Water treatment method and equipment based on analogue simulation and deep learning, and medium

Abstract

The invention discloses a water treatment method and equipment based on analogue simulation and deep learning and a medium, and relates to the technical field of urban water supply systems and water treatment. The method comprises the steps of obtaining design data related to water treatment plant facilities, and based on the design data, performing simulation modeling on the water treatment whole-process facilities to obtain a facility simulation model; obtaining process data related to processes of the water treatment plant, and performing reaction simulation on the facility simulation model based on the process data to obtain a water treatment simulation model; the method comprises the following steps: acquiring raw water quality data and equipment state data of a water treatment plant in real time, and inputting the raw water quality data and the equipment state data into a pre-trained deep reinforcement learning controller to generate a facility or process control instruction; and inputting the facility or process control instruction into the water treatment simulation model for execution, and monitoring effluent quality data in real time so as to dynamically adjust a control strategy according to the effluent quality data.

Description

Technical Field

[0001] The present application relates to the field of urban water supply systems and water treatment technology, and in particular to a water treatment method, equipment and medium based on simulation and deep learning. Background Art

[0002] Intelligent control of urban water supply systems is key to improving water treatment efficiency and water quality stability. Although most water plants have achieved basic automation (such as remote control of water intake pump stations), their control logic is mostly concentrated in single-variable open-loop or closed-loop systems (such as single flow regulation or pH feedback control). Such systems are difficult to cope with the complex scenarios of multiple coupled processes in water production processes, especially in hydraulic scheduling, chemical addition, and other multivariable closed-loop control links that need to synchronously respond to water quality parameters (turbidity, residual chlorine, ammonia nitrogen, etc.), equipment status, and environmental interference. Existing technologies still rely heavily on manual experience to adjust parameters.

[0003] Traditional methods use mechanistic models to aid decision-making, but water treatment involves the strong coupling of physical reactions (fluid dynamics), chemical reactions (disinfection kinetics), and biological processes (activated carbon adsorption). Existing modeling approaches often focus on a single process (e.g., simulating only the flow pattern in a sedimentation tank) and lack the ability to collaboratively simulate the entire process, including coagulation, filtration, and disinfection. This results in insufficient model prediction accuracy. When raw water quality fluctuates or equipment anomalies occur, the model struggles to map the nonlinear changes in the real system in real time, making it impossible to provide a high-confidence basis for control strategies.

[0004] In addition, conventional control systems have significant limitations in multi-variable closed-loop scenarios (such as dynamically adjusting the dosage of chemical dosage, valve opening, and pump frequency based on real-time water quality). Rule-based control logic is difficult to cover complex working conditions, and traditional optimization algorithms (such as PID control) are prone to falling into local optimality when dealing with high-dimensional, strongly coupled variables, resulting in wasted chemical consumption, increased energy consumption, and fluctuations in effluent water quality. Although some studies have attempted to introduce machine learning methods, they have not solved the problem of efficient coordination between simulation environments and real control systems, resulting in a disconnect between training data and real-time control and weak strategy generalization capabilities.

[0005] Therefore, insufficient automation level, unstable water purification effect and low water production efficiency in urban water supply systems have become technical problems that technical personnel in this field urgently need to solve. Summary of the Invention

[0006] The embodiments of the present application provide a water treatment method, equipment, and medium based on simulation and deep learning to solve the following technical problems: insufficient automation level, unstable water purification effect, and low water production efficiency in urban water supply systems.

[0007] In the first aspect, an embodiment of the present application provides a water treatment method based on simulation and deep learning, the method including: obtaining design data related to water treatment plant facilities, and based on the design data, simulating and modeling the water treatment facilities throughout the entire process to obtain a facility simulation model; obtaining process data related to the water treatment plant process, and based on the process data, performing reaction simulation on the facility simulation model to obtain a water treatment simulation model; obtaining raw water quality data and equipment status data of the water treatment plant in real time, and inputting the raw water quality data and equipment status data into a pre-trained deep reinforcement learning controller to generate facility or process control instructions; inputting the facility or process control instructions into the water treatment simulation model for execution, and monitoring the effluent water quality data in real time to dynamically adjust the control strategy according to the effluent water quality data.

[0008] In one embodiment of the present application, based on the design data, the water treatment full-process facilities are simulated and modeled to obtain a facility simulation model, specifically including: based on the design data, constructing a 3D model of each facility in the water treatment full-process, and defining the connection relationship between the 3D models of each facility; based on the connection relationship, simulating the internal structure and connection method of each 3D model of the facility to complete the geometric accuracy and topological relationship of the facility simulation model.

[0009] In one embodiment of the present application, a reaction simulation is performed on a facility simulation model based on process data to obtain a water treatment simulation model, specifically including: performing an operation simulation on the facility simulation model based on the process data, and constructing a water treatment mathematical model based on water treatment principles and process data to simulate the treatment effect of each water treatment node; obtaining historical water treatment data, and determining the simulation parameters in the water treatment mathematical model based on the historical water treatment data; wherein the historical water treatment data includes: historical water quality data, historical equipment status data and historical control instruction data; the historical water quality data includes: historical raw water quality data and historical effluent water quality data.

[0010] In one embodiment of the present application, a water treatment mathematical model is constructed based on water treatment principles and process data, specifically including: constructing a physical reaction model according to the fluid dynamics characteristics of the coagulation and sedimentation process; constructing a chemical reaction model based on the chemical kinetics principles of the disinfection and dosing process; embedding the physical reaction model and the chemical reaction model into the facility simulation model to simulate the real-time water quality changes of the raw water after passing through each treatment node.

[0011] In one embodiment of the present application, the method also includes: pre-training a deep reinforcement learning controller, specifically including: constructing a deep reinforcement learning network framework based on the Actor-Critic framework; wherein the Actor-Critic framework includes: an Actor network and a Critic network; defining a target score for the convergence of the Actor network training and defining a time difference error threshold for the convergence of the Critic network training; based on historical water treatment data, training the deep reinforcement learning network framework until a converged deep reinforcement learning control is obtained.

[0012] In one embodiment of the present application, a deep reinforcement learning network framework is trained based on historical water treatment data, specifically including: constructing a virtual training environment based on a water treatment simulation model, and inputting historical raw water quality data into the virtual training environment to generate a simulated water quality state; according to the simulated water quality state, outputting a simulated control action through the Actor network, and inputting the simulated effluent water quality data, simulated control action, historical effluent water quality data and historical control instruction data into the Critic network to obtain a simulated instantaneous score value; adjusting the weight parameters of the Actor network according to the difference between the instantaneous score value and the target score until the instantaneous score value reaches the target score.

[0013] In one embodiment of the present application, the control strategy is dynamically adjusted according to the effluent water quality data, specifically including: inputting the facility or process control instructions and the effluent water quality data into the Critic network to generate a feedback score for the facility or process control instructions; adjusting the weight parameters of the Actor network through the back propagation algorithm based on the deviation value between the feedback score and the target score; and applying the updated control instructions output by the updated Actor network to valve opening adjustment and reagent dosage control to form a multi-parameter closed-loop optimization loop.

[0014] In one embodiment of the present application, after obtaining the water treatment simulation model, the method also includes: associating the water treatment simulation model with the control system of the water treatment plant, specifically including: establishing a data interaction channel between the water treatment simulation model and the water treatment plant control system, so as to transmit the facility or process control instructions to the control system of the water treatment plant in real time through the data interaction channel; in the control system, configuring conversion logic that converts the received facility or process control instructions into device executable instructions, and configuring execution feedback logic to feed back the effluent water quality data to the water treatment simulation model after the facility or process control instructions are executed.

[0015] In a second aspect, an embodiment of the present application also provides a water treatment device based on simulation and deep learning, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a water treatment method based on simulation and deep learning such as any one of the above items.

[0016] In a third aspect, an embodiment of the present application further provides a non-volatile computer storage medium storing computer-executable instructions. When the computer-executable instructions are executed, a water treatment method based on simulation and deep learning as described above is implemented.

[0017] The embodiments of the present application provide a water treatment method, device, and medium based on simulation and deep learning, which have the following beneficial effects: By applying deep learning algorithms and simulation theory, the proposed intelligent water treatment method based on deep learning and simulation theory provides a solution for water plant automation and intelligent control for urban water supply systems. This can significantly improve the automation and intelligence level of the water supply process, optimize water treatment efficiency, reduce energy consumption and costs, and enhance the safety of water supply and the stability of water purification effects. In addition, the present invention can bring long-term environmental and economic benefits to urban water supply systems and provide a new path for the sustainable development of the water supply industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A flow chart of a water treatment method based on simulation and deep learning provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the internal structure of a water treatment device based on simulation and deep learning is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] The embodiments of the present application provide a water treatment method, equipment, and medium based on simulation and deep learning to solve the following technical problems: insufficient automation level, unstable water purification effect, and low water production efficiency in urban water supply systems.

[0023] The technical solutions proposed in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0024] Figure 1 This is a flow chart of a water treatment method based on simulation and deep learning provided in the embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides a water treatment method based on simulation and deep learning, which specifically includes the following steps:

[0025] Step 101: Obtain design data related to water treatment plant facilities, and simulate and model the entire water treatment process facilities based on the design data to obtain a facility simulation model.

[0026] In one embodiment of the present application, based on the design data, the water treatment full-process facilities are simulated and modeled to obtain a facility simulation model, specifically including: based on the design data, constructing a 3D model of each facility in the water treatment full-process, and defining the connection relationship between the 3D models of each facility; based on the connection relationship, simulating the internal structure and connection method of each 3D model of the facility to complete the geometric accuracy and topological relationship of the facility simulation model.

[0027] Step 102: Acquire process data related to the water treatment plant process, and perform reaction simulation on the facility simulation model based on the process data to obtain a water treatment simulation model.

[0028] In one embodiment of the present application, a reaction simulation is performed on a facility simulation model based on process data to obtain a water treatment simulation model, specifically including: performing an operation simulation on the facility simulation model based on the process data, and constructing a water treatment mathematical model based on water treatment principles and process data to simulate the treatment effect of each water treatment node; obtaining historical water treatment data, and determining the simulation parameters in the water treatment mathematical model based on the historical water treatment data; wherein the historical water treatment data includes: historical water quality data, historical equipment status data and historical control instruction data; the historical water quality data includes: historical raw water quality data and historical effluent water quality data.

[0029] In one embodiment of the present application, a water treatment mathematical model is constructed based on water treatment principles and process data, specifically including: constructing a physical reaction model according to the fluid dynamics characteristics of the coagulation and sedimentation process; constructing a chemical reaction model based on the chemical kinetics principles of the disinfection and dosing process; embedding the physical reaction model and the chemical reaction model into the facility simulation model to simulate the real-time water quality changes of the raw water after passing through each treatment node.

[0030] In this embodiment, in terms of intelligent control of water production equipment (corresponding to steps 101 and 102 above), this application proposes a water treatment process simulation method based on task flow to realize the simulation of the entire water production process of the water plant. The simulation requires sorting out the standard water production process flow, performing high-precision modeling according to the standard working scenarios of each process, and building a fluid dynamics model to simulate the hydraulic scheduling process between each process in the water plant. At the same time, based on the working mechanism of each process, a water treatment physical-chemical model is designed, and on this basis, comprehensive monitoring of the water quality status, water balance, water scheduling, and water treatment of the entire water production process of the water plant is achieved. The research on water treatment process simulation based on task flow mainly includes simulation modeling of the entire water production process, construction of physical-chemical models that conform to the working principles of each water production process, and efficient simulation calculation of mechanism models. Water treatment process simulation involves a large number of physical-chemical reactions. In order to more accurately simulate the operating status of water quality, water quantity, and water conditions in the process from raw water to drinkable water, the present invention first geometrically models the physical entities such as various water tanks, wells, valves, etc. in the treatment process within the water plant. Then, based on the working principles of each process, a physical-chemical model is designed to model and analyze the water treatment process reactions in steps and tasks.

[0031] Specifically, the first step is to simulate and model the entire water treatment process. The simulation modeling of the entire water treatment process aims to accurately and completely express and present the structure, layout and process flow of the water treatment facilities in a digital way, providing a real modeling scenario for the next step of water treatment process reaction simulation. First, all design data related to the water treatment facilities are collected, including equipment specifications, pipeline layout, process flow charts, etc.; secondly, based on the collected data, a 3D model of the water treatment facilities is established, including drawing various components such as equipment, pipelines, valves, and defining the connection relationship between them; then, based on the preliminary model, more details are added, such as the internal structure of the equipment, the direction and connection method of the pipeline, etc., to ensure the geometric accuracy of the model and the correctness of the topological relationship; finally, it is compared with the actual design drawings, on-site photos, etc. to verify the accuracy of the water plant geometric model.

[0032] The second is the water treatment process reaction simulation based on the physical-chemical model. The water treatment process reaction simulation based on the physical-chemical model is a complex process involving multiple links and steps. First, relevant data is collected for the entire process of the water plant, including raw water quality, design parameters of each water tank in the water plant, equipment performance, operating conditions, etc.; then, according to the water treatment principles and process flow, a mathematical model is established to describe the treatment effects and relationships of each link, as well as the treatment process and simulation requirements. These models include but are not limited to physical models, chemical models or biological models; then, the parameters in the model are determined based on the collected data, and the model is verified using actual data to ensure that the model can accurately simulate the actual water treatment process; finally, the verified model is used for simulation operation to simulate the data changes of water quality status, water balance and water scheduling in the water treatment process under different conditions.

[0033] Step 103: Obtain raw water quality data and equipment status data of the water treatment plant in real time, and input the raw water quality data and equipment status data into a pre-trained deep reinforcement learning controller to generate facility or process control instructions.

[0034] In one embodiment of the present application, the method also includes: pre-training a deep reinforcement learning controller, specifically including: constructing a deep reinforcement learning network framework based on the Actor-Critic framework; wherein the Actor-Critic framework includes: an Actor network and a Critic network; defining a target score for the convergence of the Actor network training and defining a time difference error threshold for the convergence of the Critic network training; based on historical water treatment data, training the deep reinforcement learning network framework until a converged deep reinforcement learning control is obtained.

[0035] In one embodiment of the present application, a deep reinforcement learning network framework is trained based on historical water treatment data, specifically including: constructing a virtual training environment based on a water treatment simulation model, and inputting historical raw water quality data into the virtual training environment to generate a simulated water quality state; according to the simulated water quality state, outputting a simulated control action through the Actor network, and inputting the simulated effluent water quality data, simulated control action, historical effluent water quality data and historical control instruction data into the Critic network to obtain a simulated instantaneous score value; adjusting the weight parameters of the Actor network according to the difference between the instantaneous score value and the target score until the instantaneous score value reaches the target score.

[0036] In this embodiment, raw water quality data refers to a set of physical and chemical parameters monitored in real time from the water plant inlet, including but not limited to turbidity, pH value, residual chlorine concentration, ammonia nitrogen content and temperature parameters. These data are periodically collected by IoT devices such as multispectral sensors and ion-selective electrodes deployed on water pipelines, and uploaded to the central control system via the water plant data bus. Equipment status data covers key equipment operating parameters such as valve opening percentage, water pump speed, membrane filtration pressure, etc., and are fed back in real time by the PLC controller and SCADA system. It can be understood that such dynamic data together constitute the input state space of the deep reinforcement learning controller, and its data specification and preprocessing logic are directly related to the historical water treatment data structure to ensure the homology of training data and application data.

[0037] It should be noted that the core architecture of the pre-trained deep reinforcement learning controller uses the Actor-Critic framework to achieve decision optimization. Specifically, the Actor network acts as a strategy generator, receives a multi-dimensional feature vector formed by the fusion of water quality data and equipment status data (for example, the current composite state of increased turbidity accompanied by insufficient valve opening), and outputs control instructions in the continuous action space. For example, when an abnormal ammonia nitrogen concentration in the inlet water is detected, the Actor network may generate two types of instructions: one is a process control instruction, such as adjusting the frequency conversion parameters of the disinfectant dosing pump; the other is a facility control instruction, such as adjusting the opening percentage of the sedimentation tank outlet valve. This multi-variable collaborative decision-making mechanism responds to the subsequent multi-parameter closed-loop control requirements.

[0038] It's understandable that the controller's pre-training process involves building a virtual environment based on a water treatment simulation model, enabling the Actor network to learn from historical water quality fluctuations and establish a mapping relationship between "water quality-equipment" states and optimal control actions. For example, when simulating a high-turbidity raw water shock in the simulation environment, the network, through trial and error, learns the decision sequence of "opening valves first, then slowly adjusting the reagent," avoiding the wasteful practice of directly increasing the reagent dosage in traditional control. It's important to emphasize that in actual applications, the controller directly calls upon the pre-set network weight parameters, eliminating the need for online training and ensuring real-time response speed.

[0039] Specifically, the generation of control instructions follows physical constraints: for example, valve opening instructions are limited to the equipment's safe operating range (0-100%), and agent dosage instructions comply with the maximum dosage threshold. This constraint mechanism, embedded in the Actor network output layer, ensures that instructions are within the equipment's executable range. This design effectively resolves potential equipment conflicts during the coordinated control of multiple processes, for example, preventing the risk of pipeline overpressure caused by simultaneously executing "accelerating the pump speed" and "closing the valve."

[0040] Step 104: Input the facility or process control instructions into the water treatment simulation model for execution, and monitor the effluent water quality data in real time to dynamically adjust the control strategy according to the effluent water quality data.

[0041] In one embodiment of the present application, after obtaining the water treatment simulation model, the method also includes: associating the water treatment simulation model with the control system of the water treatment plant, specifically including: establishing a data interaction channel between the water treatment simulation model and the water treatment plant control system, so as to transmit the facility or process control instructions to the control system of the water treatment plant in real time through the data interaction channel; in the control system, configuring conversion logic that converts the received facility or process control instructions into device executable instructions, and configuring execution feedback logic to feed back the effluent water quality data to the water treatment simulation model after the facility or process control instructions are executed.

[0042] In one embodiment of the present application, the control strategy is dynamically adjusted according to the effluent water quality data, specifically including: inputting the facility or process control instructions and the effluent water quality data into the Critic network to generate a feedback score for the facility or process control instructions; adjusting the weight parameters of the Actor network through the back propagation algorithm based on the deviation value between the feedback score and the target score; and applying the updated control instructions output by the updated Actor network to valve opening adjustment and reagent dosage control to form a multi-parameter closed-loop optimization loop.

[0043] In this embodiment, the water treatment simulation model serves as a digital twin operation carrier, and its core is composed of a physical reaction model and a chemical reaction model coupled. Specifically, when the facility or process control instruction generated in step 103 (such as "increase the valve opening of the disinfection contact pool" or "reduce the flocculant dosage") is input into the model, the model responds dynamically based on the following mechanism: the physical reaction model simulates the water flow velocity field distribution caused by the change in valve opening according to the fluid dynamics equation, and the chemical reaction model calculates the agent concentration diffusion and reaction rate through the chemical kinetics formula. For example, if the instruction requires an increase in the ozone dosage, the model will simulate the dissolution efficiency of ozone in the water body and its oxidation reaction process with organic matter, and predict the trend of changes in the residual chlorine concentration in the effluent water.

[0044] It is understandable that the simulated effluent water quality data (such as predicted turbidity, residual chlorine value, etc.) generated after the model executes the instructions needs to be compared with the real-time monitoring effluent water quality data of the actual production link. This real-time data is collected by online water quality instruments installed at the outlet of the sedimentation tank, the back end of the filter tank and the clear water tank (such as ultraviolet spectrometer, turbidity sensor), and its data type is exactly the same as the historical effluent water quality data. It should be noted that this step forms the basis of the dynamic adjustment strategy: when the actual effluent residual chlorine concentration continues to be lower than the simulation prediction value, it indicates that there is room for optimization of the disinfection process control instructions.

[0045] Specifically, the process of dynamically adjusting the control strategy: first, the actual effluent water quality data and the executed control instructions are input into the Critic network, and the network outputs a quantitative evaluation based on the preset scoring rules. For example, if the actual residual chlorine concentration deviates from the predicted range and is lower than the national standard limit, the Critic network will give a low score. The score is calculated by the time difference error (TD-error) to calculate the deviation from the target score, triggering the backpropagation algorithm to adjust the weight parameters of the Actor network. This mechanism makes the network more inclined to choose actions such as "increasing the disinfection dose" or "extending the contact time" in subsequent decision-making, rather than the original invalid instructions.

[0046] It should be emphasized that the adjusted control strategy acts on the physical system in real time through the established data interaction channel: the updated Actor network outputs new instructions (such as "increase the frequency of the sodium hypochlorite dosing pump"), which are directly sent to the field equipment through the control system's instruction conversion logic (such as the OPC-UA protocol). At the same time, the execution effect of the new instructions is fed back to the water treatment simulation model, forming a closed-loop optimization circuit of "instruction execution → water quality monitoring → strategy iteration → instruction update". For example, when heavy rain causes fluctuations in raw water turbidity, the system can complete multiple rounds of strategy tuning within half an hour until the effluent water quality is stable and meets the standards.

[0047] This closed-loop mechanism is understandably well-suited for multi-process collaboration. For example, when adjusting the valve opening in the coagulation process, the simulation model simultaneously calculates its impact on the load of the subsequent filtration process. If the prediction indicates that the filter pressure differential will exceed the limit, a compensation instruction to "reduce the inlet flow" is dynamically generated. This cross-process collaborative optimization effectively addresses the overall efficiency decline caused by single-point control in traditional water plants.

[0048] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, this application embodiment also provides a water treatment device based on simulation and deep learning, the structure of which is as follows Figure 2 shown.

[0049] Figure 2 This is a schematic diagram of the internal structure of a water treatment device based on simulation and deep learning provided in an embodiment of the present application. Figure 2 As shown, the equipment includes:

[0050] at least one processor 201;

[0051] and, a memory 202 communicatively coupled to the at least one processor;

[0052] The memory 202 stores instructions that can be executed by at least one processor, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to:

[0053] Obtain design data related to water treatment plant facilities, and based on the design data, simulate and model the water treatment facilities throughout the entire process to obtain a facility simulation model; obtain process data related to the water treatment plant procedures, and based on the process data, perform reaction simulation on the facility simulation model to obtain a water treatment simulation model; obtain the raw water quality data and equipment status data of the water treatment plant in real time, and input the raw water quality data and equipment status data into a pre-trained deep reinforcement learning controller to generate facility or process control instructions; input the facility or process control instructions into the water treatment simulation model for execution, and monitor the effluent water quality data in real time to dynamically adjust the control strategy based on the effluent water quality data.

[0054] Some embodiments of the present application provide corresponding Figure 1 A non-volatile computer storage medium stores computer executable instructions, wherein the computer executable instructions are configured as follows:

[0055] Obtain design data related to water treatment plant facilities, and based on the design data, simulate and model the water treatment facilities throughout the entire process to obtain a facility simulation model; obtain process data related to the water treatment plant procedures, and based on the process data, perform reaction simulation on the facility simulation model to obtain a water treatment simulation model; obtain the raw water quality data and equipment status data of the water treatment plant in real time, and input the raw water quality data and equipment status data into a pre-trained deep reinforcement learning controller to generate facility or process control instructions; input the facility or process control instructions into the water treatment simulation model for execution, and monitor the effluent water quality data in real time to dynamically adjust the control strategy based on the effluent water quality data.

[0056] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other. Each embodiment focuses on the differences from the other embodiments. In particular, the IoT device and media embodiments are generally similar to the method embodiments, so their description is relatively simple. For relevant portions, refer to the description of the method embodiments.

[0057] The system and medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the system and medium also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0058] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0059] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0060] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0062] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0063] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0064] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0065] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0066] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A water treatment method based on simulation and deep learning, characterized in that: The method comprises: Acquiring design data related to water treatment plant facilities, and based on the design data, performing simulation modeling of the water treatment full-process facilities to obtain a facility simulation model; Acquiring process data related to a process of a water treatment plant, and performing a reaction simulation on the facility simulation model based on the process data to obtain a water treatment simulation model; Obtaining raw water quality data and equipment status data from a water treatment plant in real time, and inputting the raw water quality data and equipment status data into a pre-trained deep reinforcement learning controller to generate facility or process control instructions; The facility or process control instructions are input into the water treatment simulation model for execution, and the effluent water quality data is monitored in real time to dynamically adjust the control strategy according to the effluent water quality data.

2. A water treatment method based on simulation and deep learning according to claim 1, characterized in that: Based on the design data, a full-process water treatment facility simulation model is constructed to obtain a facility simulation model, specifically including: Based on the design data, construct 3D models of each facility in the entire water treatment process and define the connection relationship between the 3D models of each facility; Based on the connection relationship, the internal structure and connection mode of each facility 3D model are simulated to complete the geometric accuracy and topological relationship of the facility simulation model.

3. The water treatment method based on simulation and deep learning according to claim 1, characterized in that: Based on the process data, a reaction simulation is performed on the facility simulation model to obtain a water treatment simulation model, specifically comprising: Based on the process data, the facility simulation model is run simulated, and a water treatment mathematical model is constructed based on water treatment principles and the process data to simulate the treatment effect of each water treatment node; Acquire historical water treatment data, and determine simulation parameters in the water treatment mathematical model based on the historical water treatment data; wherein the historical water treatment data includes: historical water quality data, historical equipment status data and historical control instruction data; the historical water quality data includes: historical raw water quality data and historical effluent water quality data.

4. A water treatment method based on simulation and deep learning according to claim 3, characterized in that: A water treatment mathematical model is constructed based on the water treatment principle and the process data, specifically including: A physical reaction model is constructed based on the fluid dynamics characteristics of the coagulation and sedimentation process; Construct a chemical reaction model based on the chemical kinetics principle of the disinfection and dosing process; The physical reaction model and the chemical reaction model are embedded in the facility simulation model to simulate the real-time water quality changes of the raw water after passing through each treatment node.

5. The water treatment method based on simulation and deep learning according to claim 3, characterized in that: The method further comprises: Pre-training the deep reinforcement learning controller includes: Constructing a deep reinforcement learning network framework based on the Actor-Critic framework; wherein the Actor-Critic framework includes: an Actor network and a Critic network; Defining a target score for the Actor network training convergence and a time difference error threshold for the Critic network training convergence; Based on the historical water treatment data, the deep reinforcement learning network framework is trained until converged deep reinforcement learning control is obtained.

6. The water treatment method based on simulation and deep learning according to claim 5, characterized in that: Based on the historical water treatment data, training the deep reinforcement learning network framework specifically includes: Constructing a virtual training environment based on the water treatment simulation model, and inputting the historical raw water quality data into the virtual training environment to generate a simulated water quality state; According to the simulated water quality state, the simulated control action is output through the Actor network, and the simulated effluent water quality data, the simulated control action, the historical effluent water quality data and the historical control instruction data are input into the Critic network to obtain a simulated real-time score value; The weight parameters of the Actor network are adjusted according to the difference between the instantaneous score and the target score until the instantaneous score reaches the target score.

7. The water treatment method based on simulation and deep learning according to claim 5, characterized in that: Dynamically adjust the control strategy according to the effluent water quality data, specifically including: Inputting the facility or process control instruction and the effluent water quality data into a Critic network to generate a feedback score for the facility or process control instruction; Based on the deviation between the feedback score and the target score, the weight parameters of the Actor network are adjusted through the back-propagation algorithm; The updated control instructions output by the updated Actor network are applied to valve opening adjustment and reagent dosage control to form a multi-parameter closed-loop optimization circuit.

8. The water treatment method based on simulation and deep learning according to claim 1, characterized in that: After obtaining the water treatment simulation model, the method further includes: Associating the water treatment simulation model with the control system of the water treatment plant includes: Establishing a data interaction channel between the water treatment simulation model and the water treatment plant control system, so as to transmit the facility or process control instructions to the water treatment plant control system in real time through the data interaction channel; In the control system, a conversion logic is configured to convert the received facility or process control instructions into equipment executable instructions, and an execution feedback logic is configured to feed back the effluent water quality data to the water treatment simulation model after the facility or process control instructions are executed.

9. A water treatment device based on simulation and deep learning, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; Wherein, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a water treatment method based on simulation and deep learning as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: When the computer-executable instructions are executed, a water treatment method based on simulation and deep learning as described in any one of claims 1 to 8 is implemented.

Images