Advanced oxidation water treatment disinfection system and method thereof

Abstract

The application relates to an advanced oxidation water treatment disinfection system and a method thereof, and belongs to the technical field of water treatment control. The system comprises a modular reaction unit array, a distributed sensing unit, an intelligent control unit and a collaborative execution unit. The modular reaction unit array comprises at least two parallelly arranged pipeline AOP reactors with the same structure. The intelligent control unit is used for generating initial process parameters through feedforward prediction, dynamically correcting and resource scheduling optimization in combination with feedback of mixed states and water outlet effects, finally generating and issuing control instructions to drive collaborative actions of each execution mechanism. The application realizes flexible adjustment of a wide range of processing capacity through modular array and intelligent scheduling, guarantees uniform mixing and eliminates disinfection dead angles through distributed monitoring and active disturbance, and dynamically optimizes key parameters through intelligent closed-loop control to ensure effects while significantly reducing drug consumption and energy consumption.

Description

Technical Field

[0001] This invention belongs to the field of water treatment control technology, specifically relating to an advanced oxidation water treatment disinfection system and method. Background Technology

[0002] Advanced oxidation processes (AOP) are widely used in water treatment and disinfection because they generate highly oxidizing hydroxyl radicals, which can effectively degrade recalcitrant pollutants and inactivate pathogenic microorganisms. However, existing AOP disinfection technologies and devices have the following main drawbacks: (1) Inefficient control and poor adaptability: Most systems use fixed parameters such as oxidant dosage and reaction time, which cannot be precisely adjusted according to the dynamic changes in influent water quality and flow rate in real time. This can easily lead to incomplete disinfection when the water quality is complex or the flow rate is at its peak, while it can cause waste of oxidant when the water quality is stable or the load is low.

[0003] (2) Uneven mixing and unstable effect: Especially in large-capacity treatment units, it is difficult to achieve rapid and uniform mixing between the oxidant and the water to be treated, resulting in low efficiency of hydroxyl radical generation and uneven spatial distribution, with dead corners where local disinfection is insufficient, affecting the stability of the overall treatment effect.

[0004] (3) Lack of feedback and lagging adjustment: Existing systems generally lack effective real-time process monitoring and closed-loop feedback adjustment mechanisms, making it impossible to grasp the disinfection effect in a timely manner and dynamically correct process parameters. The system has weak anti-interference ability and large fluctuations in effluent water quality.

[0005] Therefore, there is an urgent need to develop an AOP-controlled disinfection system that can dynamically adapt to changes in water quality and quantity, ensure a uniform and efficient reaction system, and achieve real-time monitoring and intelligent feedback of the process. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides an advanced oxidation water treatment and disinfection system and method. It achieves wide-range flexible adjustment of treatment capacity through modular arrays and intelligent scheduling; ensures uniform mixing and eliminates disinfection dead zones through distributed monitoring and active perturbation; and dynamically optimizes key parameters through intelligent closed-loop control, significantly reducing chemical and energy consumption while ensuring effectiveness.

[0007] The objective of this invention can be achieved through the following technical solutions: The first aspect of this disclosure provides an advanced oxidation water treatment and disinfection system, including a modular reaction unit array, a distributed sensing unit, an intelligent control unit, and a collaborative execution unit; The modular reaction unit array includes at least two parallel, identical pipe-type AOP reactors. Each reactor has a premixing section, a main reaction section, and a flow stabilizing section arranged sequentially along the water flow direction. The main reaction section integrates an ultraviolet catalytic component. Each reactor is independently configured with an inlet regulating valve, an oxidant dosing branch, and a circulation disturbance interface. The intelligent control unit is configured as an industrial controller based on an edge computing architecture. The industrial controller has a hybrid intelligent control model built inside, which is used to generate initial process parameters through feedforward prediction, and to perform dynamic correction and resource scheduling optimization by combining feedback from the mixed state and the water output effect. Finally, it generates and issues control commands to drive the coordinated action of each actuator. The hybrid intelligent control model includes: Feedforward prediction sub-model: Based on the real-time influent flow and water quality parameters of the influent monitoring module, a pre-trained machine learning model is used to obtain feedforward prediction instructions; Feedback correction sub-model: Based on the ORP spatial distribution data obtained by the monitoring modules of each reactor unit, the overall mixing uniformity index is calculated, and the feedforward prediction results are dynamically corrected according to the mixing uniformity index and the feedback data of the effluent monitoring module to generate the final control instruction set. Scheduling optimization sub-model: Based on the real-time total influent flow rate and the operating status of each reactor, a multi-objective optimization algorithm is used to dynamically determine the number of reactors to be activated, the flow allocation ratio of each reactor, and the operating mode.

[0008] As a preferred technical solution of the present invention, the distributed sensing unit is used to continuously collect influent water quality and quantity data, dynamically monitor the spatial distribution of oxidant inside each reactor, and simultaneously track the final effluent disinfection effect indicators, and transmit multi-source real-time sensing data to the intelligent control unit.

[0009] As a preferred embodiment of the present invention, the distributed sensing unit includes an influent monitoring module, a unit monitoring module, and an effluent monitoring module, wherein: The influent monitoring module, located in the main influent pipeline, is used to monitor the influent flow rate and water quality parameters in real time. The water quality parameters include at least chemical oxygen demand, total organic carbon, turbidity, and pH value. The unit monitoring module, corresponding to each reactor, includes at least two redox potential probes located in the main reaction section to monitor the uniformity index of oxidant distribution inside the reactor. The effluent monitoring module, located in the main effluent pipeline, is used to monitor effluent indicators in real time, including the concentration of disinfection byproduct precursors and the concentration of microbial indicators.

[0010] As a preferred embodiment of the present invention, the step of obtaining the feedforward prediction instruction includes the following steps: Real-time acquisition and construction of multi-dimensional influent features: After cleaning the real-time influent flow rate and water quality parameters, derived features are obtained through calculation. Then, all features are standardized to form a unified real-time feature vector. Synchronous prediction based on a multi-task learning architecture: Real-time feature vectors are input into a pre-trained machine learning model. The machine learning model adopts a multi-task learning structure with a shared feature extraction layer and dual prediction branches. The shared feature extraction layer is responsible for extracting high-order, abstract pattern information from the features. The dual prediction branches simultaneously infer the initial value of oxidant demand and the recommended hydraulic residence time based on the shared information.

[0011] As a preferred embodiment of the present invention, the feedback correction sub-model for generating the final control instruction set includes the following steps: Multi-source data synchronization and fusion: synchronously receive feedforward prediction commands, real-time process monitoring data and system status data. Among them, the feedforward prediction commands include the initial value of oxidant demand and the recommended hydraulic residence time, the real-time process monitoring data includes the redox potential values ​​and effluent water quality at multiple points inside each reactor, and the system status data includes the current operating conditions of each actuator. Deviation analysis: The spatial mixing uniformity index is calculated based on the oxidation-reduction potential value. If the spatial mixing uniformity index is lower than the set threshold, it indicates that the oxidant is unevenly distributed in the reaction system. At the same time, the effluent water quality data is compared with the target value to evaluate the actual disinfection effect of the current cycle. Parameter optimization and correction: Based on real-time mixing status and disinfection effect feedback, the total amount of oxidant added, mixing intensity and process parameters are adjusted in a coordinated manner; Among them, the correction of oxidant dosage is as follows: the initial oxidant requirement is compensated according to the mixing uniformity. If the mixing is not up to standard, the total dosage is increased by a certain proportion. At the same time, the distribution ratio of each reactor's addition branch is adjusted, and units with poor mixing are given preferential treatment. Adjustment of mixing intensity: Based on the mixing uniformity and ORP data of each unit, generate adjustment commands for the circulation pump frequency or stirring intensity; Coordinated fine-tuning of process parameters: Based on the actual disinfection effect of the effluent and the state of the reactor, the hydraulic retention time and ultraviolet intensity parameters are coordinated and fine-tuned. Generate the final control instruction set: Package all the corrected parameters into a unified final control instruction set, which includes execution instructions for the oxidant dosing subsystem, fluid preparation subsystem, and ultraviolet catalysis subsystem, and send them to the collaborative execution unit in real time.

[0012] As a preferred embodiment of the present invention, the scheduling optimization sub-model dynamically determines the number of reactors to be activated, the flow allocation ratio of each reactor, and the operating mode through a multi-objective optimization algorithm, including the following steps: Resource status assessment: The theoretical total reaction volume is calculated based on the product of the recommended hydraulic retention time and the total influent flow rate. At the same time, the availability status and rated volume of all modular reactor units are polled in real time to generate a list of available resources. Scheduling scheme generation: Based on the theoretical total reaction volume and available resource list, multi-objective optimization calculations are initiated to execute optimization decisions, which include: Determine the number and combination of active units: Calculate the number of modular reactor units that satisfy the theoretical total reaction volume, and select the optimal combination from the available units; Decision-making operation connection mode: Select the parallel or series mode of the reactor array based on the characteristics of the influent water quality and the recommended hydraulic retention time; Flow distribution scheme: In parallel mode, the influent flow is dynamically distributed according to the real-time status of each activated unit.

[0013] As a preferred technical solution of the present invention, the collaborative execution unit drives a multi-channel metering pump, an electric regulating valve, a variable frequency circulating pump and an ultraviolet lamp power supply based on a control instruction set, thereby controlling the oxidant dosage, water flow distribution, internal circulation disturbance intensity and ultraviolet catalytic intensity respectively.

[0014] As a preferred embodiment of the present invention, the cooperative execution unit includes an oxidant dosing subsystem, a fluid dynamic adjustment subsystem, and an ultraviolet catalytic intensity regulation subsystem, wherein: The oxidant dosing subsystem receives control commands generated by the intelligent control unit and drives a multi-channel variable frequency metering pump to add the calculated total oxidant dose in an unbalanced manner according to the flow rate ratio of each reactor and the mixing uniformity compensation coefficient. The fluid dynamic distribution subsystem receives control commands generated by the intelligent control unit, controls the opening degree of the inlet water regulating valve of each reactor to distribute the flow, and controls the speed of the variable frequency circulating pump connected to the reactor circulation disturbance interface to regulate the internal flow state. The ultraviolet catalytic intensity regulation subsystem adjusts the power or the number of ultraviolet lamps turned on or off in each reactor by receiving control commands generated by the intelligent control unit.

[0015] A second aspect of this disclosure provides an advanced oxidation water treatment disinfection method, applied to an advanced oxidation water treatment disinfection system as described above, comprising the following steps: S1. Synchronous acquisition and feature engineering of multi-source data: Real-time data from distributed sensing units are acquired synchronously, including influent water quality parameters, measured values ​​of multiple redox potential probes inside each activated reactor, and concentration of disinfection byproduct precursors in the effluent; then, real-time feature engineering is performed on the influent data. After data cleaning, derived features are calculated and standardized to generate real-time feature vectors for model input. S2. Feedforward prediction to generate initial process baseline: Input the real-time feature vector into the feedforward prediction sub-model, extract high-order abstract information from the features through the shared feature extraction layer of the feedforward prediction sub-model, and then output the feedforward prediction instruction containing the initial value of oxidant demand and the recommended hydraulic residence time based on the dual prediction branches, which together constitute the initial process parameter set. S3. Scheduling optimization decision-making resource allocation: The scheduling optimization sub-model receives the total influent flow rate and recommended hydraulic retention time, calculates the required theoretical total reaction volume, and evaluates the availability status of all modular reactor units in real time. Based on the principle of multi-objective optimization, it dynamically decides the optimal number and combination of reactor units to be activated, determines the parallel or series operation mode, and formulates a preliminary flow allocation scheme, thereby forming a scheduling instruction set. S4. Feedback Correction and Parameter Fine-tuning: The feedback correction sub-model synchronously receives initial process parameters, scheduling instruction set, real-time oxidation-reduction potential spatial distribution data, effluent water quality data, and the status of each actuator. It calculates the mixing uniformity index within the reaction system by analyzing the oxidation-reduction potential values ​​and integrates multiple effluent indicators to evaluate the real-time disinfection effect. Simultaneously, it compensates for the total oxidant dosage based on the mixing state and optimizes the distribution of each branch, diagnoses mixing defect modes to directionally adjust the intensity of circulation disturbance, and fine-tunes the hydraulic residence time and ultraviolet catalytic intensity in conjunction with the effluent effect, ultimately generating a final control instruction set containing all execution instructions. S5. Collaborative Execution: The final control instruction set and scheduling instructions are synchronously sent to the collaborative execution unit. The oxidant precision dosing subsystem drives the multi-channel metering pump to perform dosing; the fluid dynamic allocation subsystem controls the inlet water regulating valve and circulation disturbance pump of each reactor; and the ultraviolet catalytic intensity regulation subsystem adjusts the power of each ultraviolet lamp group.

[0016] Furthermore, it also includes the following steps: Effect evaluation and model self-learning: After a control cycle ends, the disinfection effect is re-evaluated based on the latest effluent data. If the effect continues to deviate from the target, the key parameters in the feedback correction sub-model are automatically fine-tuned. At the same time, the effective complete data in this cycle will be stored in the case database for periodic retraining of the feedforward prediction sub-model.

[0017] The beneficial effects of this invention are as follows: This invention achieves flexible adjustment of treatment capacity over a wide range by using modular reactor arrays and intelligent scheduling optimization, adapting to large flow fluctuations and solving the problem of narrow adjustment range of fixed-volume devices. Secondly, by introducing a combination of distributed oxidation-reduction potential monitoring and active circulation disturbance control, real-time sensing and active optimization of the mixing state inside the reactor are achieved, ensuring uniform mixing of oxidant and pollutants, significantly improving the utilization efficiency of hydroxyl radicals, and eliminating disinfection dead zones. Finally, through a hybrid intelligent control model composed of feedforward prediction, feedback correction, and self-learning mechanisms, dynamic and precise control and closed-loop optimization of key process parameters such as oxidant dosing and hydraulic retention time are achieved. This allows the system to automatically adjust to the optimal operating state based on real-time influent water quality and quantity, ensuring stable and compliant disinfection while significantly reducing reagent and energy consumption. Attached Figure Description

[0018] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0019] Figure 1 This is a schematic diagram of the structure of an advanced oxidation water treatment and disinfection system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the steps for obtaining feedforward prediction instructions provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the steps of an advanced oxidation water treatment and disinfection method provided in an embodiment of the present invention. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0021] This embodiment provides an advanced oxidation water treatment and disinfection system, such as Figure 1 As shown, it includes a modular reaction unit array, a distributed sensing unit, an intelligent control unit, and a collaborative execution unit.

[0022] The modular reaction unit array includes at least two parallel, identical pipe-type AOP reactors. Each reactor has a premixing section, a main reaction section, and a flow stabilization section arranged sequentially along the water flow direction. The main reaction section integrates an ultraviolet catalytic component. Each reactor is independently configured with an inlet regulating valve, an oxidant dosing branch, and a circulation disturbance interface.

[0023] Understandably, the circulation disturbance interface is a dedicated hardware channel and its control system independently configured on each modular reactor unit for actively intervening in and optimizing the internal fluid mixing state. The circulation disturbance interface consists of at least one pair of connection ports located on the sidewall or end of the reactor, connected to the inlet and outlet of a variable frequency centrifugal pump via external piping, forming a side circulation pipeline that can be independently opened, closed, and its flow rate adjusted. By adjusting the pump's rotational speed, the circulation shear intensity and flow pattern of the fluid inside the reactor can be actively altered. In this embodiment, by introducing the circulation disturbance interface, the system transforms from a reaction vessel with a fixed geometry into an intelligent reaction unit with a state-sensing and internally adjustable environment, actively optimizing the mixing state of the oxidant and pollutants, enhancing the mass transfer process, and eliminating "dead zone" problems caused by flow rate changes, uneven dosing, or reactor structure.

[0024] It should be noted that the premixing section is located at the very front of the reactor, immediately adjacent to the inlet and oxidant dosing port. It is equipped with a static mixer or jet nozzles to vigorously mix the influent with the reagent injected through the oxidant dosing branch, achieving instantaneous and high-intensity initial mixing of the oxidant and raw water. The main reaction section, located after the premixing section, is the core part of the reactor with the largest volume and longest length. It integrates ultraviolet catalytic components (such as UV lamps of specific wavelengths and TiO2 catalysts supported on a carrier). Under ultraviolet photocatalysis, it efficiently and continuously generates hydroxyl radicals, completing the oxidative degradation of pollutants and the inactivation of pathogenic microorganisms. The flow stabilization section is located at the end of the reactor, before the outlet. It is equipped with a perforated plate or flow straightening grid to uniformly distribute the flow velocity, smooth the water flow, stabilize the reaction, and serve as a final monitoring and safety buffer zone.

[0025] The distributed sensing unit is used to continuously collect influent water quality and quantity data, dynamically monitor the spatial distribution of oxidant inside each reactor, and simultaneously track the final effluent disinfection effect indicators, transmitting multi-source real-time sensing data to the intelligent control unit.

[0026] The distributed sensing unit includes an influent monitoring module, a unit monitoring module, and an effluent monitoring module, wherein: The influent monitoring module, located in the main influent pipeline, is used to monitor the influent flow rate and water quality parameters in real time. The water quality parameters include at least chemical oxygen demand, total organic carbon, turbidity, and pH value. The unit monitoring module, corresponding to each reactor, includes at least two oxidation-reduction potential (ORP) probes located in the main reaction section to monitor the uniformity index of oxidant distribution inside the reactor. The effluent monitoring module, located in the main effluent pipeline, is used to monitor key effluent indicators in real time, including the concentration of disinfection byproduct precursors and the concentration of microbial indicators.

[0027] The intelligent control unit is configured as an industrial controller based on an edge computing architecture. The industrial controller has a hybrid intelligent control model built inside, which is used to generate initial process parameters through feedforward prediction, and to perform dynamic correction and resource scheduling optimization by combining feedback from the mixed state and the water output effect. Finally, it generates and issues precise control commands to drive the coordinated action of each actuator.

[0028] The hybrid intelligent control model includes: Feedforward prediction sub-model: Based on the real-time influent flow rate and water quality parameters of the influent monitoring module, a pre-trained machine learning model is used to obtain feedforward prediction instructions.

[0029] The method for obtaining feedforward prediction instructions, such as Figure 2 As shown, it includes the following steps: Real-time acquisition and construction of multi-dimensional influent features: After cleaning the real-time influent flow rate and water quality parameters, derived features (such as pollutant load, rate of change of key water quality parameters, etc.) are obtained through calculation. Then, all features are standardized to form a unified real-time feature vector. Synchronous prediction based on a multi-task learning architecture: Real-time feature vectors are input into a pre-trained machine learning model. The machine learning model adopts a multi-task learning structure with a shared feature extraction layer and dual prediction branches. The shared feature extraction layer is responsible for extracting high-order, abstract pattern information from the features. The dual prediction branches simultaneously infer the initial value of oxidant demand and the recommended hydraulic residence time based on the shared information.

[0030] It should be noted that the pre-training of a machine learning model includes the following steps: Training sample construction: Select steady-state operating periods (historical stable operating periods where the effluent quality consistently meets the standards and key energy and chemical consumption indicators are within the optimization range), extract influent feature vectors, and match them with the process parameters statistically obtained during this period to generate training samples; Multi-task learning model architecture design: The model includes a shared feature extraction layer and dual prediction branches. The shared feature extraction layer consists of several fully connected layers and is responsible for learning and extracting representations that can simultaneously reflect oxidant demand and reaction kinetics features from the influent features. The dual prediction branches include an oxidant demand prediction branch and a residence time prediction branch. The dual prediction branches each use independent neural networks, taking the abstract features output by the shared feature extraction layer as input, and specifically learning to map to continuous values.

[0031] Model Training and Optimization: The loss function is defined as a weighted sum of the losses from the two tasks. Process constraints are incorporated as regularization terms into the loss function; for example, a penalty is imposed on the recommended hydraulic residence time output if it exceeds the physical limits allowed by the reactor array design. Dropout and weight decay techniques are employed to prevent overfitting. The constructed training samples are divided into training, validation, and test sets in a proportional ratio (e.g., 8:1:1), and iterative training is performed using backpropagation and optimizers such as Adam. During training, the validation set is used to monitor the model's generalization ability, and early stopping is used to prevent overfitting.

[0032] Feedback Correction Sub-model: Based on the ORP spatial distribution data obtained by the monitoring modules of each reactor unit, the overall mixing uniformity index is calculated, and the feedforward prediction results are dynamically corrected according to the mixing uniformity index and the feedback data of the effluent monitoring module to generate the final control instruction set.

[0033] The feedback correction sub-model generates the final control instruction set by the following steps: Multi-source data synchronization and fusion: Synchronously receive feedforward prediction commands, real-time process monitoring data, and system status data. The feedforward prediction commands include the initial value of oxidant demand and the recommended hydraulic residence time. The real-time process monitoring data includes the redox potential values ​​and effluent water quality at multiple sites inside each reactor. The system status data includes the current operating status of each actuator (pump, valve, UV lamp).

[0034] Deviation analysis: The spatial mixing uniformity index is calculated based on the oxidation-reduction potential value. If the spatial mixing uniformity index is lower than the set threshold, it indicates that the oxidant is unevenly distributed in the reaction system, and there is a risk of disinfection dead zones. At the same time, the model compares the effluent water quality data with the target value to evaluate the actual disinfection effect of the current cycle.

[0035] The mixing uniformity index is first calculated by using the ratio of the standard deviation to the mean of each collected oxidation-reduction potential value as the coefficient of variation, and then normalized to obtain the mixing uniformity index. The actual disinfection effect is evaluated using a multi-index fusion method. The system monitors key indicators in the effluent in real time: residual oxidant concentration, used to calculate the oxidant utilization efficiency; concentration of specific characteristic pollutants, whose rate of change in influent and effluent reflects the overall effectiveness of the oxidation process; and indirect microbial activity indicators, such as adenosine triphosphate (ATP) concentration, used to infer the degree of microbial inactivation. The measured values ​​of these three indicators are compared with their respective target values, and different weighting coefficients are assigned according to their importance in the process. A comprehensive evaluation value is calculated using a weighted fusion algorithm. The closer this value is to 1, the better the overall disinfection effect.

[0036] Parameter optimization and correction: Based on real-time mixing status and disinfection effect feedback, the total amount and distribution of oxidant, circulation disturbance intensity, hydraulic residence time and ultraviolet intensity are adjusted in a coordinated manner to achieve closed-loop collaborative optimization.

[0037] The correction of oxidant dosage involves compensating for the initial oxidant requirement based on the mixing uniformity. If the mixing is poor, the total dosage is increased by a certain proportion, and the distribution ratio of the dosing branches in each reactor is adjusted to favor units with poor mixing.

[0038] Specifically, there is a negative correlation between mixing uniformity and the required oxidant dosage. Based on the real-time calculated mixing uniformity, the system calculates a compensation coefficient using a predetermined compensation function (usually determined by historical data regression or computational fluid dynamics simulation). The preferred form of this function is: Compensation coefficient β = a / H(t) + b, where a and b are positive coefficients fitted from experimental data, and H(t) is the mixing uniformity exponent. This compensation coefficient is greater than or equal to 1 and increases as mixing uniformity decreases. Multiplying the compensation coefficient by the initial oxidant requirement yields the corrected oxidant dosage.

[0039] Fluid and mixing intensity adjustment: Based on the mixing uniformity and ORP data of each unit, generate adjustment commands for the circulation pump frequency or stirring intensity.

[0040] Specifically, the system diagnoses the specific causes of poor mixing (such as insufficient overall stirring intensity, short-circuit flow, or dead zones) based on the mixing uniformity index and the spatial data distribution patterns of each ORP probe, and generates corresponding hydrodynamic adjustment commands accordingly. Different mixing defect patterns exhibit characteristic characteristics in the spatial distribution of ORP. For example, if all ORP values ​​are somewhat discrete and the mixing uniformity is lower than the set value, it is judged that the overall mixing energy is insufficient, and the command will increase the circulation pump frequency or agitator speed; if the ORP values ​​show a significant gradient along the water flow direction, it indicates the presence of short-circuit flow, and the flow pattern will be improved by adjusting the inlet guide device or increasing the premixing intensity; if only a few probes have abnormally low ORP values, there may be dead zones in the corresponding areas, and the stirring device or jet nozzle in that area will be adjusted individually. The amplitude of the adjustment command is directionally adjusted according to the ORP distribution pattern.

[0041] Coordinated fine-tuning of process parameters: Based on the actual disinfection effect of the effluent and the state of the reactor, parameters such as hydraulic retention time and ultraviolet intensity are coordinated and fine-tuned.

[0042] Specifically, the fine-tuning principle is based on the feedback control concept in process control: if the actual disinfection effect is consistently lower than the target value, and the oxidant utilization rate has reached the standard, it is determined that the reaction time may be insufficient, and the hydraulic retention time will be appropriately extended within the allowable range; if the actual disinfection effect is insufficient and the degradation rate of characteristic pollutants is lower than the threshold, while the influent UV absorption is strong, it is determined that the UV catalytic effect is insufficient, and the UV radiation intensity will be increased. When adjusting the UV intensity, the adjustment amount is compensated by considering the cumulative operating time of the UV lamp and its efficiency decay curve to ensure that the actual irradiation dose reaches the expected level. All fine-tuning follows a multivariate coordinated control strategy to minimize energy and chemical consumption while ensuring the disinfection effect.

[0043] Generate the final control instruction set: Package all the corrected parameters into a unified final control instruction set, which includes execution instructions for the oxidant dosing subsystem, fluid preparation subsystem, and ultraviolet catalysis subsystem (such as the precise flow rate of each metering pump, the opening degree of each regulating valve, the power of each ultraviolet lamp, etc.), and send them to the collaborative execution unit in real time.

[0044] Scheduling optimization sub-model: Based on the real-time total influent flow rate and the operating status of each reactor, a multi-objective optimization algorithm dynamically determines the number of reactors to be activated, the flow allocation ratio of each reactor, and the operating mode, including the following steps: Resource status assessment: The theoretical total reaction volume is calculated based on the product of the recommended hydraulic retention time and the total influent flow rate. At the same time, the availability status (such as whether it is online, whether it is in cleaning or maintenance, and its current load rate) and rated volume of all modular reactor units are polled in real time to generate a list of available resources. Scheduling scheme generation: Based on the theoretical total reaction volume and available resource list, initiate multi-objective optimization calculations to execute optimization decisions, including: Determine the number and combination of active units: Calculate the number of modular reactor units that meet the theoretical total reaction volume, and select the optimal combination from the available units. The selection considers not only volume matching but also the historical operating time of each unit, prioritizing units with shorter cumulative operating times to balance wear and tear.

[0045] Decision-making operation connection mode: Based on the characteristics of the influent water quality (such as biodegradability and toxicity) and the recommended hydraulic retention time, select the parallel mode (emphasizing treatment capacity and mixing effect) or series mode (emphasizing staged treatment and reaction depth) for the reactor array.

[0046] Understandably, the reactor array in the decision-making operation connection mode specifically refers to a dynamic operation combination in which a group of reactors is dynamically selected and activated from the modular reaction unit array by the scheduling optimization sub-model according to the current operating conditions during real-time operation, and organized in a specific logical connection manner such as parallel or series connection.

[0047] Flow distribution scheme: In parallel mode, the influent flow rate is dynamically distributed according to the real-time status of each activated unit (such as internal pressure loss and catalyst activity), rather than simply distributed evenly, so as to achieve load balance of each unit and overall hydraulic optimization.

[0048] It should be noted that the multi-objective optimization calculation aims to minimize total energy consumption and reagent consumption, and maximize the stability of treatment effect. The constraints are the upper limit of the volume of each reactor and the hydraulic load. The linear weighted sum method or heuristic rules (such as prioritizing the use of units with low cumulative running time) are used for rapid decision-making.

[0049] The collaborative execution unit drives a multi-channel metering pump, an electric regulating valve, a variable frequency circulating pump, and an ultraviolet lamp power supply based on a set of control instructions, thereby regulating the oxidant dosage, water flow distribution, internal circulation disturbance intensity, and ultraviolet catalytic intensity, respectively.

[0050] The collaborative execution unit includes an oxidant dosing subsystem, a fluid dynamic adjustment subsystem, and an ultraviolet catalytic intensity regulation subsystem, wherein: The oxidant dosing subsystem receives control commands generated by the intelligent control unit and drives a multi-channel variable frequency metering pump to add the calculated total oxidant dose in an unbalanced manner according to the flow rate ratio of each reactor and the mixing uniformity compensation coefficient. The fluid dynamic distribution subsystem receives control commands generated by the intelligent control unit, controls the opening degree of the inlet water regulating valve of each reactor to distribute the flow, and controls the speed of the variable frequency circulating pump connected to the reactor circulation disturbance interface to adjust the internal flow state and improve the mixing effect. The ultraviolet catalytic intensity regulation subsystem adjusts the power or the number of ultraviolet lamps turned on or off in each reactor by receiving control commands generated by the intelligent control unit.

[0051] This embodiment also provides an advanced oxidation water treatment disinfection method, such as... Figure 3 As shown, it includes the following steps: S1. Multi-source data synchronous acquisition and feature engineering processing: Real-time data from distributed sensing units is acquired synchronously, including influent flow rate, influent water quality parameters with multiple dimensions such as chemical oxygen demand and total organic carbon, measurements from multiple redox potential probes inside each activated reactor, and key indicators such as the concentration of disinfection byproduct precursors in the effluent. Real-time feature engineering processing is then performed on the influent data. After data cleaning, derived features such as pollutant load are calculated and standardized to form a real-time feature vector for model input.

[0052] S2. Feedforward Prediction to Generate Initial Process Baseline: Real-time feature vectors are input into the feedforward prediction sub-model. High-order abstract information is extracted from the features through the shared feature extraction layer of the feedforward prediction sub-model. Then, based on dual prediction branches, feedforward prediction instructions containing the initial value of oxidant demand and the recommended hydraulic residence time are simultaneously output, together forming the initial process parameter set. This step provides a forward-looking process baseline for the system to cope with changing influent conditions.

[0053] S3. Scheduling and Optimization Decision-Making Resource Allocation: The scheduling and optimization sub-model receives the total influent flow rate and recommended hydraulic retention time, calculates the required theoretical total reaction volume, and evaluates the availability status of all modular reactor units in real time. Based on the principle of multi-objective optimization, it dynamically decides the optimal number and combination of reactor units to be activated, determines the parallel or series operation mode, and formulates a preliminary flow allocation scheme, thereby forming a scheduling instruction set to ensure that the system's processing capacity can flexibly match real-time demand.

[0054] S4. Feedback Correction and Parameter Fine-tuning: The feedback correction sub-model synchronously receives initial process parameters, scheduling instructions, real-time oxidation-reduction potential spatial distribution data, effluent water quality data, and the status of each actuator. It calculates the mixing uniformity index within the reaction system by analyzing oxidation-reduction potential values ​​and integrates multiple effluent indicators to evaluate the real-time disinfection effect. Based on this, the initial process parameters are dynamically corrected: the total oxidant dosage is compensated according to the mixing state, and the distribution to each branch is optimized. Mixing defect patterns are diagnosed to directionally adjust the intensity of circulation disturbances. The hydraulic residence time and UV catalytic intensity are fine-tuned in conjunction with the effluent effect, ultimately generating a final control instruction set containing all precise execution instructions.

[0055] S5. Collaborative Execution: The final control command set and scheduling commands are synchronously issued to the collaborative execution unit. The oxidant precision dosing subsystem drives the multi-channel metering pump to perform dosing; the fluid dynamic allocation subsystem controls the inlet regulating valves and circulation disturbance pumps of each reactor to achieve flow distribution and internal flow optimization; the ultraviolet catalytic intensity adjustment subsystem adjusts the power of each ultraviolet lamp group. All actuators operate collaboratively to ensure that the control intention is accurately implemented.

[0056] S6. Effect Evaluation and Model Self-Learning: After a control cycle ends, the disinfection effect is re-evaluated based on the latest effluent data. If the effect continues to deviate from the target, the key parameters in the feedback correction sub-model are automatically fine-tuned. Simultaneously, the complete "input-output-effect" data valid within this cycle is stored in the case database for periodic retraining of the feedforward prediction sub-model, thereby enabling the entire system's prediction and control capabilities to continuously optimize and improve with the accumulation of operational experience.

[0057] Understandably, the key parameters for automatic fine-tuning specifically refer to the core adjustable coefficients within the feedback correction sub-model that transform real-time monitoring data into precise control commands. These include the compensation function coefficient used to correlate mixing uniformity with oxidant dosage, the weight determining the oxidant distribution ratio in each reactor branch, the gain coefficient controlling the adjustment amplitude of circulation disturbance intensity, the fusion weight of multiple indicators for evaluating disinfection effectiveness, and various thresholds that trigger fine-tuning of process parameters.

[0058] This invention achieves flexible adjustment of treatment capacity over a wide range by using modular reactor arrays and intelligent scheduling optimization, adapting to large flow fluctuations and solving the problem of narrow adjustment range of fixed-volume devices. Secondly, by introducing a combination of distributed oxidation-reduction potential monitoring and active circulation disturbance control, real-time sensing and active optimization of the mixing state inside the reactor are achieved, ensuring uniform mixing of oxidant and pollutants, significantly improving the utilization efficiency of hydroxyl radicals, and eliminating disinfection dead zones. Finally, through a hybrid intelligent control model composed of feedforward prediction, feedback correction, and self-learning mechanisms, dynamic and precise control and closed-loop optimization of key process parameters such as oxidant dosing and hydraulic retention time are achieved. This allows the system to automatically adjust to the optimal operating state based on real-time influent water quality and quantity, ensuring stable and compliant disinfection while significantly reducing reagent and energy consumption.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An advanced oxidation water treatment and disinfection system, characterized in that: It includes a modular reaction unit array, a distributed sensing unit, an intelligent control unit, and a collaborative execution unit; The modular reaction unit array includes at least two parallel, identical pipe-type AOP reactors. Each reactor has a premixing section, a main reaction section, and a flow stabilizing section arranged sequentially along the water flow direction. The main reaction section integrates an ultraviolet catalytic component. Each reactor is independently configured with an inlet regulating valve, an oxidant dosing branch, and a circulation disturbance interface. The intelligent control unit is configured as an industrial controller based on an edge computing architecture. The industrial controller has a hybrid intelligent control model built inside, which is used to generate initial process parameters through feedforward prediction, and to perform dynamic correction and resource scheduling optimization by combining feedback from the mixed state and the water output effect. Finally, it generates and issues control commands to drive the coordinated action of each actuator. The hybrid intelligent control model includes: Feedforward prediction sub-model: Based on the real-time influent flow and water quality parameters of the influent monitoring module, a pre-trained machine learning model is used to obtain feedforward prediction instructions; Feedback correction sub-model: Based on the ORP spatial distribution data obtained by the monitoring modules of each reactor unit, the overall mixing uniformity index is calculated, and the feedforward prediction results are dynamically corrected according to the mixing uniformity index and the feedback data of the effluent monitoring module to generate the final control instruction set. Scheduling optimization sub-model: Based on the real-time total influent flow rate and the operating status of each reactor, a multi-objective optimization algorithm is used to dynamically determine the number of reactors to be activated, the flow allocation ratio of each reactor, and the operating mode.

2. The advanced oxidation water treatment disinfection system according to claim 1, characterized in that: The distributed sensing unit is used to continuously collect influent water quality and quantity data, dynamically monitor the spatial distribution of oxidant inside each reactor, and simultaneously track the final effluent disinfection effect indicators, transmitting multi-source real-time sensing data to the intelligent control unit.

3. The advanced oxidation water treatment and disinfection system according to claim 2, characterized in that: The distributed sensing unit includes an influent monitoring module, a unit monitoring module, and an effluent monitoring module, wherein: The influent monitoring module, located in the main influent pipeline, is used to monitor the influent flow rate and water quality parameters in real time. The water quality parameters include at least chemical oxygen demand, total organic carbon, turbidity, and pH value. The unit monitoring module, corresponding to each reactor, includes at least two redox potential probes located in the main reaction section to monitor the uniformity index of oxidant distribution inside the reactor. The effluent monitoring module, located in the main effluent pipeline, is used to monitor effluent indicators in real time, including the concentration of disinfection byproduct precursors and the concentration of microbial indicators.

4. The advanced oxidation water treatment disinfection system according to claim 1, characterized in that: The process of obtaining the feedforward prediction instruction includes the following steps: Real-time acquisition and construction of multi-dimensional influent features: After cleaning the real-time influent flow rate and water quality parameters, derived features are obtained through calculation. Then, all features are standardized to form a unified real-time feature vector. Synchronous prediction based on a multi-task learning architecture: Real-time feature vectors are input into a pre-trained machine learning model. The machine learning model adopts a multi-task learning structure with a shared feature extraction layer and dual prediction branches. The shared feature extraction layer is responsible for extracting high-order, abstract pattern information from the features. The dual prediction branches simultaneously infer the initial value of oxidant demand and the recommended hydraulic residence time based on the shared information.

5. The advanced oxidation water treatment disinfection system according to claim 1, characterized in that: The feedback correction sub-model generates the final control instruction set by the following steps: Multi-source data synchronization and fusion: synchronously receive feedforward prediction commands, real-time process monitoring data and system status data. Among them, the feedforward prediction commands include the initial value of oxidant demand and the recommended hydraulic residence time, the real-time process monitoring data includes the redox potential values ​​and effluent water quality at multiple points inside each reactor, and the system status data includes the current operating conditions of each actuator. Deviation analysis: The spatial mixing uniformity index is calculated based on the oxidation-reduction potential value. If the spatial mixing uniformity index is lower than the set threshold, it indicates that the oxidant is unevenly distributed in the reaction system. At the same time, the effluent water quality data is compared with the target value to evaluate the actual disinfection effect of the current cycle. Parameter optimization and correction: Based on real-time mixing status and disinfection effect feedback, the total amount of oxidant added, mixing intensity and process parameters are adjusted in a coordinated manner; Among them, the correction of oxidant dosage is as follows: the initial oxidant requirement is compensated according to the mixing uniformity. If the mixing is not up to standard, the total dosage is increased by a certain proportion. At the same time, the distribution ratio of each reactor's addition branch is adjusted, and units with poor mixing are given preferential treatment. Adjustment of mixing intensity: Based on the mixing uniformity and ORP data of each unit, generate adjustment commands for the circulation pump frequency or stirring intensity; Coordinated fine-tuning of process parameters: Based on the actual disinfection effect of the effluent and the state of the reactor, the hydraulic retention time and ultraviolet intensity parameters are coordinated and fine-tuned. Generate the final control instruction set: Package all the corrected parameters into a unified final control instruction set, which includes execution instructions for the oxidant dosing subsystem, fluid preparation subsystem, and ultraviolet catalysis subsystem, and send them to the collaborative execution unit in real time.

6. The advanced oxidation water treatment disinfection system according to claim 1, characterized in that: The scheduling optimization sub-model dynamically determines the number of reactors to be activated, the flow allocation ratio of each reactor, and the operating mode through a multi-objective optimization algorithm, including the following steps: Resource status assessment: The theoretical total reaction volume is calculated based on the product of the recommended hydraulic retention time and the total influent flow rate. At the same time, the availability status and rated volume of all modular reactor units are polled in real time to generate a list of available resources. Scheduling scheme generation: Based on the theoretical total reaction volume and available resource list, multi-objective optimization calculations are initiated to execute optimization decisions, which include: Determine the number and combination of active units: Calculate the number of modular reactor units that satisfy the theoretical total reaction volume, and select the optimal combination from the available units; Decision-making operation connection mode: Select the parallel or series mode of the reactor array based on the characteristics of the influent water quality and the recommended hydraulic retention time; Flow distribution scheme: In parallel mode, the influent flow is dynamically distributed according to the real-time status of each activated unit.

7. The advanced oxidation water treatment disinfection system according to claim 1, characterized in that: The collaborative execution unit drives a multi-channel metering pump, an electric regulating valve, a variable frequency circulating pump, and an ultraviolet lamp power supply based on a set of control instructions, thereby regulating the oxidant dosage, water flow distribution, internal circulation disturbance intensity, and ultraviolet catalytic intensity, respectively.

8. The advanced oxidation water treatment disinfection system according to claim 7, characterized in that: The collaborative execution unit includes an oxidant dosing subsystem, a fluid dynamic adjustment subsystem, and an ultraviolet catalytic intensity regulation subsystem, wherein: The oxidant dosing subsystem receives control commands generated by the intelligent control unit and drives a multi-channel variable frequency metering pump to add the calculated total oxidant dose in an unbalanced manner according to the flow rate ratio of each reactor and the mixing uniformity compensation coefficient. The fluid dynamic distribution subsystem receives control commands generated by the intelligent control unit, controls the opening degree of the inlet water regulating valve of each reactor to distribute the flow, and controls the speed of the variable frequency circulating pump connected to the reactor circulation disturbance interface to regulate the internal flow state. The ultraviolet catalytic intensity regulation subsystem adjusts the power or the number of ultraviolet lamps turned on or off in each reactor by receiving control commands generated by the intelligent control unit.

9. An advanced oxidation water treatment disinfection method, applied to an advanced oxidation water treatment disinfection system as described in any one of claims 1-8, characterized in that: Includes the following steps: S1. Synchronous acquisition and feature engineering of multi-source data: Real-time data from distributed sensing units are acquired synchronously, including influent water quality parameters, measured values ​​of multiple redox potential probes inside each activated reactor, and concentration of disinfection byproduct precursors in the effluent; then, real-time feature engineering is performed on the influent data. After data cleaning, derived features are calculated and standardized to generate real-time feature vectors for model input. S2. Feedforward prediction to generate initial process baseline: Input the real-time feature vector into the feedforward prediction sub-model, extract high-order abstract information from the features through the shared feature extraction layer of the feedforward prediction sub-model, and then output the feedforward prediction instruction containing the initial value of oxidant demand and the recommended hydraulic residence time based on the dual prediction branches, which together constitute the initial process parameter set. S3. Scheduling optimization decision-making resource allocation: The scheduling optimization sub-model receives the total influent flow rate and recommended hydraulic retention time, calculates the required theoretical total reaction volume, and evaluates the availability status of all modular reactor units in real time. Based on the principle of multi-objective optimization, it dynamically decides the optimal number and combination of reactor units to be activated, determines the parallel or series operation mode, and formulates a preliminary flow allocation scheme, thereby forming a scheduling instruction set. S4. Feedback Correction and Parameter Fine-tuning: The feedback correction sub-model synchronously receives initial process parameters, scheduling instruction set, real-time oxidation-reduction potential spatial distribution data, effluent water quality data, and the status of each actuator. It calculates the mixing uniformity index within the reaction system by analyzing the oxidation-reduction potential values ​​and integrates multiple effluent indicators to evaluate the real-time disinfection effect. Simultaneously, it compensates for the total oxidant dosage based on the mixing state and optimizes the distribution of each branch, diagnoses mixing defect modes to directionally adjust the intensity of circulation disturbance, and fine-tunes the hydraulic residence time and ultraviolet catalytic intensity in conjunction with the effluent effect, ultimately generating a final control instruction set containing all execution instructions. S5. Collaborative Execution: The final control instruction set and scheduling instructions are synchronously sent to the collaborative execution unit. The oxidant precision dosing subsystem drives the multi-channel metering pump to perform dosing; the fluid dynamic allocation subsystem controls the inlet water regulating valve and circulation disturbance pump of each reactor; and the ultraviolet catalytic intensity regulation subsystem adjusts the power of each ultraviolet lamp group.

10. The advanced oxidation water treatment disinfection method according to claim 9, characterized in that: It also includes the following steps: Effect evaluation and model self-learning: After a control cycle ends, the disinfection effect is re-evaluated based on the latest effluent data. If the effect continues to deviate from the target, the key parameters in the feedback correction sub-model are automatically fine-tuned. At the same time, the effective complete data in this cycle will be stored in the case database for periodic retraining of the feedforward prediction sub-model.

Images