasonic liquid level sensor and a peristaltic pump; the delivery pump has a flow range of 0.5-5 L / h and an accuracy of ±0.05 L / h; a paddle stirrer is provided, and the rotating speed is 50-100 rpm, and the paddle stirrer is operated for 2 minutes every 10 minutes; The sediment collection unit comprises a sediment tank, a screw conveyor and a drying device; the drying device is a cyclone hot air dryer, and the hot air temperature is 80 DEG C; a weighing sensor is integrated to monitor the recovery rate in real time.

Images