Dosing control method, system, equipment and medium for sewage treatment process

Abstract

The invention provides a dosing control method, system and equipment for a sewage treatment process and a medium, and the method comprises the following steps: obtaining multi-source heterogeneous data, and processing the multi-source heterogeneous data to construct a high-dimensional state vector; inputting the high-dimensional state vector into a drug demand dynamic prediction model to obtain a predicted concentration sequence; obtaining a target control instruction according to the current high-dimensional state vector and the predicted concentration sequence; the target control instruction is issued to a controller of the target dosing pump, and dosing control is carried out through the controller; the method solves the industrial problems of response lag, excessive or insufficient addition, incapability of coping with sudden change of water quality, high operation cost and the like existing in the traditional artificial experience control or simple PID (Proportion Integration Differentiation) control. According to the method, under the rigid constraint of absolutely ensuring that the effluent quality is stable and reaches the standard, the medicament consumption can be reduced to the greatest extent, the operation cost is reduced, the secondary pollution is avoided, and the automation and intelligence level of the operation of the whole sewage treatment system is improved.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and in particular to a method, system, equipment and medium for controlling the dosing of chemicals in a wastewater treatment process. Background Technology

[0002] To minimize the cost of chemicals per ton of water treatment while ensuring consistently high effluent quality, a technological shift is required, moving from traditional, inefficient, and delayed chemical dosing methods to real-time, precise, and predictive intelligent dosing. Current chemical dosing control schemes widely used in the wastewater treatment industry suffer from the following main problems: (1) Traditional manual experience control: Since humans cannot continuously process massive amounts of real-time data and make optimal calculations, and experience is difficult to standardize and pass on, it will heavily rely on the personal experience of operators. The dosage setting is highly subjective and fluctuates greatly. The adjustment frequency is low (usually once every few hours), and it is impossible to track minute-level changes in water quality. The control quality declines at night or when personnel are not energetic.

[0003] (2) Feedforward / feedback control based on fixed parameters or simple proportions (such as classic PID): response lag and low control accuracy. The essence of PID control is to compensate for "errors that have already occurred". For wastewater treatment, a system with "large inertia and large lag", when the instrument detects that the phosphorus or nitrogen in the effluent exceeds the standard, the pollutants have already passed through the reaction tank, and adding more chemicals is useless. In order to prevent exceeding the standard, it is often necessary to maintain a high dosage benchmark for a long time, resulting in waste.

[0004] (3) Early or single automated dosing schemes: Although automatic start and stop are achieved, the set value still needs to be given manually, or it is only added according to a single parameter (such as influent flow rate) at a fixed ratio. It cannot adapt to the time-varying characteristics of influent concentration (COD, TP), resulting in waste when the influent concentration is low and insufficient dosing when the concentration is high.

[0005] (4) Lack of intelligent coordination with the whole process: The dosing unit often operates in isolation and lacks information exchange and strategy coordination with the intelligent control systems of other process units such as aeration, recirculation, and sludge discharge (if they exist). For example, when the optimization and adjustment of the aeration intelligent body leads to changes in the nitrification effect, the carbon source demand for denitrification changes accordingly, but the independent dosing system cannot sense and respond to this change.

[0006] Therefore, how to achieve real-time sensing, accurate calculation, and collaborative operation with other intelligent units throughout the wastewater treatment process, ultimately ensuring absolute water quality safety while unlocking the greatest potential for cost savings, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] To address the aforementioned technical problems, the present invention aims to provide a method, system, equipment, and medium for chemical dosing control in wastewater treatment processes. This solution overcomes industry pain points such as response lag, overdosing or underdosing, inability to handle sudden changes in water quality, and high operating costs associated with traditional manual experience-based control or simple PID control. This application, under the strict constraint of ensuring stable effluent quality (especially key indicators such as total phosphorus and total nitrogen), minimizes reagent consumption, reduces operating costs, avoids secondary pollution, and enhances the automation and intelligence level of the entire wastewater treatment system.

[0008] The first objective of this invention is to provide a method for controlling the dosing of chemicals in a wastewater treatment process; The technical solution provided by this invention is as follows: A method for controlling chemical dosing in a wastewater treatment process includes the following steps: Acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector; The high-dimensional state vector is input into the dynamic prediction model of drug demand to obtain the predicted concentration sequence. The target control command is obtained based on the current high-dimensional state vector and the predicted concentration sequence; The target control command is sent to the controller of the target dosing pump, and the dosing is controlled by the controller.

[0009] Preferably, the step of acquiring multi-source heterogeneous data and processing the multi-source heterogeneous data to construct a high-dimensional state vector specifically includes: The multi-source heterogeneous data is cleaned in real time, and key features are calculated based on the cleaned multi-source heterogeneous data; Based on the key features, a high-dimensional state vector representing the complete operating condition of the system at time t is constructed.

[0010] Preferably, the step of inputting the high-dimensional state vector into the dynamic prediction model of drug demand to obtain the predicted concentration sequence specifically includes: The agent is loaded and run in the pre-trained dynamic prediction model for drug demand; The high-dimensional state vector and key driving factors are input into the drug demand dynamic prediction model containing the agent to obtain the predicted concentration sequence.

[0011] Preferably, the method for obtaining the key driving factor is as follows: By analyzing historical process data through causal feature discovery algorithms, process variables with strong causal relationships with drug demand are identified, and these process variables are used as the key driving factors.

[0012] Preferably, obtaining the target control command based on the current high-dimensional state vector and the predicted concentration sequence specifically includes: The current high-dimensional state vector and the predicted concentration sequence are input into a deep reinforcement learning policy network to output the target delivery action; The target deployment action is converted into a target control command.

[0013] Preferably, the step of sending the target control command to the controller of the target dosing pump, and controlling the dosing through the controller, specifically includes: The target control command is sent to the PLC or frequency converter of the target dosing pump through the industrial network, and the dosing control is performed through the PLC or the frequency converter. The controller of the target dosing pump includes either the PLC or the frequency converter.

[0014] Preferably, after the step of sending the target control command to the controller of the target dosing pump and controlling the dosing through the controller, the method further includes: The drug dosing data after the execution of the target control command is continuously collected to form an empirical tuple, which is used to update the dynamic prediction model of drug demand.

[0015] The second objective of this invention is to provide a chemical dosing control system for a wastewater treatment process; The technical solution provided by this invention is as follows: A dosing control system for a wastewater treatment process includes: a treatment module, an input module, an acquisition module, and a control module; The processing module is used to acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector. The input module is used to input the high-dimensional state vector into the dynamic prediction model of drug demand in order to obtain the predicted concentration sequence; The acquisition module is used to acquire target control commands based on the current high-dimensional state vector and the predicted concentration sequence. The control module is used to send the target control command to the controller of the target dosing pump, and to control the dosing through the controller.

[0016] The third objective of this invention is to provide an electronic device; The technical solution provided by this invention is as follows: An electronic device, comprising: At least one processor; and A memory communicatively connected to the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method steps of any one of the dosing control methods for a wastewater treatment process.

[0017] A fourth objective of this invention is to provide a computer-readable storage medium; The technical solution provided by this invention is as follows: A computer-readable storage medium for storing a computer program for causing a computer to perform the steps of any one of the methods for controlling the dosing of a wastewater treatment process.

[0018] Compared with existing technologies, this invention provides a dosing control method for wastewater treatment processes, comprising the following steps: acquiring multi-source heterogeneous data and processing the data to construct a high-dimensional state vector; inputting the high-dimensional state vector into a dynamic prediction model of reagent demand to obtain a predicted concentration sequence; obtaining a target control command based on the current high-dimensional state vector and the predicted concentration sequence; and sending the target control command to the controller of the target dosing pump for dosing control. This method solves the industry pain points of traditional manual experience control or simple PID control, such as response lag, overdosing or underdosing, inability to cope with sudden changes in water quality, and high operating costs. This method can minimize reagent consumption, reduce operating costs, avoid secondary pollution, and improve the automation and intelligence level of the entire wastewater treatment system under the strict constraint of ensuring stable effluent quality (especially key indicators such as total phosphorus and total nitrogen).

[0019] The present invention also provides a dosing control system for a wastewater treatment process. Since this system and the dosing control method for the wastewater treatment process solve the same technical problem and belong to the same technical concept, they should have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A schematic flowchart of a chemical dosing control method for a wastewater treatment process provided in an embodiment of this application; Figure 2A flowchart illustrating the execution of a chemical dosing control method for another wastewater treatment process provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of a chemical dosing control system for a wastewater treatment process provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] like Figures 1 to 2 As shown in the figure, an embodiment of the present invention provides a method for controlling the dosing of chemicals in a wastewater treatment process, comprising the following steps: S1. Acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector; S2. Input the high-dimensional state vector into the dynamic prediction model of drug demand to obtain the predicted concentration sequence; S3. Obtain the target control command based on the current high-dimensional state vector and the predicted concentration sequence; S4. The target control command is sent to the controller of the target dosing pump, and the dosing is controlled by the controller.

[0024] In steps S1 to S4, the first step is to collect ① water quality and process parameters: key data from the plant's automated control system (PLC / DCS) are collected in real time using industrial communication protocols (such as OPCUA, Modbus TCP). Core data points include: 1) Influent parameters: instantaneous flow rate (Q_in), chemical oxygen demand (COD_in), total phosphorus (TP_in), ammonia nitrogen (NH3-N_in), total nitrogen (TN_in), pH value, and temperature (T). 2) Process unit parameters: oxidation-reduction potential (ORP_anoxic) and dissolved oxygen (DO_anoxic) in the anaerobic / anoxic tank; nitrate concentration (NO3-N_aerobic) and dissolved oxygen (DO_aerobic) at the end of the aerobic tank; and mixed liquor suspended solids concentration (MLSS). 3) Effluent parameters: effluent total phosphorus (TP_out), total nitrogen (TN_out), ammonia nitrogen (NH3-N_out), and chemical oxygen demand (COD_out). ② Equipment and Chemical Status: Collect the operating status (start / stop), inverter frequency (Hz) or stroke position (%), and cumulative dosage volume or weight of the metering pumps on each dosing pipeline; correlate with the real-time liquid level or weight signal of the chemical storage tank. ③ Cost and External Parameters: Obtain current peak / valley / flat time-of-use electricity price information; use multi-source heterogeneous data such as preset or updated chemical unit price (yuan / kg or yuan / liter) through the human-machine interface (HMI), and clean, integrate, and standardize this multi-source heterogeneous data to construct a high-dimensional state vector that can comprehensively reflect the system status. The second step is to input the constructed high-dimensional state vector into a pre-trained dynamic prediction model for chemical demand. Through the analysis and calculation of this model, a predicted chemical concentration sequence for a future period is obtained. The third step is to generate precise target control commands based on the current high-dimensional state vector and the predicted concentration sequence output by the model, through optimization algorithms or rule-based judgment. The fourth step is to send the generated target control command to the controller of the target dosing pump, and the controller will execute the corresponding dosing operation according to the command to achieve accurate control of the dosing process.

[0025] This application aims to systematically address the pain points of traditional chemical dosing control, which is characterized by lag, inefficiency, and isolation. Its core implementation lies in constructing an intelligent decision-making and control system with a closed loop of "perception-prediction-optimization-coordination-evolution," achieving a paradigm shift from experience-driven to data- and model-driven approaches. Specifically, this application goes beyond passively reacting to the current water quality status. Instead, it integrates machine learning predictive models to enable the system to proactively predict future key water quality indicators and process requirements, thereby transforming control actions from reactive remediation to proactive intervention and fundamentally overcoming the significant lag in the treatment process. Building upon this, the application introduces an embedded decision engine based on reinforcement learning or advanced model predictive control. This engine sets "lowest chemical consumption per ton of water" as the core optimization objective and "stable compliance with effluent water quality standards" as an inviolable hard constraint. It solves multivariate optimization problems online, dynamically generating optimal dosing instructions to directly achieve a balance between economy and compliance. To achieve global optimization rather than local optimization, this application utilizes real-time communication and strategic linkage with other intelligent agents such as aeration and reflux units to ensure coordinated rather than conflicting operations between process units, thereby unlocking the potential for cross-system collaborative energy saving and consumption reduction. Finally, the decision-making strategy of the entire intelligent agent is not static; it originates from massive simulation training in a digital twin virtual environment and, through continuous comparison of prediction and feedback effects during actual operation, forms a continuous self-evolutionary capability of cloud training and edge execution. This enables the system to adapt to changes in water quality, process adjustments, and equipment aging, maintaining optimal performance over the long term. In summary, this solution, by endowing the system with four core capabilities—prediction, optimization, coordination, and evolution—transforms reagent dosing from a simple execution action into a continuous intelligent decision-making process based on global information and future insights.

[0026] Preferably, the step of acquiring multi-source heterogeneous data and processing the multi-source heterogeneous data to construct a high-dimensional state vector specifically includes: The multi-source heterogeneous data is cleaned in real time, and key features are calculated based on the cleaned multi-source heterogeneous data; Based on the key features, a high-dimensional state vector representing the complete operating condition of the system at time t is constructed.

[0027] In practical applications, this step involves data cleaning, verification, and key feature generation to construct a high-dimensional state vector, thus laying the foundation for intelligent agents to understand the physical world. It achieves a precise understanding of the system's state through extensive data access and deep processing. ① Data quality assurance: Real-time cleaning of raw data streams, including outlier detection and filtering based on process knowledge base (such as identifying transient glitches in instruments), and intelligent interpolation for short-term communication interruptions (such as using causal-based variable regression for estimation instead of simple linear interpolation).

[0028] ② Key Feature Calculation: Based on the cleaned data, derived features that are more relevant to process guidance are calculated in real time: 1) Pollution load: TP load = Q_in * TP_in; COD load = Q_in * COD_in.

[0029] 2) Process status index: Estimate the real-time food / microorganism ratio (F / M); calculate the sludge activity trend index based on MLSS, SRT (sludge age) and temperature.

[0030] 3) Process efficiency indicators: calculate the real-time phosphorus removal rate and nitrification / denitrification potential.

[0031] ③ Construct a high-dimensional state vector: In each control cycle (e.g., 1 minute), all the processed and calculated data are integrated into a high-dimensional state vector S(t) representing the complete operating condition of the system at time t. For example: S(t) = [Q_in, TP_in, TP_out, ORP_anoxic, NO3-N_aerobic, MLSS, T, pump status, electricity price indicator, reagent level, ...]. The high-dimensional state vector S(t) is the input basis for all subsequent intelligent decisions.

[0032] Preferably, the step of inputting the high-dimensional state vector into the dynamic prediction model of drug demand to obtain the predicted concentration sequence specifically includes: The agent is loaded and run in the pre-trained dynamic prediction model for drug demand; The high-dimensional state vector and key driving factors are input into the drug demand dynamic prediction model containing the agent to obtain the predicted concentration sequence.

[0033] In practical applications, this step focuses on giving the agent the ability to "foresee the future," enabling its control to shift from passive response to proactive planning.

[0034] First, the agent is loaded and run in a pre-trained dynamic prediction model for reagent demand on a cloud platform. This model is a machine learning model that integrates time series analysis and causal inference. In this embodiment, the agent for reagent dosing is a dedicated intelligent decision-making and control module in the "Jianqi Water Treatment Full-Process AI Intelligent Control System," responsible for the precise delivery of chemical reagents (with phosphorus removal agents and denitrification carbon sources as the core). As a software entity (Agent), it is mainly deployed on the edge AI control server on the wastewater treatment plant, collaborating with the cloud-based intelligent decision-making platform to form an integrated "cloud-edge" execution and evolution architecture. This agent achieves fully automatic, adaptive, and optimized control of reagent dosing through a complete "perception-prediction-optimization-execution-learning" closed loop.

[0035] In the dynamic prediction model of drug demand, a high-dimensional state vector is input, which also contains the time series of historical state vectors [S(tk),...,S(t-1),S(t)]. At the same time as inputting the high-dimensional state vector, key driving factors identified by the causal feature discovery algorithm are also introduced to ensure that the prediction focuses on variables with causal relationship and improve robustness.

[0036] The dynamic prediction model for reagent demand predicts the trajectory of key controlled variables within a future optimization time domain (e.g., the next 4 hours, with a total of 240 steps). Finally, the core output is the predicted concentration sequence P(t+1:t+n) of total phosphorus (TP_out) in the future effluent. Here, the trajectory refers to the sequence generated by interaction, which includes actions, predicted states, etc.

[0037] It is important to note that the dynamic forecasting model for drug demand also outputs confidence intervals for the predicted values, which are used for subsequent robust optimization to adopt a more conservative strategy when the prediction uncertainty is high.

[0038] Preferably, the method for obtaining the key driving factor is as follows: By analyzing historical process data through causal feature discovery algorithms, process variables with strong causal relationships with drug demand are identified, and these process variables are used as the key driving factors.

[0039] In practical applications, the causal feature discovery algorithm uses feature variance as the criterion for causal features in historical process data. This achieves feature decoupling without the need for complex neural network structures, identifying process variables with strong causal relationships to reagent requirements. These process variables are then used as key driving factors, ensuring that predictions focus on causally related variables and improving robustness. The specific calculation process of this causal feature discovery algorithm (i.e., causal feature decoupling) has been applied for in a prior patent (CN121436046A) and will not be elaborated here.

[0040] Preferably, obtaining the target control command based on the current high-dimensional state vector and the predicted concentration sequence specifically includes: The current high-dimensional state vector and the predicted concentration sequence are input into a deep reinforcement learning policy network to output the target delivery action; The target deployment action is converted into a target control command.

[0041] In practical applications, this step is the "decision-making brain" of the intelligent agent, transforming predictive information into specific control instructions that are cost-optimal and absolutely compliant.

[0042] At each decision time t, based on the current high-dimensional state vector S(t) and the predicted concentration sequence P(t+1:t+n), a rolling optimization problem in the finite time domain is constructed.

[0043] 1. Decision variable: The sequence of drug dosage for the next m control periods (m≤n) is U(t:t+m-1)=[u(t),u(t+1),...,u(t+m-1)].

[0044] 2. Objective function (minimization): MinJ=Σ[ω1*C_pharmaceutical(u(i))*f_electricity price(i)+ω2*Φ(TP_pred(i)-standard)+ω3*(u(i)-u(i-1))²],i=ttot+m-1; Wherein, C_chemical (u(i)) represents the chemical cost directly related to the dosage u(i); f_electricity price (i) represents the electricity price coefficient for the current period, coupling operating costs with time-of-use electricity prices; Φ(·) represents an asymmetric penalty function. When the predicted effluent index TP_pred(i) is better than the standard, the penalty is 0 or minimal; when it approaches or exceeds the standard, the penalty increases exponentially, thus making "meeting the standard" an inviolable hard constraint; (u(i)-u(i-1))² represents the penalty for large fluctuations in the control quantity, ensuring stable equipment operation and extending its lifespan; ω1, ω2, ω3 represent weighting coefficients, which are determined in the cloud through long-term historical data optimization, balancing economy, safety, and stability.

[0045] 3. Constraints: Process dynamics constraints (implied by the prediction model).

[0046] Equipment physical constraints: u_min≤u(i)≤u_max.

[0047] Safety constraint: Remaining amount (i) in the drug storage tank ≥ safety margin.

[0048] 4. Solving for the optimal control command: A deep reinforcement learning (DRL) policy network is employed: The agent is equipped with an offline-trained DRL policy network π_θ. This network takes a high-dimensional state vector S(t) and the predicted concentration sequence as input, and directly outputs the current optimal dosing action u*(t) = π_θ(S(t),P). The policy network π_θ is trained in a high-fidelity digital twin environment built on the training platform through tens of millions of simulations and trials, with the goal of minimizing the long-term comprehensive cost (i.e., the long-term expectation of the above objective function MinJ).

[0049] Alternatively, an online model predictive control (MPC) solver can be used: the above optimization problem is solved in real time on an edge server to obtain the optimal control sequence U*, and the theoretically optimal action u*(t) in the sequence is executed immediately.

[0050] 5. Command Safety Verification and Output: The theoretically optimal action u*(t) executed above, i.e. the target dosing action, must pass through a final safety limit to ensure that it does not exceed the rated capacity of the dosing pump. Subsequently, this theoretically optimal action u*(t), i.e. the target dosing action, is converted into a specific equipment command, i.e., a target control command, such as "set the frequency of the dephosphorization dosing pump A to 38.5Hz".

[0051] Preferably, the step of sending the target control command to the controller of the target dosing pump, and controlling the dosing through the controller, specifically includes: The target control command is sent to the PLC or frequency converter of the target dosing pump through the industrial network, and the dosing control is performed through the PLC or the frequency converter. The controller of the target dosing pump includes either the PLC or the frequency converter.

[0052] In practical applications, this step ensures that intelligent decisions are safely and reliably translated into physical actions and can respond to on-site anomalies.

[0053] 1. Control command issuance: Through a highly reliable industrial network, the safety-verified command, i.e. the target control command, is issued to the PLC or frequency converter of the target dosing pump. The edge server confirms the receipt of the command and monitors the execution status.

[0054] 2. Intelligent sensor fault diagnosis and seamless switching: This is a key innovation to ensure high system availability.

[0055] 3. Real-time monitoring: Continuously monitor the data quality, communication status, and reading rationality of key input sensors (such as TP online analyzer and NO3-N online analyzer).

[0056] 4. Fault diagnosis and switching: Once the system diagnoses a core sensor failure or serious data distortion (for example, the TP reading deviates significantly from the trends of ORP, MLSS and other strongly correlated variables) through the built-in causal correlation analysis model, it does not simply alarm and shut down.

[0057] 5. Seamless Switching to Soft Measurement Mode: The system automatically and seamlessly switches the core control logic to a soft measurement model based on other reliable sensors. For example, when the online TP meter fails, it immediately switches to a soft measurement value of TP estimated in real time based on multi-source data such as influent TP load, anaerobic tank ORP, MLSS, and historical dosage, which serves as the input for prediction and control, ensuring uninterrupted operation of the control loop.

[0058] Preferably, after the step of sending the target control command to the controller of the target dosing pump and controlling the dosing through the controller, the method further includes: The drug dosing data after the execution of the target control command is continuously collected to form an empirical tuple, which is used to update the dynamic prediction model of drug demand.

[0059] In practical applications, this step endows the agent with the ability to "learn for life," enabling its performance to continuously improve over time.

[0060] 1. Operational effect tracking and KPI calculation: The system continuously collects data such as actual effluent quality, actual chemical consumption, and equipment operating parameters after command execution, forming an empirical tuple of (state S(t), action u*(t), result S(t+1), actual cost).

[0061] 2. Data Feedback and Upload: After being encrypted, this experience data is synchronously uploaded to the central data warehouse of the cloud training platform.

[0062] 3. Cloud-based model retraining and optimization: The training platform periodically (e.g., weekly or monthly) retrains and optimizes the prediction model in step two and the DRL policy network (or MPC model parameters) in step three using massive amounts of newly generated operational data from across the plant. The training process takes place in a more macroscopic virtual digital twin that includes more extreme operating condition simulations, and uses adversarial training and causal counterfactual reasoning to improve the model's generalization ability and robustness.

[0063] 4. Model Validation and Dynamic Updates: After rigorous offline simulation testing and validation, the newly trained model is compiled and securely deployed to the edge server, replacing the original model. This enables silent, seamless upgrades and continuous evolution of the dosing agent control strategy, allowing it to automatically adapt to long-term changes in influent water quality characteristics, process adjustments, and equipment performance evolution.

[0064] like Figure 3 As shown, this embodiment of the invention provides a chemical dosing control system for a wastewater treatment process, comprising: a processing module, an input module, an acquisition module, and a control module; The processing module is used to acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector. The input module is used to input the high-dimensional state vector into the dynamic prediction model of drug demand in order to obtain the predicted concentration sequence; The acquisition module is used to acquire target control commands based on the current high-dimensional state vector and the predicted concentration sequence. The control module is used to send the target control command to the controller of the target dosing pump, and to control the dosing through the controller.

[0065] In practical applications, the dosing control system for wastewater treatment processes includes a processing module, an input module, an acquisition module, and a control module. The input module is connected to both the processing and acquisition modules. The acquisition module is also connected to the control module. The processing module acquires multi-source heterogeneous data, processes it to construct a high-dimensional state vector, and then transmits this high-dimensional state vector to the input module. The input module inputs the high-dimensional state vector into a dynamic prediction model for reagent demand to obtain a predicted concentration sequence, which is then transmitted to the acquisition module. The acquisition module obtains the target control command based on the current high-dimensional state vector and the predicted concentration sequence, and then transmits the target control command to the control module. The control module issues the target control command to the controller of the target dosing pump for dosing control. This system, through the coordinated operation of the processing, input, acquisition, and control modules, solves industry pain points such as response lag, over- or under-dosing, inability to cope with sudden changes in water quality, and high operating costs associated with traditional manual experience control or simple PID control. This method can minimize reagent consumption, reduce operating costs, avoid secondary pollution, and improve the automation and intelligence level of the entire wastewater treatment system while ensuring that the effluent quality meets the standards (especially key indicators such as total phosphorus and total nitrogen) under strict constraints.

[0066] Furthermore, embodiments of this application also disclose an electronic device, Figure 4 This is a structural diagram of an electronic device according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of use of this application.

[0067] Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of this application. The electronic device 20 specifically includes: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the dosing control method of the wastewater treatment process disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment can specifically be an electronic computer.

[0068] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a dosing control channel for the wastewater treatment process between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0069] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0070] The operating system 221 manages and controls the various hardware devices on the electronic device 20 and the computer program 222 to enable the processor 21 to perform calculations and processing on the data 223 in the memory 22. It can be Windows Server, Netware, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the dosing control method for the wastewater treatment process executed by the electronic device 20 as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the dosing control device for the wastewater treatment process from external devices, as well as data collected by its own input / output interface 25.

[0071] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0072] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned dosing control method for the wastewater treatment process. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0073] It should be understood that the use of terms such as "method," "apparatus," "unit," and / or "module" in this application is merely to distinguish one method of different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0074] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "a," and / or "the" are not specifically singular and may include the plural. Generally, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements. An element defined by the phrase "comprising an..." does not exclude the presence of other identical elements in the process, method, product, or apparatus that includes the element.

[0075] Hereinafter, 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.

[0076] If a flowchart is used in this application, it is used to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0077] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for controlling chemical dosing in a wastewater treatment process, characterized in that, Includes the following steps: Acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector; The high-dimensional state vector is input into the dynamic prediction model of drug demand to obtain the predicted concentration sequence. The target control command is obtained based on the current high-dimensional state vector and the predicted concentration sequence. The target control command is sent to the controller of the target dosing pump, and the dosing is controlled by the controller.

2. The dosing control method for the wastewater treatment process according to claim 1, characterized in that, The process of acquiring multi-source heterogeneous data and processing the multi-source heterogeneous data to construct a high-dimensional state vector specifically includes: The multi-source heterogeneous data is cleaned in real time, and key features are calculated based on the cleaned multi-source heterogeneous data; Based on the key features, a high-dimensional state vector representing the complete operating condition of the system at time t is constructed.

3. The dosing control method for the wastewater treatment process according to claim 1, characterized in that, The step of inputting the high-dimensional state vector into the dynamic prediction model of drug demand to obtain the predicted concentration sequence specifically includes: The agent is loaded and run in the pre-trained dynamic prediction model for drug demand; The high-dimensional state vector and key driving factors are input into the drug demand dynamic prediction model containing the agent to obtain the predicted concentration sequence.

4. The dosing control method for the wastewater treatment process according to claim 3, characterized in that, The method for obtaining the key driving factors is as follows: By analyzing historical process data through causal feature discovery algorithms, process variables with strong causal relationships with drug demand are identified, and these process variables are used as the key driving factors.

5. The dosing control method for the wastewater treatment process according to claim 1, characterized in that, The step of obtaining the target control command based on the current high-dimensional state vector and the predicted concentration sequence specifically includes: The current high-dimensional state vector and the predicted concentration sequence are input into a deep reinforcement learning policy network to output the target delivery action; The target deployment action is converted into a target control command.

6. The dosing control method for the wastewater treatment process according to claim 1, characterized in that, The step of sending the target control command to the controller of the target dosing pump, and controlling the dosing through the controller, specifically includes: The target control command is sent to the PLC or frequency converter of the target dosing pump through the industrial network, and the dosing control is performed through the PLC or the frequency converter. The controller of the target dosing pump includes either the PLC or the frequency converter.

7. The dosing control method for the wastewater treatment process according to claim 1, characterized in that, After the target control command is sent to the controller of the target dosing pump, and the dosing control is performed by the controller, the process further includes: The drug dosing data after the execution of the target control command is continuously collected to form an empirical tuple, which is used to update the dynamic prediction model of drug demand.

8. A dosing control system for a wastewater treatment process, characterized in that, include: Processing module, input module, acquisition module, and control module; The processing module is used to acquire multi-source heterogeneous data and process the multi-source heterogeneous data to construct a high-dimensional state vector. The input module is used to input the high-dimensional state vector into the dynamic prediction model of drug demand in order to obtain the predicted concentration sequence; The acquisition module is used to acquire target control commands based on the current high-dimensional state vector and the predicted concentration sequence. The control module is used to send the target control command to the controller of the target dosing pump, and to control the dosing through the controller.

9. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium is used to store a computer program that causes a computer to perform the method described in any one of claims 1-7.

Images