Separation system for high-efficiency anaerobic reactors

Abstract

This invention provides a novel separation system for high-efficiency anaerobic reactors, belonging to the field of separation technology for anaerobic reactors. The system includes a solid-liquid preliminary separation module, a biogas purification and impurity separation module, a data acquisition and transmission module, a data analysis and processing module, a fusion algorithm control module, a decision-making and instruction generation module, and an actuator module. This invention aims to overcome the shortcomings of existing technologies and provide a high-efficiency separation system that integrates multiple modules working collaboratively and possesses deep algorithm fusion characteristics.

Description

Technical Field

[0001] This invention relates to the field of separation technology for anaerobic reactors, and in particular to a novel separation system for high-efficiency anaerobic reactors. Background Technology

[0002] Traditional anaerobic reactor separation systems suffer from problems such as low separation efficiency, easy clogging, and unstable operation. Existing technologies mostly employ physical sedimentation or simple filtration for solid-liquid separation, which is insufficient for thoroughly removing impurities from biogas and fails to meet the requirements of high-efficiency anaerobic reactions. Furthermore, even when multiple algorithms are used in existing systems, they are often simply superimposed without data interaction and collaborative processing mechanisms, failing to deeply mine the value of the data. This results in imprecise adjustments to system operating parameters and limited overall performance. Summary of the Invention

[0003] This invention provides a novel separation system for high-efficiency anaerobic reactors, aiming to overcome the shortcomings of existing technologies and provide a high-efficiency separation system that integrates multiple modules working collaboratively and has deep algorithm fusion characteristics.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A novel separation system for high-efficiency anaerobic reactors includes a solid-liquid preliminary separation module, a biogas purification and impurity separation module, a data acquisition and transmission module, a data analysis and processing module, a fusion algorithm control module, a decision and instruction generation module, and an actuator module.

[0006] The solid-liquid preliminary separation module is used to perform solid-liquid preliminary separation on the mixed liquid output from the anaerobic reactor;

[0007] The biogas purification and impurity separation module is used to purify and separate impurities from the biogas discharged from the anaerobic reactor.

[0008] The data acquisition and transmission module is used to collect data from the solid-liquid preliminary separation module and the biogas purification and impurity separation module, as well as system operation data, and transmit them to the data analysis and processing module.

[0009] The data analysis and processing module is used to preprocess the data, extract key feature information, and transmit it to the fusion algorithm control module.

[0010] The fusion algorithm control module uses an adaptive control algorithm as its core, and integrates fuzzy optimization, surface fitting and random forest techniques to process the received data, generate parameter adjustment instructions and transmit them to the decision and instruction generation module.

[0011] The decision-making and instruction generation module integrates parameter adjustment instructions and the actual system operation to generate the final control instruction and transmit it to the actuator module.

[0012] The actuator module controls the operating status of the relevant equipment in the solid-liquid preliminary separation module and the biogas purification and impurity separation module according to the final control command.

[0013] In this specification, the solid-liquid preliminary separation module adopts a combination of hydrocyclone separation and membrane filtration. The mixed liquid first enters the hydrocyclone separator to separate particulate solids, and the liquid after preliminary separation then enters the membrane filtration unit to retain suspended solids.

[0014] In this specification, the biogas purification and impurity separation module includes a water washing tower and an activated carbon adsorption tower. The biogas first enters the water washing tower to remove water-soluble impurities, and then passes through the activated carbon adsorption tower to adsorb organic impurities and odor substances.

[0015] In this specification, the fuzzy optimization algorithm in the fusion algorithm control module performs preliminary optimization of the parameters of the adaptive control algorithm based on the pressure difference data of the solid-liquid preliminary separation module, the biogas component concentration data of the biogas purification and impurity separation module, and the system operation data, through a fuzzy rule base.

[0016] In this specification, the surface fitting technology in the fusion algorithm control module performs multidimensional analysis on the pressure difference data of the solid-liquid preliminary separation module and the biogas component concentration data of the biogas purification and impurity separation module, both historical and real-time, to construct a data surface model, explore the nonlinear relationship between system operating parameters and separation effect, and provide parameter adjustment direction for the adaptive control algorithm.

[0017] In this specification, the random forest technology in the fusion algorithm control module is trained and learned based on a large amount of historical operating data and real-time monitoring data to predict the solid-liquid separation effect and biogas composition change trend under different combinations of operating parameters, and the prediction results are provided to the adaptive control algorithm for parameter adjustment.

[0018] In this specification, the data preprocessing operations of the data analysis and processing module include data cleaning, filtering, integration, extraction of key feature information, removal of abnormal data, and standardization of data format.

[0019] In this specification, the decision and instruction generation module dynamically adjusts the weights of the output results of each algorithm in the fusion algorithm control module based on reinforcement learning, and comprehensively generates the final control instruction.

[0020] In this specification, the solid-liquid preliminary separation module controlled by the actuator module includes a water pump, valves, and a cyclone separator motor, while the biogas purification and impurity separation module controlled by the actuator module includes inlet and outlet valves of the water washing tower and a switching device for the activated carbon adsorption tower.

[0021] In this specification, the data of the solid-liquid preliminary separation module and the biogas purification and impurity separation module, as well as the system operation data, include the pressure difference of the membrane filtration unit of the solid-liquid preliminary separation module, the inlet flow rate of the hydrocyclone separator, the turbidity of the filtrate of the membrane filtration unit, the filtration temperature of the membrane filtration unit, the separation time of the hydrocyclone separator, the methane concentration at the outlet of the activated carbon adsorption tower in the biogas purification and impurity separation module, the liquid level of the water washing tower, the temperature inside the activated carbon adsorption tower, the washing time of the water washing tower, the service life of the activated carbon, the temperature inside the anaerobic reactor, the system operating time, the current of the water pump in the actuator module, the energy consumption of the entire system, and the pH value inside the anaerobic reactor.

[0022] In summary, the present invention has at least the following beneficial effects:

[0023] Breakthrough in High-Efficiency Separation Performance: This novel separation system achieves highly efficient separation of solid, liquid, and gas phases through the collaborative operation of multiple modules. In the preliminary solid-liquid separation module, a combination of cyclone separation and membrane filtration first utilizes the centrifugal force of the cyclone separator to remove larger solid particles, and then the membrane filtration unit retains small suspended solids. Compared to traditional single physical sedimentation methods, this improves solid-liquid separation efficiency, effectively reduces the suspended solids content in the liquid, and lowers the burden on subsequent processing. In the biogas purification and impurity separation module, the combination of a water washing tower and an activated carbon adsorption tower provides dual purification of water-soluble and organic impurities in the biogas, increasing the methane concentration in the biogas and significantly improving the gas production quality of the anaerobic reactor, meeting higher standards of energy utilization.

[0024] Advantages of Deep Integration of Intelligent Algorithms: The system incorporates surface fitting and random forest technologies, deeply integrating them with adaptive control and fuzzy optimization algorithms. The resulting fusion algorithm control module enables precise dynamic adjustment of system operating parameters. Surface fitting technology, by constructing a multi-dimensional data surface model, deeply explores the nonlinear relationships between solid-liquid separation efficiency, biogas purification effect, and various operating parameters. For example, it can accurately analyze the impact of parameter combinations such as membrane filtration pressure and hydrocyclone speed on solid-liquid separation efficiency, providing a scientific theoretical basis for parameter adjustment. Random forest technology, trained on a large amount of historical and real-time data, can accurately predict the solid-liquid separation effect and biogas composition trends under different parameter combinations. For instance, it can predict the fluctuations in biogas methane concentration after adjusting the water level in the washing tower and the activated carbon adsorption time, giving the system a forward-looking capability. These algorithms work together to improve the accuracy and stability of system operating parameter adjustments compared to traditional simple superposition algorithms, significantly reducing the frequency and cost of manual intervention and achieving intelligent and automated operation of the separation system.

[0025] System architecture design advantages: Modular design and closed-loop data control are key innovations of this system. Each functional module has a clear division of labor and works closely together. Data acquisition and transmission modules, data analysis and processing modules, etc., provide accurate data support for the algorithm module. The processing results from the algorithm module are transmitted to the execution module via the decision and instruction generation module, forming a complete data closed loop. This architecture makes the system structure clear, facilitates maintenance and upgrades, and allows for rapid location and handling of faults or optimizations in any module, reducing downtime.

[0026] Significant economic and environmental benefits: The efficient separation performance and intelligent operation bring substantial economic benefits. Improved biogas quality increases its value as an energy source, allowing for direct use in power generation and heating, improving energy recovery and utilization rates, and increasing enterprise profits. Enhanced operational stability reduces equipment failure frequency and maintenance costs, minimizing production losses due to equipment downtime. From an environmental perspective, the system's efficient separation of the solid, liquid, and gas phases reduces pollutant emissions, such as lowering the pollutant content in the separated liquid and preventing secondary pollution. High-quality biogas energy replaces traditional fossil fuels, reducing greenhouse gas emissions and contributing to carbon reduction goals, aligning with sustainable development principles and possessing significant social and environmental importance.

[0027] Outstanding technological innovation: The deep integration of multiple advanced algorithms and technologies has enabled this system to achieve a major technological breakthrough. Most existing anaerobic reactor separation systems lack the deep collaborative application of algorithms, while this system innovatively applies technologies such as surface fitting and random forests to optimize the separation process. Through data sharing and collaborative processing among algorithms, a new fusion algorithm has been formed, solving the problems of inaccurate parameter adjustment and lack of foresight in traditional systems. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of a novel separation system for a high-efficiency anaerobic reactor involved in this invention.

[0030] Figure 2 This is a schematic diagram of the algorithm model construction process involved in this invention.

[0031] Figure 3 This is a schematic diagram of the algorithm fusion and training process involved in this invention.

[0032] Figure 4 This is a schematic diagram of the algorithm application process involved in this invention. Detailed Implementation

[0033] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0034] The following disclosure provides many different implementations or examples for carrying out different structures of the embodiments of the present invention. To simplify the disclosure of the embodiments of the present invention, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the embodiments of the present invention. Furthermore, reference numerals and / or reference letters may be repeated in different examples of the embodiments of the present invention; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various implementations and / or arrangements discussed.

[0035] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0036] like Figure 1 As shown, this embodiment provides a novel separation system for a high-efficiency anaerobic reactor, including a solid-liquid preliminary separation module, a biogas purification and impurity separation module, a data acquisition and transmission module, a data analysis and processing module, a fusion algorithm control module, a decision and instruction generation module, and an actuator module;

[0037] The solid-liquid preliminary separation module is used to perform solid-liquid preliminary separation on the mixed liquid output from the anaerobic reactor;

[0038] The biogas purification and impurity separation module is used to purify and separate impurities from the biogas discharged from the anaerobic reactor.

[0039] The data acquisition and transmission module is used to collect data from the solid-liquid preliminary separation module and the biogas purification and impurity separation module, as well as system operation data, and transmit them to the data analysis and processing module.

[0040] The data analysis and processing module is used to preprocess the data, extract key feature information, and transmit it to the fusion algorithm control module.

[0041] The fusion algorithm control module uses an adaptive control algorithm as its core, and integrates fuzzy optimization, surface fitting and random forest techniques to process the received data, generate parameter adjustment instructions and transmit them to the decision and instruction generation module.

[0042] The decision-making and instruction generation module integrates parameter adjustment instructions and the actual system operation to generate the final control instruction and transmit it to the actuator module.

[0043] The actuator module controls the operating status of the relevant equipment in the solid-liquid preliminary separation module and the biogas purification and impurity separation module according to the final control command.

[0044] In some embodiments, the solid-liquid preliminary separation module adopts a combination of hydrocyclone separation and membrane filtration. The mixture first enters the hydrocyclone separator to separate particulate solids, and the liquid after preliminary separation then enters the membrane filtration unit to retain suspended solids.

[0045] In some embodiments, the biogas purification and impurity separation module includes a water washing tower and an activated carbon adsorption tower. The biogas first enters the water washing tower to remove water-soluble impurities, and then passes through the activated carbon adsorption tower to adsorb organic impurities and odor substances.

[0046] In some embodiments, the fuzzy optimization algorithm in the fusion algorithm control module performs preliminary optimization of the parameters of the adaptive control algorithm based on the pressure difference data of the solid-liquid preliminary separation module, the biogas component concentration data of the biogas purification and impurity separation module, and the system operation data, using a fuzzy rule base.

[0047] In some embodiments, the surface fitting technology in the fusion algorithm control module performs multidimensional analysis on the pressure difference data of the solid-liquid preliminary separation module and the biogas component concentration data of the biogas purification and impurity separation module, both historical and real-time, to construct a data surface model, explore the nonlinear relationship between system operating parameters and separation effect, and provide parameter adjustment direction for the adaptive control algorithm.

[0048] In some embodiments, the random forest technique in the fusion algorithm control module is trained and learned based on a large amount of historical operating data and real-time monitoring data to predict the solid-liquid separation effect and biogas composition change trend under different combinations of operating parameters, and the prediction results are provided to the adaptive control algorithm for parameter adjustment.

[0049] In some embodiments, the data preprocessing operations of the data analysis and processing module include data cleaning, filtering, integration, extraction of key feature information, removal of abnormal data, and standardization of data format.

[0050] In some embodiments, the decision and instruction generation module dynamically adjusts the weights of the output results of each algorithm in the fusion algorithm control module based on reinforcement learning, and generates the final control instruction in a comprehensive manner.

[0051] In some embodiments, the solid-liquid preliminary separation module controlled by the actuator module includes a water pump, valves, and a cyclone separator motor, while the biogas purification and impurity separation module controlled by the actuator module includes a water washing tower inlet and outlet valve and an activated carbon adsorption tower switching device.

[0052] In some embodiments, the data from the solid-liquid preliminary separation module and the biogas purification and impurity separation module, as well as the system operation data, include the pressure difference of the membrane filtration unit in the solid-liquid preliminary separation module, the inlet flow rate of the hydrocyclone separator, the turbidity of the filtrate in the membrane filtration unit, the filtration temperature of the membrane filtration unit, the separation time of the hydrocyclone separator, the methane concentration at the outlet of the activated carbon adsorption tower in the biogas purification and impurity separation module, the liquid level in the water washing tower, the temperature inside the activated carbon adsorption tower, the washing time in the water washing tower, the service life of the activated carbon, the temperature inside the anaerobic reactor, the system operating time, the current of the water pump in the actuator module, the energy consumption of the entire system, and the pH value inside the anaerobic reactor.

[0053] The technical concept of this invention is as follows:

[0054] This novel separation system aims to overcome the shortcomings of existing technologies and provide a highly efficient separation system that integrates multiple modules working collaboratively and possesses deep algorithm fusion characteristics. Specifically, it includes the following main modules:

[0055] The initial solid-liquid separation module employs a combination of hydrocyclone separation and membrane filtration. The mixture output from the reactor first enters the hydrocyclone separator, where centrifugal force separates larger solid particles. The pre-separated liquid then enters the membrane filtration unit to further retain smaller suspended solids. This module uses a pressure sensor to monitor the pressure difference across the membrane in real time and transmits the pressure difference data to the data acquisition and transmission module.

[0056] Biogas purification and impurity separation module: Biogas discharged from the anaerobic reactor first enters a water washing tower to remove some water-soluble impurities, and then passes through an activated carbon adsorption tower for further adsorption and removal of organic impurities and odor substances. A gas composition sensor is installed at the outlet of the activated carbon adsorption tower to detect the concentration of gases such as methane, carbon dioxide, and hydrogen sulfide in the biogas in real time, and transmits the detection data to the data acquisition and transmission module.

[0057] Data acquisition and transmission module: responsible for collecting pressure difference data from the solid-liquid preliminary separation module, gas composition data from the biogas purification and impurity separation module, and relevant operating data generated by other modules, and accurately and quickly transmitting this data to the data analysis and processing module.

[0058] Data Analysis and Processing Module: Receives data from the data acquisition and transmission module, performs preprocessing operations such as cleaning, filtering, and integration, and extracts key feature information to provide basic data for subsequent algorithm fusion processing. The processed data is then transmitted to the fusion algorithm control module.

[0059] The integrated algorithm control module uses an adaptive control algorithm as its core framework, deeply integrating fuzzy optimization, surface fitting, and random forest techniques. During operation, the fuzzy optimization algorithm first performs preliminary optimization of the adaptive control algorithm's parameters, adjusting the approximate range of parameters based on the system's fuzzy rules and experience. Surface fitting technology performs multi-dimensional analysis of historical and real-time data, constructing a data surface model to uncover the nonlinear relationships between solid-liquid separation efficiency, biogas purification effect, and various operating parameters, providing a more precise direction for parameter adjustment in the adaptive control algorithm. The random forest prediction module, trained on a large amount of data, predicts the system's future operating state and separation effect under different parameter combinations. The adaptive control algorithm integrates the preliminary adjustment results of fuzzy optimization, the relationship analysis of surface fitting, and the prediction information of random forest to dynamically generate precise parameter adjustment commands, which are then transmitted to the decision and command generation module.

[0060] Decision and instruction generation module: Receives parameter adjustment instructions from the fusion algorithm control module, combines them with the current actual operating conditions of the system, makes comprehensive decisions, generates precise control instructions, and transmits them to the actuator module.

[0061] The actuator module precisely controls the operation of the water pumps, valves, and cyclone separator motors of the solid-liquid preliminary separation module, as well as the inlet and outlet valves of the water washing tower and the switching device of the activated carbon adsorption tower in the biogas purification and impurity separation module, based on the instructions from the decision and instruction generation module, thereby achieving automated operation of the separation system.

[0062] The connections between the modules are as follows: the mixed liquor outlet of the anaerobic reactor is connected to the inlet of the solid-liquid preliminary separation module; the liquid outlet of the solid-liquid preliminary separation module discharges the treated liquid, and the solid outlet discharges the separated solid matter; the biogas outlet of the anaerobic reactor is connected to the inlet of the biogas purification and impurity separation module; the outlet of the biogas purification and impurity separation module outputs purified biogas; the solid-liquid preliminary separation module and the biogas purification and impurity separation module are connected to the data acquisition and transmission module via data transmission lines; the data acquisition and transmission module is connected to the data analysis and processing module; the data analysis and processing module is connected to the fusion algorithm control module; the fusion algorithm control module is connected to the decision and instruction generation module; the decision and instruction generation module is connected to the actuator module via control signal lines.

[0063] Regarding data transmission, the solid-liquid preliminary separation module and the biogas purification and impurity separation module transmit monitoring data to the data acquisition and transmission module. The data is then transmitted to the data analysis and processing module for preprocessing. The processed data enters the fusion algorithm control module, where multiple algorithms are integrated to process the data. The generated results are transmitted to the decision-making and instruction generation module, and finally, the execution module executes the instructions, forming a complete data closed loop. This ensures efficient data utilization at each stage and enables dynamic optimization and efficient and stable operation of the system.

[0064] In summary, all hardware components are existing equipment. Common models of hydrocyclones, such as the Hydro-Cyclone series, membrane filtration units, such as PVDF hollow fiber ultrafiltration membranes, and turbidity sensors, such as the HACH 2100Q portable turbidity meter, are used. Other sensors, such as pressure sensors (pressure transmitters), flow meters, water washing towers, activated carbon adsorption towers, and gas composition sensors, as well as various specific actuators, will not be listed here; existing equipment can be used. The focus of this invention is on the collaborative work of the various modules and the innovative algorithm fusion within the fusion algorithm control module.

[0065] In some embodiments, the fusion algorithm model implementation process in the fusion algorithm control module is as follows:

[0066] I. Algorithm Model Construction Process (e.g.) Figure 2 (As shown)

[0067] 1.1 Construction of Multimodal Feature Space

[0068] Constructing the feature space for anaerobic reactor separation systems , where each sample feature vector It consists of the following specific features:

[0069] Solid-liquid separation characteristics:

[0070] The pressure difference in the membrane filtration unit of the solid-liquid preliminary separation module is collected in real time by a pressure sensor within the module. An increase in pressure difference indicates an increase in solid particles adhering to the membrane surface, leading to increased filtration resistance and potential clogging, directly impacting the solid-liquid separation efficiency.

[0071] The inlet flow rate of the hydrocyclone separator is measured by a flow meter. The flow rate affects the centrifugal force, which in turn affects the separation effect of solid particles. Too high or too low a flow rate will result in insufficient separation.

[0072] The turbidity of the filtrate from the membrane filtration unit is detected by a turbidity sensor. Turbidity reflects the content of residual suspended solids in the liquid after solid-liquid separation; higher turbidity indicates poorer separation efficiency.

[0073] The filtration temperature of the membrane filtration unit affects the viscosity of the liquid, which in turn affects the filtration speed and separation effect.

[0074] The separation time of a hydrocyclone separator is crucial; insufficient separation time will result in incomplete separation of solid particles.

[0075] Characteristics related to biogas purification:

[0076] The methane concentration at the outlet of the activated carbon adsorption tower in the biogas purification and impurity separation module is monitored by a gas composition sensor. Methane concentration is a key indicator for measuring biogas quality; the higher the concentration, the greater the value of biogas as an energy source.

[0077] The liquid level in the water scrubbing tower is obtained by a liquid level sensor. The liquid level affects the scrubbing effect of the water scrubbing tower on water-soluble impurities in biogas. If the liquid level is too low, the scrubbing may be insufficient, while if the liquid level is too high, it may affect the flow of biogas.

[0078] The temperature inside the activated carbon adsorption tower is collected by a temperature sensor. Temperature changes affect the adsorption capacity of activated carbon; excessively high or low temperatures can reduce adsorption efficiency.

[0079] The washing time of the water washing tower is crucial; too short a washing time will not be sufficient to remove water-soluble impurities.

[0080] The usage cycle of activated carbon is important; using it for too long will reduce its adsorption effect.

[0081] System operating status characteristics:

[0082] The temperature within the anaerobic reactor plays a decisive role in the activity and reaction rate of anaerobic microorganisms. A suitable temperature range ensures that microorganisms efficiently decompose organic matter and produce biogas.

[0083] System running time is used to analyze the changing patterns of system parameters at different operating stages, such as the differences in parameters between the initial equipment break-in stage and the stable operation stage.

[0084] The current of the water pump in the actuator module reflects the pump's workload and operating status. Abnormal fluctuations in current may indicate a pump malfunction or abnormal operation.

[0085] Energy consumption of the entire system, used to evaluate the economic efficiency of system operation.

[0086] The pH value inside the anaerobic reactor affects the growth and metabolism of anaerobic microorganisms.

[0087] 1.2 Basic Framework of Adaptive Control

[0088] The Model Reference Adaptive Control (MRAC) architecture is adopted, and the core equations are:

[0089] ;

[0090] in: To control the output, specific commands include those for the membrane filtration pressure adjustment module, the hydrocyclone speed control module, the filtration temperature adjustment module, and the separation time control module; as well as commands for the water level adjustment module, the activated carbon adsorption time control module, the water washing time control module, and the activated carbon replacement reminder module; and commands for adjusting the power of the water pump in the actuator. These commands directly affect each module, adjusting its operating status to achieve efficient separation.

[0091] As an adaptive gain matrix, it dynamically adjusts the weights of each feature's influence on the control output based on the system's real-time operating status. For example, when the membrane filtration pressure difference is large, its corresponding weight is increased, causing the system to prioritize adjusting the membrane filtration pressure.

[0092] This is a compensation term used to correct the deviation between the model and the actual system, thereby improving the accuracy of control.

[0093] It is the vector that includes specific features such as solid-liquid separation, biogas purification, and system operating status.

[0094] 1.3 Fuzzy Optimized Embedding

[0095] Designing a fuzzy inference system Make dynamic adjustments:

[0096] Fuzzy rule base: Establishment Fuzzy rules .

[0097] Rules related to solid-liquid separation:

[0098] If the pressure difference of the membrane filtration unit ( "High" and cyclone separator inlet flow rate ( If the flow rate is set to "low", the weight of the membrane filtration pressure regulation command will be increased, while the weight of the cyclone separator speed control command will be decreased. This is because a high pressure difference indicates severe membrane clogging, in which case the membrane filtration pressure should be adjusted first. On the other hand, a low flow rate may indicate that the separation effect is less affected by the flow rate, so the weight of the speed control command will be decreased.

[0099] If the pressure difference of the membrane filtration unit ( "High" and the filtration temperature of the membrane filtration unit ( If the value is "low", the weight of the membrane filtration pressure regulation command will be increased, and the weight of the filtration temperature regulation command will be appropriately increased, because low temperature may increase the liquid viscosity, affecting the filtration speed and thus increasing the pressure difference.

[0100] If the separation time of the cyclone separator ( "Short" and cyclone separator inlet flow rate ( If the value is set to "high", the weight of the separation time control command of the cyclone separator will be increased, and the weight of the inlet flow control command will be reduced to avoid insufficient separation due to excessive flow or insufficient separation time.

[0101] Regulations related to biogas purification:

[0102] If the methane concentration at the outlet of the activated carbon adsorption tower ( "Low" and the water washing tower liquid level height ( If the value is set to "low", then the weighting of the water level adjustment command and the activated carbon adsorption time control command will be increased. A low methane concentration indicates poor purification effect, while a low water level in the water washing tower may lead to insufficient washing. Therefore, both parameters need to be adjusted simultaneously.

[0103] If the temperature inside the activated carbon adsorption tower ( "High" and the service life of activated carbon ( If the usage period is "long", an activated carbon replacement prompt will be issued, and the weight of the activated carbon adsorption time control command will be appropriately reduced, because high temperature and long usage period will reduce the adsorption capacity of activated carbon.

[0104] If the washing time of the water scrubbing tower ( "Short" and the water washing tower liquid level height ( If the setting is "high", then the weight of the washing time control command of the water washing tower will be increased, and the weight of the water level adjustment command will be appropriately reduced to ensure that the biogas is fully washed when the water level is appropriate.

[0105] System running status related rules:

[0106] If the temperature inside the anaerobic reactor ( "High" and system energy consumption ( If the value is "high", the weight of the water pump power adjustment command will be reduced, and the operating status of other equipment will be checked at the same time, because high temperature and high energy consumption may indicate abnormal equipment operation or unreasonable parameter settings.

[0107] If the current of the water pump in the actuator module ( "Abnormal fluctuations" and system uptime ( If the signal is "long", a water pump maintenance prompt will be issued, and the water pump power regulation command will be adjusted appropriately to prevent water pump failure from affecting system operation.

[0108] If the pH value in the anaerobic reactor ( "Deviating from the suitable range" and the temperature inside the anaerobic reactor ( If the pH value is "normal", then adjust the instructions for adding relevant chemical reagents according to the deviation of the pH value to maintain a suitable reaction environment.

[0109] Among them, fuzzy sets such as "high", "low", "short", "long", "abnormal fluctuation", and "deviation from appropriate range" divide the numerical range of each feature and determine the membership function based on experience and experimental data.

[0110] Fuzzy reasoning: A product inference engine and the centroid method are used for defuzzification to finally obtain the adjusted result. .

[0111] Fuzzy rules and membership degree calculation:

[0112] Assume there is Fuzzy rules The influence of membrane filtration pressure difference and hydrocyclone inlet flow rate on solid-liquid separation. Taking the relevant weight adjustment as an example, the rules In the form of:

[0113] If the pressure difference of the membrane filtration unit ( )yes And the inlet flow rate of the cyclone separator ( )yes ,but ,in, This is a fuzzy set, such as "high" or "low". For the input real-time data... and Calculate its membership degree in each fuzzy set. .

[0114] Indicates the first Under fuzzy rules, the weighting coefficient of the membrane filtration pressure regulation command. This coefficient is used to adjust the pressure control command of the membrane filtration unit in the solid-liquid preliminary separation module, and its value range is usually [0,1] or set according to actual control requirements. Mechanism of action: When the membrane filtration pressure difference... Satisfying fuzzy sets (e.g., "high" or "low") and the inlet flow rate of the cyclone separator satisfy hour, This determines the priority and impact of the membrane filtration pressure regulation command in the adaptive control output under the current operating conditions. For example: if "High pressure difference" If the data usage is "low", then... It will be set to a larger value (such as 0.8) so that the system will prioritize increasing the membrane filtration pressure to alleviate clogging.

[0115] Indicates the first The weighting coefficient of the cyclone separator speed control command under the given fuzzy rules. This coefficient is used to adjust the speed control command of the cyclone separator, and its value range is also set based on the control requirements (e.g., [0,1]). Mechanism of action: When... and When the corresponding fuzzy set is satisfied, This determines the priority of cyclone separator speed regulation. For example: if "Pressure difference is normal" For "high traffic", then It will be set to a larger value (such as 0.7) so that the system can enhance the centrifugal separation effect by increasing the rotation speed and avoid insufficient separation caused by high flow rate.

[0116] and The specific values ​​need to be determined based on the operating characteristics and experimental data of the anaerobic reactor separation system, following the principles of: operating condition matching:

[0117] For different combinations of pressure differentials and flow rates (such as "high pressure differential + high flow rate" or "low pressure differential + low flow rate"), corresponding settings are configured based on historical data or expert experience. and This ensures that the weights are consistent with the optimal control strategy under actual operating conditions. Normalization constraint:

[0118] In the same rule, and Normalization conditions must be met (e.g.) + =1), to ensure the rationality of weight allocation. For example, when the system needs to prioritize adjusting the membrane pressure, Increase Correspondingly reduced to avoid control command conflicts. Dynamic adjustability:

[0119] Activation weighting through fuzzy reasoning (e.g.) ) and It will dynamically combine based on real-time operating conditions to form the final control weights, thereby achieving adaptive optimization of the separation system.

[0120] For example: Suppose there are two fuzzy rules: Rule 1 ( =1):

[0121] like It is "high" ( )and It is "low" ( ),but Meaning: When the membrane pressure differential is high and the flow rate is low, the membrane pressure regulation weight... Speed ​​adjustment weight Prioritize addressing membrane blockage issues.

[0122] Rule 2 ( =2):

[0123] like It is “middle” ( )and It is "high" ( ),but Meaning: When the pressure difference is moderate and the flow rate is high, the membrane pressure regulation weight... Speed ​​adjustment weight The separation efficiency under high flow rate can be improved by increasing the rotation speed.

[0124] By calculating the activation degree of each rule in real time The final weighted result It integrates weight suggestions from multiple rules to achieve precise control over system parameters.

[0125] Assuming a Gaussian membership function is used:

[0126] ;

[0127] in, It is the first Rule number 1 A fuzzy set The central value, It is its width parameter.

[0128] Product inference engine calculation rule activation:

[0129] The product inference engine calculates the activation degree of each rule. This is used to measure the applicability of the rule under the current input. The formula for calculating rule activation is:

[0130] ;

[0131] here, It is the number of features participating in that rule, in the example above. =2. According to the rules For example, its activation level This indicates the degree to which the rule is met under the current membrane filtration pressure differential and cyclone separator inlet flow rate.

[0132] Calculate the weighted average output:

[0133] Based on the activation level of each rule Gain vector corresponding to the rule The centroid method is used to defuzzify and calculate the adjusted value. :

[0134] ;

[0135] by The first element Taking the weight of the corresponding membrane filtration pressure regulation command as an example, the specific calculation process is as follows:

[0136] ;

[0137] in, It is the first Gain vector corresponding to each rule The first element in the formula represents the weight of the membrane filtration pressure regulation command under this rule. Similarly, the formula can be calculated... The values ​​of other elements are used to obtain the complete adjusted adaptive gain matrix. It is used by adaptive control algorithms to adjust system parameters.

[0138] 1.4 Surface Fitting Modeling

[0139] Constructing a multidimensional surface model using tensor product spline functions:

[0140] ;

[0141] The input feature vector specifically includes characteristic data reflecting the operating status of the anaerobic reactor separation system, such as the pressure difference of the membrane filtration unit in the solid-liquid preliminary separation module, the inlet flow rate of the hydrocyclone separator, and the methane concentration at the outlet of the activated carbon adsorption tower. : Feature vector The dimension, i.e., the number of features; The index variable for summation, used to iterate through different combinations of basis functions; These represent the number of basis functions in the 1st, 2nd, ..., m1th dimensions, respectively. Tensor product spline coefficients are parameters of the model; : indicates the first The first dimension One-dimensional B-spline basis functions.

[0142] By analyzing the membrane filtration pressure difference in historical data ( ), cyclone separator inlet flow rate ( ), filtration temperature of the membrane filtration unit ( ), separation time of the hydrocyclone separator ( The relationship between solid-liquid separation characteristics and solid-liquid separation efficiency was fitted; and the methane concentration at the outlet of the activated carbon adsorption tower was fitted. ), water washing tower liquid level height ( ), temperature inside the activated carbon adsorption tower ( ), washing time of the water washing tower ( ), the service life of activated carbon ( The relationship between biogas purification-related characteristics and biogas purification efficiency was fitted. For example, the study analyzed how membrane filtration pressure difference, filtration temperature, and separation time jointly affect solid-liquid separation efficiency, identifying the potential patterns of parameter changes on separation efficiency and providing a theoretical basis for adjusting system operating parameters. Simultaneously, the relationships between the operating state characteristics of each system and their indirect impacts on solid-liquid separation and biogas purification processes were analyzed, providing support for overall system optimization.

[0143] 1.5 Random Forest Prediction Model

[0144] A random forest containing T decision trees is constructed. Historical operational data is used as training samples, including combinations of parameters such as membrane filtration pressure, hydrocyclone speed, filtration temperature, separation time, water level in the scrubbing tower, activated carbon adsorption time, scrubbing tower washing time, activated carbon usage cycle, and water pump power, along with corresponding solid-liquid separation efficiencies (e.g., filtrate turbidity), methane concentration in biogas, system energy consumption, and equipment failure probability. Through training, the random forest can predict future solid-liquid separation performance (e.g., trends in filtrate turbidity), biogas composition trends (e.g., methane concentration fluctuations), system energy consumption changes, and potential equipment failure times under different combinations of operating parameters. This helps the system plan parameter adjustment strategies and equipment maintenance schedules in advance.

[0145] II. Algorithm Fusion and Training Process (e.g.) Figure 3 (As shown)

[0146] 2.1 Feature Importance Fusion

[0147] The Analytic Hierarchy Process (AHP) was used to determine the importance weight of each feature in the anaerobic reactor separation system:

[0148] Construct a judgment matrix ,element Indicates the first The feature relative to the first A comparison of the degree of influence of each feature on the overall system performance. This indicates the number of features involved in the comparison. Pairwise comparisons are made between features related to solid-liquid separation, biogas purification, and system operating status. For example, comparing the pressure difference of the membrane filtration unit (…). ) and methane concentration at the outlet of the activated carbon adsorption tower ( The degree of impact on the overall system performance depends on whether the pressure difference has a greater impact on the solid-liquid separation effect, while the methane concentration has a critical impact on biogas quality. The determination should be based on the relative importance of the two factors. and The value; then compare the pressure difference of the membrane filtration unit ( ) and the temperature inside the anaerobic reactor ( Because temperature affects the entire anaerobic reaction process, it indirectly affects solid-liquid separation and biogas purification. If the effect of temperature is relatively greater, then the corresponding... and The value of is then calculated. This process is repeated to compare all features and construct a complete judgment matrix.

[0149] Calculate the feature weight vector ,satisfy In this way, the influence weight of each feature on system operation and decision-making in the algorithm is clearly defined, making the algorithm focus more on key features.

[0150] 2.2 Integrated Learning Framework

[0151] Design a three-tier converged architecture:

[0152] Bottom-level fusion: Feature-level fusion combines the original data according to their importance weights. Priority is given to features that have a significant impact on separation performance and system performance, such as membrane filtration pressure difference. ), methane concentration at the outlet of the activated carbon adsorption tower ( ), temperature inside the anaerobic reactor ( The fused feature vector is obtained through weighted summation and other methods. This approach highlights key information and reduces the interference of redundant data on the algorithm. Furthermore, for highly correlated features, such as the pressure difference of the membrane filtration unit and the turbidity of the filtrate, dimensionality reduction techniques such as principal component analysis can be performed to further optimize the feature vectors.

[0153] Mid-level fusion: Algorithm layer fusion, where intermediate results are output after parallel processing of each algorithm.

[0154] Adaptive control output: Based on the fusion feature vector, preliminary adjustment instructions for the operating parameters of the solid-liquid separation and biogas purification modules are calculated, including adjustments to parameters such as membrane filtration pressure, hydrocyclone speed, filtration temperature, separation time, water level in the washing tower, activated carbon adsorption time, washing tower washing time, activated carbon replacement, and water pump power. To fuse the feature vector, it is composed of multimodal features combined according to their importance weights, that is: ,in For original features (such as) Membrane filtration pressure difference, methane concentration, etc. (characteristics) The feature weights calculated for the Analytic Hierarchy Process (AHP) ).

[0155] Fuzzy optimization output: Based on fuzzy rules, the adaptive control parameters are optimized. The parameter adjustment commands of the adaptive control output are refined and corrected according to different operating conditions and characteristic combinations to better match actual operating conditions. For example, in the initial stage and stable period of system operation, the adjustment amplitude and direction of the adaptive control parameters are optimized according to different characteristic states. The fuzzy gain matrix has dimensions of and . Same, generated by a fuzzy inference system. Its elements Yes The correction factor is calculated as follows: ,in Determined by a fuzzy rule base. For example, when the membrane filtration pressure difference... When in the "extremely high" range, This means enhancing the sensitivity of membrane filtration pressure regulation.

[0156] Surface fitting output: Based on the surface model analysis, the relationship between current parameters and separation effect is analyzed, and parameter adjustment suggestions are given. The surface model is used to predict the solid-liquid separation efficiency and biogas purification effect under different parameter combinations, providing the theoretically optimal parameter adjustment direction for the system. It is a multidimensional tensor product spline function.

[0157] Random Forest Output: This predicts the future system state under different parameter adjustment schemes, including solid-liquid separation effect, changes in biogas composition, system energy consumption, and equipment failure probability. For example, it predicts the changes in filtrate turbidity and methane concentration over a future period after adjusting membrane filtration pressure and activated carbon adsorption time, as well as the likelihood of equipment failure. The random forest prediction function is an ensemble of T decision trees: ,in Let be the prediction function of the t-th decision tree. Through bootstrap sampling and random feature selection training, nonlinear relationships (such as the interaction between methane concentration and water washing tower liquid level and activated carbon temperature) can be captured. The number of decision trees.

[0158] High-level integration: Decision-level integration based on dynamic weighting using reinforcement learning. The weights of each algorithm's output are dynamically adjusted according to the system's real-time operating status and set objectives. Multiple reward mechanisms are established, such as positive rewards for improved solid-liquid separation efficiency, reduced system energy consumption, and negative rewards for increased equipment failure probability. The reinforcement learning algorithm adjusts the weights of each algorithm's output based on these reward feedbacks. For example, when the system is in an unstable state, such as a sharp increase in membrane filtration pressure differential, the weight of the adaptive control algorithm's output is increased to quickly adjust the membrane filtration pressure; when optimizing long-term operating effects, such as improving the stability of methane concentration in biogas, the weights of surface fitting and random forest algorithms are increased. Parameter adjustments are made by comprehensively considering multiple factors to ultimately obtain a final control output that comprehensively considers the advantages of each algorithm. .

[0159] 2.3 Joint Training Process

[0160] The fusion model is trained using a multi-objective optimization framework:

[0161] ;

[0162] in: This is a set of model parameters, covering adaptive control algorithms. , The rule parameters of the fuzzy optimization algorithm (such as the membership function parameters of the fuzzy set, the weights of the fuzzy rules, etc.), and the coefficients of the surface fitting model. The tree structure parameters of the random forest model (such as the splitting node conditions of the decision tree, the depth of the tree, etc.). These parameters determine the various algorithms ( The performance and behavior of the algorithm (number of algorithms).

[0163] For the first The loss functions of each algorithm are as follows: Adaptive control algorithm uses the deviation between the actual control parameters and the target parameters as the loss, such as the deviation between the actual value of the membrane filtration pressure and the set optimal pressure value; Random forest uses the error between the predicted value and the actual value as the loss, such as the error between the predicted methane concentration and the actual detected methane concentration; Fuzzy optimization algorithm uses the matching degree of fuzzy rules and the comprehensive evaluation of the control effect as the loss; Surface fitting model uses the error between the fitted surface and the actual data points as the loss.

[0164] The weights are set according to the importance of each algorithm in the system. Key algorithms, such as adaptive control algorithms, can be assigned higher weights; auxiliary optimization algorithms have their weights appropriately reduced. Furthermore, during training, the weights are dynamically adjusted based on the algorithm's performance; for example, if an algorithm performs exceptionally well under specific conditions, its weight is increased.

[0165] The weight of the regularization term.

[0166] For regularization, L1 or L2 regularization is used to prevent overfitting and ensure that the model maintains good generalization ability under different conditions. During training, the regularization parameter is adjusted to balance the model's fitting ability and generalization ability.

[0167] III. Algorithm Application Process (e.g.) Figure 4 (As shown)

[0168] 3.1 Real-time data processing

[0169] The data acquisition and transmission module collects real-time data such as pressure difference, filtrate turbidity, filtration temperature, and separation time from the solid-liquid preliminary separation module; methane concentration, water washing tower liquid level, activated carbon adsorption tower temperature, water washing tower washing time, and activated carbon usage cycle from the biogas purification and impurity separation module; and anaerobic reactor temperature, system operating time, water pump current, system energy consumption, and anaerobic reactor pH value from the system operation status. This data is then transmitted to the data analysis and processing module.

[0170] The data analysis and processing module performs preprocessing operations such as data cleaning, filtering, and integration. It removes outlier data, such as invalid data where the membrane filtration pressure difference shows abrupt changes and does not conform to normal variation patterns. For missing data, it uses interpolation methods (such as linear interpolation, polynomial interpolation, etc.) to fill in the gaps. Simultaneously, it standardizes the data, converting data with different dimensions to a unified dimension. For example, it uses normalization methods to map the data to the [0,1] interval, or uses standardization methods to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, making it comparable and facilitating subsequent algorithm analysis and processing.

[0171] 3.2 Parallel Inference Computation

[0172] The adaptive control algorithm module calculates parameter adjustment instructions for the solid-liquid preliminary separation module and the biogas purification and impurity separation module based on real-time feature data and the adaptive control framework. Based on current data such as membrane filtration pressure differential and hydrocyclone inlet flow rate, the direction and magnitude of adjustment for parameters such as membrane filtration pressure and hydrocyclone rotation speed are preliminarily determined.

[0173] The fuzzy optimization algorithm module optimizes the parameters of the adaptive control algorithm based on fuzzy rules and real-time data, and outputs the adjusted instructions. For example, if the pressure difference of the membrane filtration unit is detected to be in the "high" range and the inlet flow rate of the cyclone separator is in the "low" range, the weight of the membrane filtration pressure adjustment command is adjusted from the initial value according to the corresponding rule in the fuzzy rule base. Increase to At the same time, the weight of the cyclone separator speed control command is changed from Reduce to And output the adjusted parameter instructions.

[0174] The surface fitting analysis module uses the constructed multidimensional surface model to analyze the real-time collected feature data. Substitute these parameters into the model. The predicted solid-liquid separation efficiency and biogas purification effect corresponding to different parameter combinations on the surface are calculated and compared with the actual operating results. For example, if the predicted solid-liquid separation efficiency corresponding to the current combination of parameters such as membrane filtration pressure and cyclone separator speed on the surface is lower than the actual requirement, the surface fitting analysis module will provide parameter adjustment suggestions based on the surface change trend, suggesting that the membrane filtration pressure should be increased by p units and the cyclone separator speed should be decreased by s units. .

[0175] The random forest prediction module inputs real-time feature data into a trained random forest model. Each decision tree processes the data according to its own splitting rules, obtaining its own prediction result. The final prediction result is obtained by averaging the prediction results of T decision trees. For example, predicting that the methane concentration at the outlet of the activated carbon adsorption tower may drop to concentration c within the next hour under current operating parameters, or predicting the probability of the water pump failing within the next n hours. .

[0176] 3.3 Dynamic Weighted Decision Making

[0177] Decision and instruction generation module receives , , , Then, the output weights of each algorithm are dynamically adjusted based on the reinforcement learning algorithm. The reinforcement learning algorithm maintains a state-action-reward mapping table, with the current system running state (including various feature data, device operating parameters, etc.) as the state and the parameter adjustment schemes output by each algorithm as the actions.

[0178] After the system performs a certain action, it provides a corresponding reward based on the actual operating effect. For example, if the solid-liquid separation efficiency improves after executing the adjusted parameter command, the system will perform a certain action. Furthermore, the system energy consumption has been reduced. A positive reward will be given. ( , (This refers to the reward coefficient set according to the system optimization goals); if the equipment malfunctions or the separation effect deteriorates, a negative reward will be given.

[0179] Reinforcement learning algorithms update the weights of each algorithm's output based on reward feedback using methods such as policy gradient algorithms. For example, if the parameter adjustment scheme output by the adaptive control algorithm module continuously improves the system's performance, its weight is increased. When the random forest prediction module accurately predicts under complex conditions, helping the system to avoid problems in advance, its weight should be increased. Similarly, when the fuzzy optimization algorithm module and the surface fitting analysis module receive positive rewards, their weights can be increased respectively. and .

[0180] Finally, the final control command is generated through a weighted summation method. .

[0181] 3.4 Control Command Execution

[0182] The actuator module receives the final control command. Then, the instructions are parsed. If the instruction contains membrane filtration pressure adjustment information, the actuator module will control the pressure adjustment device of the membrane filtration unit to adjust the membrane filtration pressure to the specified value; if it involves a cyclone separator speed control instruction, it will drive the motor of the cyclone separator to adjust its speed.

[0183] For the biogas purification and impurity separation module, if the instruction requires adjusting the water level in the water washing tower, the actuator module will control the inlet and outlet valves of the water washing tower to change the liquid level. If the activated carbon adsorption time needs to be adjusted, the operating time of the adsorption tower will be controlled. At the same time, for equipment such as water pumps, their power or operating status will be adjusted according to the instructions. Through precise control of each device, the separation system can be automated and dynamically optimized, ensuring that the system is always in a highly efficient and stable working state.

[0184] The embodiments described above are for illustrative purposes only and are not intended to limit the invention. Therefore, any changes in numerical values ​​or substitutions of equivalent elements should still fall within the scope of this invention.

[0185] The above detailed description will enable those skilled in the art to understand that the present invention can indeed achieve the aforementioned objectives and has complied with the provisions of the Patent Law.

[0186] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention. The above descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

[0187] It should be noted that the above description of the process is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to the process under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0188] The basic concepts have been described above. Obviously, for those skilled in the art who have read this application, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore, such modifications, improvements, and corrections still fall within the spirit and scope of the exemplary embodiments of this application.

[0189] Furthermore, this application uses specific terms to describe its embodiments. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different positions in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application can be appropriately combined.

[0190] Furthermore, those skilled in the art will understand that aspects of this application can be described and illustrated through several patentable types or situations, including any new and useful combination of processes, machines, products, or substances, or any new and useful improvements thereof. Therefore, aspects of this application can be implemented entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. All of the above hardware or software can be referred to as a “unit,” “module,” or “system.” Furthermore, aspects of this application can take the form of a computer program product embodied in one or more computer-readable media, wherein computer-readable program code is contained therein.

[0191] The computer program code required for the operation of each part of this application can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, and Python; general programming languages ​​such as C; Visual Basic, Fortran2103, Perl, COBOL2102, PHP, and ABAP; dynamic programming languages ​​such as Python, Ruby, and Groovy; or other programming languages. This program code can run entirely on the user's computer, or as a standalone software package on the user's computer, or partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer can be connected to the user's computer via any network, such as a local area network (LAN) or wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as Software as a Service (SaaS).

[0192] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this application are not intended to limit the order of the processes and methods of this application. Although some currently considered useful embodiments of the invention have been discussed in the foregoing disclosure by way of various examples, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the substance and scope of the embodiments of this application. For example, although the implementation of the various components described above can be embodied in a hardware device, it can also be implemented as a purely software solution, such as an installation on an existing server or mobile device.

[0193] Similarly, it should be noted that, in order to simplify the description of the present application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of the present application sometimes combines multiple features into a single embodiment, drawing, or description thereof. However, this approach of the present application should not be construed as reflecting an intention that the claimed subject matter requires more features than expressly recited in each claim. Rather, the subject of the invention should possess fewer features than in any single embodiment described above.

Claims

1. A separation system for a high-rate anaerobic reactor, characterized in that, The system comprises a solid-liquid preliminary separation module, a biogas purification and impurity separation module, a data acquisition and transmission module, a data analysis and processing module, a fusion algorithm control module, a decision and instruction generation module, and an actuator module. The solid-liquid preliminary separation module is used for solid-liquid preliminary separation of the mixed liquid output by the anaerobic reactor. The biogas purification and impurity separation module is used for purification and impurity separation of the biogas discharged by the anaerobic reactor. The data acquisition and transmission module is used for collecting data of the solid-liquid preliminary separation module and the biogas purification and impurity separation module and system operation data, and transmitting to the data analysis and processing module. The data analysis and processing module is used for preprocessing data, extracting key feature information, and transmitting to the fusion algorithm control module. The fusion algorithm control module takes a self-adaptive control algorithm as the core, fuses fuzzy optimization, surface fitting and random forest technology, processes the received data, generates parameter adjustment instructions and transmits to the decision and instruction generation module, and the specific processing is as follows: Feature importance fusion: the importance weight of each feature in the anaerobic reactor separation system is determined by the analytic hierarchy process; Design a three-level fusion architecture: Feature layer fusion: the original data is combined according to the importance weight; Algorithm layer fusion: the intermediate results are output after parallel processing of each algorithm; Decision layer fusion: based on dynamic weighting of reinforcement learning, the weight of each algorithm output result is dynamically adjusted according to the real-time running state and the set target of the system; The fuzzy optimization algorithm in the fusion algorithm control module preliminarily optimizes the parameters of the self-adaptive control algorithm according to the pressure difference data of the solid-liquid preliminary separation module, the biogas component concentration data of the biogas purification and impurity separation module, and the system operation data through the fuzzy rule base; The surface fitting technology in the fusion algorithm control module performs multi-dimensional analysis on the historical and real-time pressure difference data of the solid-liquid preliminary separation module and the biogas component concentration data of the biogas purification and impurity separation module, constructs a data surface model, and mines the nonlinear relationship between the system operation parameters and the separation effect, providing parameter adjustment direction for the self-adaptive control algorithm; The random forest technology in the fusion algorithm control module trains and learns based on historical operation data and real-time monitoring data, predicts the solid-liquid separation effect and biogas component change trend under different operation parameter combinations, and provides the prediction results to the self-adaptive control algorithm for parameter adjustment; The decision and instruction generation module generates the final control instruction and transmits it to the actuator module according to the parameter adjustment instruction and the actual system operation condition; the decision and instruction generation module dynamically adjusts the weight of each algorithm output result in the fusion algorithm control module based on reinforcement learning, and generates the final control instruction by weighted summation; The actuator module controls the running state of the related equipment of the solid-liquid preliminary separation module and the biogas purification and impurity separation module according to the final control instruction.

2. The separation system for high-rate anaerobic reactors according to claim 1, characterized in that, The solid-liquid preliminary separation module adopts a combination of cyclone separation and membrane filtration, and the mixed liquid first enters the cyclone separator to separate particulate solids, and the preliminarily separated liquid then enters the membrane filtration unit to intercept suspended solids.

3. The separation system for a high-rate anaerobic reactor according to claim 2, wherein, The biogas purification and impurity separation module comprises a water washing tower and an activated carbon adsorption tower, and the biogas enters the water washing tower first to remove water-soluble impurities, and then passes through the activated carbon adsorption tower to adsorb organic impurities and odor substances.

4. The separation system for high-rate anaerobic reactors according to claim 1, characterized in that, The data preprocessing operation of the data analysis and processing module comprises data cleaning, screening, integration and extraction of key feature information, removal of abnormal data and unification of data format.

5. The separation system for high-rate anaerobic reactors according to claim 3, characterized in that, The solid-liquid preliminary separation module equipment controlled by the actuator module comprises a water pump, a valve and a cyclone separator motor, and the biogas purification and impurity separation module equipment controlled by the actuator module comprises water inlet and outlet valves of the water washing tower and an activated carbon adsorption tower switching device.

6. The separation system for a high-rate anaerobic reactor according to claim 5, characterized in that, The data and system operation data of the solid-liquid preliminary separation module and the biogas purification and impurity separation module comprise a pressure difference of a membrane filtration unit of the solid-liquid preliminary separation module, an inlet flow of the cyclone separator, a filtrate turbidity of the membrane filtration unit, a filtration temperature of the membrane filtration unit, a separation time of the cyclone separator, a methane concentration at an outlet of the activated carbon adsorption tower in the biogas purification and impurity separation module, a liquid level height of the water washing tower, a temperature in the activated carbon adsorption tower, a washing time of the water washing tower, a service period of the activated carbon, a temperature in the anaerobic reactor, a system operation time, a current of the water pump in the actuator module, an energy consumption of the whole system and a pH value in the anaerobic reactor.

Images