Industrial wastewater deep recycling system based on multi-scale membrane separation technology and control method

Abstract

The invention discloses an industrial wastewater deep recycling system based on a multi-scale membrane separation technology and a control method, and relates to the field of wastewater treatment. The system comprises a control unit, and a pretreatment unit, a low-molecular pollutant separation unit, a nanofiltration membrane unit, a reverse osmosis membrane unit and a post-treatment unit which are respectively arranged from a wastewater inlet end to a wastewater outlet end. A first buffer storage tank, a second buffer storage tank, a third buffer storage tank and a fourth buffer storage tank are respectively arranged between adjacent units of the pretreatment unit, the low-molecular pollutant separation unit, the nanofiltration membrane unit, the reverse osmosis membrane unit and the post-treatment unit. The system aims to solve the problems of low efficiency, high energy consumption, unstable recycled water quality and the like of a traditional industrial wastewater recycling system through combination of a membrane separation mode and a multi-stage treatment technology, and achieves an efficient, low-consumption and stable industrial wastewater deep recycling effect.

Description

Technical Field

[0001] This application relates to the field of wastewater treatment, in particular to an industrial wastewater advanced reuse system and control method based on multi-scale membrane separation technology. Background Art

[0002] With the rapid development of global industrialization, the discharge of industrial wastewater and the complexity of pollutants in the wastewater are increasing continuously, which brings severe challenges to water resource management and environmental protection. Industrial wastewater not only contains a large amount of organic matter, inorganic matter and suspended solids, but may also contain various toxic and harmful components that are difficult to remove, such as heavy metals, drug residues, chemical additives, etc. With the increasing strictness of environmental regulations and the intensification of water resource shortage problems, the reuse and resource utilization of industrial wastewater become particularly important. However, the current wastewater treatment technologies still have many deficiencies in dealing with complex wastewater and are difficult to meet the growing demand for industrial wastewater reuse.

[0003] Traditional industrial wastewater treatment methods mainly include physicochemical methods, biological methods and membrane separation technology. Physicochemical methods usually react chemical agents with pollutants or use physical methods (such as filtration, precipitation, etc.) to remove pollutants in the wastewater. Although these methods can remove suspended solids, dissolved organic matter, etc. in the wastewater to a certain extent, the treatment effect is often unstable. Especially when the composition of the wastewater changes greatly, the effect of these methods will decrease significantly. Physicochemical methods are also often accompanied by high energy consumption and high chemical agent usage, and some components in the wastewater are difficult to be completely removed, resulting in the water quality being difficult to meet the reuse standard.

[0004] As a relatively traditional and widely used wastewater treatment method, biological methods degrade organic pollutants in the wastewater through microorganisms and usually have a relatively low operating cost. However, for wastewater with high concentration, toxic substances or drastic changes, the treatment ability of biological methods is greatly limited. The treatment process of biological methods is generally slow, and has strict requirements on environmental conditions such as water temperature, pH value, dissolved oxygen, etc., and a large amount of sludge is still generated during the treatment process, increasing the cost and difficulty of subsequent disposal.

[0005] As an emerging water treatment technology, membrane separation technology has been widely used in the treatment of industrial wastewater due to its advantages such as high separation efficiency, energy conservation, and no secondary pollution. Membrane separation technology can selectively separate dissolved substances and suspended solids in water according to the pore size of the membrane. Common membranes include microfiltration membranes, ultrafiltration membranes, nanofiltration membranes, and reverse osmosis membranes. However, despite the promising prospects of membrane separation technology in the field of water treatment, single membrane separation technology still has certain limitations. First of all, the problem of membrane fouling severely limits the long-term stable operation of the membrane system. As the use time of the membrane system increases, substances such as organic matter, bacteria, and inorganic matter in the wastewater will accumulate on the membrane surface, resulting in a decrease in membrane flux and even complete blockage of the membrane pores, affecting the separation efficiency of the membrane. The occurrence of membrane fouling not only affects the wastewater treatment effect but also requires additional cleaning and maintenance, increasing the operation and maintenance costs of the system.

[0006] In addition, the energy consumption of membrane separation technology is relatively high. Especially, reverse osmosis membranes and nanofiltration membranes need to operate under high pressure. When treating high-concentration wastewater, the energy consumption of the membrane system will increase significantly, which poses challenges to the economy and sustainability of membrane separation technology. Although single membrane technology can effectively remove certain pollutants, the removal effect of multiple complex pollutants in wastewater is not ideal, and it is necessary to improve the water quality through multi-stage membrane separation or combination with other treatment methods.

[0007] With the diversification and complexity of the treatment requirements of industrial wastewater, traditional single membrane technology and other water treatment methods cannot meet the requirements of high efficiency, low energy consumption, and stability for wastewater reuse. Therefore, the introduction of multi-scale membrane separation technology, that is, by using different types of membrane technologies (such as microfiltration, ultrafiltration, nanofiltration, reverse osmosis membranes, etc.) at different treatment stages, can make full use of the advantages of each membrane in wastewater treatment, achieve step-by-step separation and multiple purification of wastewater, and avoid the disadvantages of single membrane technology. Multi-scale membrane separation technology can effectively reduce the incidence of membrane fouling and, by reasonably configuring the membrane pore size and operating pressure, enable different pollutants to be most effectively removed in different membrane units.

[0008] However, despite the significant advantages of multi-scale membrane separation technology in wastewater treatment, how to precisely control the operating state of each stage of the membrane unit to ensure high separation efficiency and extend the service life of the membrane remains a difficult point in technology implementation. The composition of wastewater is complex and the water quality fluctuates greatly, and traditional wastewater treatment methods are often difficult to adapt to these changes in real time. Summary of the Invention

[0009] The purpose of this application aims to at least overcome one deficiency existing in the technology, and provides an industrial wastewater advanced reuse system and a control method based on multi-scale membrane separation technology. This system aims to combine the membrane separation method with multi-stage treatment technology to solve the problems existing in the traditional industrial wastewater reuse system, such as low efficiency, high energy consumption, and unstable quality of the recycled water, so as to achieve an efficient, low-consumption, and stable effect of industrial wastewater advanced reuse.

[0010] In the first aspect, this application discloses an industrial wastewater advanced reuse system based on multi-scale membrane separation technology. This system includes a control unit, and a pretreatment unit, a low-molecular pollutant separation unit, a nanofiltration membrane unit, a reverse osmosis membrane unit, and a post-treatment unit are respectively arranged from the wastewater inlet end to the wastewater outlet end.

[0011] A first buffer storage tank, a second buffer storage tank, a third buffer storage tank, and a fourth buffer storage tank are respectively arranged between adjacent units of the pretreatment unit, the low-molecular pollutant separation unit, the nanofiltration membrane unit, the reverse osmosis membrane unit, and the post-treatment unit. Liquid buffer transmission between adjacent units is achieved through the first buffer storage tank, the second buffer storage tank, the third buffer storage tank, and the fourth buffer storage tank.

[0012] In the above structure, the first buffer storage tank, the second buffer storage tank, the third buffer storage tank, and the fourth buffer storage tank are used to temporarily store part of the wastewater when the wastewater treatment load is high or the water quality fluctuates greatly, so as to avoid system overload or untimely operation of the membrane unit.

[0013] The pretreatment unit includes a pretreatment tank. An inlet is arranged at the left end of the pretreatment tank, and an outlet is arranged at the opposite end. Between the inlet and the outlet, a coarse grid, a sedimentation area, and a chemical reaction area are horizontally arranged in sequence.

[0014] After being treated by the pretreatment unit, the wastewater is transported to the first buffer tank through a pipeline, and then sent from the first buffer tank to the low-molecular pollutant separation unit.

[0015] The low-molecular pollutant separation unit, the nanofiltration membrane unit, and the reverse osmosis membrane unit have the same structure, and each includes a total inlet pipe and a total outlet pipe with a pump, and a number of filtration modules connected in parallel between the total inlet pipe and the total outlet pipe.

[0016] Furthermore, a filtration membrane for realizing filtration is arranged in the filtration module. The inlet of each filtration module is connected to the total inlet pipe through a branch inlet pipe with a valve, and the outlet is connected to the total outlet pipe through a sub-outlet pipe with a valve.

[0017] Furthermore, the valves of the branch inlet pipe and the sub-outlet pipe are adjusted by the control unit to control the waste water flow rate and the working state of the filtration module. When a certain filtration module is blocked or needs to be cleaned, the control unit closes the corresponding sub-branch inlet pipe and outlet pipe valves to stop the module from working and prevent waste water from passing through the module.

[0018] The pore sizes of the filter membranes provided in the filtration module vary according to the unit functions. Among them, the pore size of the low molecular pollutant separation unit is 0.1 to 10 microns, the pore size of the nanofiltration membrane unit is 1 to 10 nanometers, and the pore size of the reverse osmosis membrane unit is 0.1 to 0.5 nanometers.

[0019] Hydraulic sensors, turbidity sensors, and conductivity sensors are configured at the inlet of each filtration module. Membrane surface pressure sensors and temperature sensors are also installed in each filtration module and are matched with the membrane. Among them, the hydraulic sensor monitors the waste water inlet pressure in real time to ensure that the membrane unit works under appropriate pressure; the turbidity sensor detects the concentration of suspended solids in the waste water to judge the membrane pollution condition; the conductivity sensor evaluates the degree of water pollution; the membrane surface pressure sensor monitors the degree of membrane pollution; the temperature sensor ensures that the waste water temperature is within the working range of the membrane unit; a flow regulating valve and a pressure sensor are configured on the main inlet pipe to monitor and adjust the waste water flow rate and pressure in real time.

[0020] The data collected by all sensors are transmitted to the control unit through communication lines or modules for analysis and processing.

[0021] To realize the membrane cleaning function, a backwash water storage tank and a backwash pipeline are also included in the system; the backwash water storage tank is used to store clean water for cleaning. The backwash water storage tank is connected to the filtration modules of each unit through the backwash main pipe. A backwash pump and a backwash main valve are configured on the backwash main pipe. A backwash branch pipe is provided at the outlet of each filtration module. The backwash branch pipe is connected in parallel with the sub-outlet pipe through a connector with a valve. The valve in the backwash branch pipe is controlled by the control unit. When performing backwashing, the control unit starts the backwash pump, opens the backwash main valve and the backwash branch pipe valve, and the clean water flows reversely from the outlet of the filtration module, flushes the membrane surface and then is discharged through the branch inlet pipe.

[0022] To further realize the membrane cleaning function, a chemical cleaning agent storage tank is also included in the system. The chemical cleaning agent storage tank is used to store the pre-prepared cleaning agent; the chemical cleaning agent storage tank is connected to the filtration modules of each unit through the cleaning agent main pipe. A metering pump and a cleaning agent main valve are configured on the cleaning agent main pipe.

[0023] At the liquid inlet of each filtration module, there is a cleaning agent branch pipe. The cleaning agent branch pipe is connected in parallel with the branch inlet pipe through a connector with a valve, and the valve is controlled by the control unit. When performing chemical cleaning, the control unit starts the metering pump, opens the main cleaning agent valve and the corresponding cleaning agent branch pipe valve. The cleaning agent is injected into the filtration module and discharged through the branch inlet pipe after cleaning.

[0024] To handle the cleaning waste liquid, a cleaning waste liquid collection branch pipe is set on the branch inlet pipe of each filtration module. The collection branch pipe is connected to the branch inlet pipe through a connector with a valve and converges into the main cleaning waste liquid pipe.

[0025] The main cleaning waste liquid pipe is connected to a waste liquid collection tank or discharge port outside the system, and a waste liquid discharge valve is configured on the main waste liquid pipe. When performing backwashing or chemical cleaning, the control unit closes the connection valve between the branch inlet pipe and the main inlet pipe, opens the cleaning waste liquid collection branch pipe valve. The waste liquid generated by cleaning enters the main waste liquid pipe through the collection branch pipe and is finally discharged or collected to avoid polluting the normal wastewater treatment path.

[0026] To ensure the independence of cleaning and normal operation, there are three-way valves on the branch inlet pipe and the sub-outlet pipe of each filtration module. During cleaning, the connection relationship between the filtration module and the chemical cleaning agent storage tank and the backwashing water storage tank is controlled through the three-way valve. By adjusting the three-way valve through the control unit, the filtration module is isolated from the normal wastewater treatment path and switched to the cleaning path; after cleaning is completed, the three-way valve is reset to resume normal operation. This structure ensures that the cleaning module operates independently without interfering with other modules.

[0027] The post-treatment unit includes a working tank. Above the working tank, there is a spraying mechanism for evenly distributing the treated wastewater. An ultraviolet disinfection component and an ozone generator are configured in the working tank. The ultraviolet disinfection component removes microorganisms and bacteria in the water, and the ozone generator further removes organic pollutants and residual pollutants. The water outlet of the post-treatment unit is connected to the system water outlet through a pipeline to ensure that the final recycled water quality meets the standards. A water quality sensor is configured at the water outlet to monitor indicators such as the bacterial content and chemical oxygen demand (COD) of the recycled water, and the data is transmitted to the control unit to ensure stable water quality.

[0028] All units of the system are connected through pipelines. The wastewater flows sequentially between different units, gradually removing pollutants to ensure that the final water quality meets the requirements of industrial reuse. Through the cooperation of valves, pumps, sensors and the control unit, the system flexibly adjusts the working state of the filtration module, dynamically optimizes the operation parameters according to the water quality changes, combines the backwashing and chemical cleaning functions to maximize the wastewater treatment efficiency, extend the service life of the membrane unit, reduce energy consumption and stabilize the quality of the recycled water.

[0029] Second aspect, based on the system disclosed in the first aspect, the present application discloses a control method for deep reuse of industrial wastewater based on multi-scale membrane separation technology. This method realizes the efficient treatment and reuse of industrial wastewater, while reducing energy consumption and extending the service life of the membrane through real-time data collection, AI deep learning and predictive analysis, and dynamic control. The steps of the method are as follows:

[0030] Step 1: Parameter collection and data packet generation. Pressure, suspended solid concentration, pollution degree, membrane surface pressure, and temperature data of the wastewater are collected in real time through hydraulic sensors, turbidity sensors, conductivity sensors, membrane surface pressure sensors, and temperature sensors configured at the liquid inlets of each filtration module. At the same time, wastewater flow and total pressure data are collected through the flow regulating valve and pressure sensor at the total liquid inlet pipe, and the bacterial content and chemical oxygen demand value of the reclaimed water are collected through the water quality sensor at the water outlet. The control unit receives the data collected by the above sensors, generates a data packet containing unit identifiers, timestamps, and corresponding parameter values, and transmits it to the control unit through the communication system for subsequent analysis and processing. The generated data packet is also stored in the historical database of the control unit as training data for the AI deep learning model.

[0031] Step 2: Data processing, membrane fouling assessment and prediction. The control unit preprocesses the received data packet, including outlier filtering, denoising, and standardization operations to adjust the data to a unified scale. Based on the preprocessed data, the control unit calculates the current membrane fouling degree of each filtration module using the membrane surface pressure sensor data and flow data. The calculation formula is:

[0032] where ΔR m (t) represents the current membrane fouling degree, ΔR m (t) is the initial membrane surface pressure, P(t) is the current membrane surface pressure, J(t) is the current membrane flux, and the membrane flux is calculated from the total liquid inlet pipe flow and the effective membrane area of the filtration module. The control unit further predicts the membrane fouling trend through the built-in AI deep learning model. This model is based on historical data and adopts a convolutional neural network combined with a long short-term memory network architecture. After training, it can predict the membrane fouling degree ΔR m (t + Δt) within a certain period in the future. The prediction input includes the current data packet and the historical data sequence of the past 72 hours, and the output is the time series curve of the membrane fouling degree. The control unit combines the current evaluation results and prediction values to analyze the change trends of the wastewater suspended solid concentration, pollution degree, and the treatment load of each unit, providing a basis for subsequent control decisions.

[0033] Step 3: The filtration module is dynamically regulated and AI-optimized for control decisions. The control unit dynamically regulates the operating states of the filtration modules of each unit according to the current membrane fouling assessment results in Step 2 and the membrane fouling trend predicted by AI; the control unit is based on the predicted membrane fouling degree ΔR m (t + Δt) and real-time turbidity data to determine the number of active filtration modules in each unit: When the prediction shows that ΔR m (t + Δt) is less than 0.3 and the turbidity is below 50 mg / L within the next 24 hours, the control unit closes the sub-outlet pipe valves of some filtration modules to reduce the number of active modules to the minimum operating requirement; when the prediction shows that ΔR m (t + tΔt) is greater than 0.6 or the turbidity exceeds 100 mg / L, the control unit turns on all filtration modules in advance to enhance the treatment capacity. The adjustment process refers to the predicted pollution peak time of the AI model to ensure that the change in the number of modules matches the dynamic wastewater load;

[0034] The control unit dynamically adjusts the operating pressure and influent flow rate according to the real-time hydraulic data and the predicted membrane flux change trend by AI: When the prediction shows that ΔR m (t + tΔt) is less than 0.2 and the membrane flux remains above 80% of the design value, the control unit reduces the total influent pipe pressure and decreases the opening of the flow valve to optimize energy consumption; when the prediction shows that ΔR m (t + tΔt) is between 0.4 and 0.6 and the membrane flux will drop below 60% of the design value, the control unit increases the operating pressure and flow rate in advance to avoid a decrease in treatment efficiency. The adjustment parameters are determined by the optimization curve trained by the AI model based on historical data and corrected according to the temperature data to ensure operation within the membrane working range;

[0035] The control unit adjusts the storage capacities of the first buffer storage tank to the fourth buffer storage tank by using the predicted wastewater flow rate and pollution load trends by AI: When the predicted flow rate will exceed 110% of the design capacity of each unit within the next 6 hours, the control unit increases the storage volume of the corresponding buffer storage tank and preferentially allocates it to the upstream storage tank to relieve the pressure on the downstream membrane unit; when the predicted flow rate drops below the design value, the wastewater in the storage tank is gradually released, and the distribution decision combines the predicted conductivity value. If the pollution degree is expected to increase, the control unit adjusts the storage strategy in advance;

[0036] The control unit switches the filtration module status in advance according to the predicted ΔR m (t + Δt) value by AI: When it is predicted that a certain module will reach the cleaning threshold within the next 12 hours, the control unit closes its sub-outlet pipe valve, switches to the standby state, and enables the redundant module to take over the operation. The switching time window is optimized by the AI model according to the pollution accumulation rate to ensure stable flow;

[0037] Through the above-mentioned regulation, AI deep learning and prediction technologies optimize control decisions, enabling the system to proactively respond to changes in wastewater load and membrane fouling, and improving operation efficiency.

[0038] Step 4: Execution of the membrane cleaning procedure. When the current membrane fouling degree ΔR m (t) or the predicted value ΔR m (t + Δt) reaches the preset threshold, the control unit starts the cleaning procedure for the corresponding filtration module. The cleaning method is determined according to the fouling degree, membrane material, and conductivity data: when the fouling degree is between 0.8 and 1.0, the control unit closes the connection valve between the branch inlet pipe and the main inlet pipe, adjusts the three-way valve to the backwashing path, starts the backwashing pump to transport clean water, and each time it lasts for 5 minutes; when the fouling degree exceeds 1.0 or the conductivity exceeds 8000 μS / cm, the control unit adjusts the three-way valve to the chemical cleaning path, starts the metering pump to inject the cleaning agent, and the cleaning time is 20 minutes. The cleaning waste liquid is discharged through the cleaning waste liquid collection branch pipe; the cleaning cycle is adjusted by the pollution accumulation rate predicted by the AI model.

[0039] Step 5: Post-treatment and water quality assurance. After the wastewater is treated by multi-scale membrane separation, it enters the post-treatment stage. The control unit drives the spray mechanism to distribute the wastewater, activates the ultraviolet disinfection component and the ozone generator to remove microorganisms and organic pollutants. The water quality sensor at the outlet monitors the bacterial content and COD value. If it exceeds the standard, the control unit adjusts the ozone dosage or the pressure of the previous membrane unit according to the AI prediction until the water quality meets the standard.

[0040] Step 6: Real-time feedback and AI adaptive optimization. After each treatment cycle, the control unit evaluates the system performance according to the sensor feedback data. If the membrane fouling or water quality fluctuation exceeds the preset range, the operating parameters are adjusted; the control unit updates the AI model using real-time data and optimizes the prediction accuracy and control strategy through online learning.

[0041] Through the above steps, this method uses AI deep learning and prediction technologies to achieve accurate prediction of membrane fouling trends and intelligent optimization of control decisions, improving wastewater treatment efficiency, membrane service life, and the stability of the quality of recycled water.

[0042] The beneficial effects listed above do not exhaust all the advantages. Other potential beneficial effects and detailed technical implementation manners will be further revealed in the embodiments or other description parts of this application. Description of the Drawings

[0043] After reading the following specific implementation manners in conjunction with the drawings, various aspects of the present disclosure will be better understood. Sometimes, the positions, sizes, and ranges of the various structures shown in the drawings, etc., do not represent the actual positions, sizes, and ranges, etc. In the drawings:

[0044] Figure 1It is a hardware connection block diagram of an embodiment disclosed in the present application.

[0045] Figure 2 It is a system working flowchart in an embodiment to be disclosed in the application. Detailed implementation manners

[0046] The present disclosure will be described below with reference to the accompanying drawings, in which several embodiments of the present disclosure are shown. However, it should be understood that the present disclosure can be presented in many different ways and is not limited to the embodiments described below; in fact, the embodiments described below are intended to make the disclosure of the present disclosure more complete and fully explain the protection scope of the present disclosure to those skilled in the art. It should also be understood that the embodiments disclosed herein can be combined in various ways to provide more additional embodiments.

[0047] It should be understood that in all the drawings, the same reference numerals represent the same elements. In the drawings, for clarity, the dimensions of some features may be deformed.

[0048] It should be understood that the terms used in the specification are only for describing specific embodiments and are not intended to limit the present disclosure. All terms used in the specification (including technical terms and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. For the sake of simplicity and / or clarity, technologies, methods, and devices known to those of ordinary skill in the relevant fields may not be discussed in detail, but where appropriate, the said technologies, methods, and devices should be regarded as part of the authorized specification.

[0049] The singular forms "a", "the", and "said" used in the specification include the plural forms unless clearly specified. The terms "including", "comprising", and "containing" used in the specification indicate the presence of the claimed features, but do not exclude the presence of one or more other features. The term "and / or" used in the specification includes any and all combinations of one or more of the related listed items.

[0050] Embodiment:

[0051] Referring to the attached Figure 1 and 2 , Embodiment

[0052] An industrial wastewater advanced reuse system based on multi-scale membrane separation technology, which is integrally composed of a pretreatment unit, a low-molecular pollutant separation unit, a nanofiltration membrane unit, a reverse osmosis membrane unit, and a post-treatment unit connected in series in sequence. Liquid buffer transmission is achieved between each unit through the first to fourth buffer storage tanks.

[0053] Each unit is controlled by a control unit, and the control unit includes a PLC part and an edge server.

[0054] The pretreatment tank of the pretreatment unit is constructed of 304 stainless steel, is 6 meters long and 1.5 meters in diameter. The left-end inlet is connected to the wastewater input source via a DN200 flanged pipe, while the right-end outlet is connected to the first buffer storage tank via a DN200 flange. Within the pretreatment tank, a coarse screen, an inclined plate settling zone, and a chemical reaction zone are arranged horizontally. The coarse screen is constructed of 316L stainless steel with a pore size of 5mm, which intercepts large suspended solids. The inclined plate settling zone consists of PP inclined plates (60° inclination) with a hydraulic retention time of 30 minutes, which can remove over 80% of suspended solids. The chemical reaction zone has a volume of 2 cubic meters and a built-in adjustable speed agitator (power of 2.2kW, speed range of 0-200rpm). A pH probe (range of 0-14, accuracy of ±0.1) monitors pH in real time. A dosing pump (flow rate of 0-10L / min) is used to dose polyaluminum chloride (50mg / L) and anionic polyacrylamide (2mg / L).

[0055] The pretreated wastewater is transported to the first buffer storage tank through a pipeline. The tank is made of vertical PE material with a capacity of 5 cubic meters. An ultrasonic level meter (measurement accuracy ±1mm) is installed on the top of the tank, and a DN150 outlet pipe is installed at the bottom of the tank, which is connected to the low molecular pollutant separation unit through a pneumatic butterfly valve.

[0056] The low molecular weight pollutant separation unit, nanofiltration membrane unit and reverse osmosis membrane unit adopt modular design. Each unit includes a total liquid inlet pipe (DN150), a total liquid outlet pipe (DN150) and 6 parallel filter modules. A centrifugal pump (power 7.5kW, flow rate 10-50m 3 / h), the pump outlet is equipped with an electromagnetic flowmeter and a pressure transmitter.

[0057] The filtration module is composed of a PVDF membrane assembly (size Φ200×1500mm). The module liquid inlet is connected to the main liquid inlet pipe through a DN50 pneumatic ball valve (for example: response time <0.5 seconds), and the liquid outlet is connected to the main liquid outlet pipe through a DN50 electric regulating valve.

[0058] The pore size of each unit's filter membrane is graded by function: 5 microns for the low-molecular-weight contaminant separation unit, 5 nanometers for the nanofiltration unit, and 0.3 nanometers for the reverse osmosis unit. Each module's inlet is equipped with a hydraulic pressure sensor, a turbidity sensor, and a conductivity sensor. Piezoresistive pressure and temperature sensors are embedded in the membrane surface. Sensor signals are transmitted via an RS485 bus to the control unit, where the data is received and processed by a PLC. The PLC's analog input module samples at a 10Hz frequency, and the data is pre-processed and uploaded to an edge server (e.g., Dell PowerEdge R750, Intel Xeon Silver 4210 processor).

[0059] The backwash water storage tank (with a volume of 3 cubic meters and made of UPVC) is connected to each unit through a DN80 backwash main pipe. A centrifugal backwash pump (for example: flow rate 30m 3 / h, head 60m) and an electric butterfly valve are installed on the main pipe. At the liquid outlet of each filtration module, there is a backwash branch pipe (DN40, made of UPVC), which is connected in parallel with the sub-liquid outlet pipe through a quick connector, and a pneumatic diaphragm valve is installed at the end of the branch pipe. The chemical cleaning agent storage tank (made of PE, with a volume of 2 cubic meters) stores 0.5% citric acid solution (pH = 2) and 1% sodium hydroxide solution, and is transported to the module liquid inlet through a metering pump (for example: flow rate 0 - 5L / min) and a PE cleaning agent main pipe. A pneumatic ball valve is set on the cleaning agent branch pipe (DN25). The cleaning waste liquid collection branch pipe (DN40, made of PP) is connected to the branch liquid inlet pipe through a pneumatic three-way valve (ASCO 353), merges into the DN100 waste liquid main pipe (made of PP), and the end is connected to the factory area waste water collection pool. An electric ball valve and a pressure sensor are configured on the main pipe.

[0060] The working tank of the post-treatment unit has a volume of 8 cubic meters and is made of 304 stainless steel. The top spray mechanism consists of a 316L stainless steel nozzle array (aperture 1mm, 50 evenly distributed). The spray pressure is adjusted by a variable frequency water pump. The ultraviolet disinfection component is installed on the left side of the tank body, and the quartz sleeve forms a 45° angle with the waste water flow channel to enhance the irradiation uniformity; the ozone generator is placed on the right side of the tank body, and the ozone output pipe is injected into the bottom of the tank through a titanium alloy aeration head (aperture 0.5mm). An on-line water quality analyzer is installed on the outlet pipe (DN200) to detect the total number of bacteria (fluorescence method, detection limit 1CFU / mL) and COD (ultraviolet absorption method, range 0 - 100mg / L), and the data is transmitted to the control unit through the Modbus TCP protocol.

[0061] When the system is running, the waste water enters the low-molecular pollutant separation unit after removing large particle impurities and suspended matters through the pretreatment unit. The control unit dynamically adjusts the start and stop of the module according to the data of the turbidity sensor. For example, when the detected turbidity value is 150NTU, the PLC starts all 6 filtration modules, the opening of the pneumatic ball valve on the branch liquid inlet pipe is adjusted to 80%, and the operating pressure is maintained at 0.4MPa. The data of the membrane surface pressure sensor (initial value 0.3MPa) is uploaded to the edge server in real time. When the AI model (a hybrid LSTM-CNN model trained by the TensorFlow framework) analyzes the data of the past 72 hours and predicts that the pollution degree will reach 0.85 in the next 12 hours, a backwash instruction is triggered: close the electric valve of the sub-liquid outlet pipe of this module, switch the three-way valve to the backwash path, and the backwash pump flushes reversely at a pressure of 0.6MPa for 5 minutes, and the waste liquid is discharged through the collection branch pipe. If the conductivity exceeds 8000μS / cm, switch to the chemical cleaning mode, and the metering pump injects the cleaning agent at a flow rate of 3L / min for 20 minutes.

[0062] More specifically, the decision-making and operation process of the system in this embodiment is as follows: The PLC collects the membrane surface pressure, turbidity, conductivity, and temperature data of each module in real time, removes noise through moving average filtering (window of 50 points), and normalizes the parameters to the [-1, 1] interval through Z-score normalization. The data packet is appended with the module ID (such as "RO-Module-02" representing the second module of the reverse osmosis unit) and a timestamp (in milliseconds) and stored in the MySQL database in JSON format. For example, when the membrane surface pressure of the fourth module of the nanofiltration unit rises from 0.5 MPa to 1.1 MPa and the turbidity detection value is 200 NTU, a data packet is generated:

[0063] Example of data packet: {"module":"NF-04","time":"2024-03-15T09:30:25.456","pressure":1.1,"turbidity":200}

[0064] The edge server runs a hybrid LSTM-CNN model trained with the TensorFlow framework. The input layer receives 72 hours of historical data (12-dimensional parameters × 2160 time points). After extracting local features through a 1D-CNN layer (convolution kernel 3×3, 64 channels), it is input into a double-layer LSTM (128 neurons in the hidden layer), and the predicted value of the membrane fouling degree for the next 24 hours (0-1 scale) is output. The model is trained using the Adam optimizer (learning rate 0.001, batch size 32), the loss function is the mean squared error (MSE), and the prediction error < 8%. For example, inputting the pressure (0.8 - 1.2 MPa) and flow rate (40 - 50 m 3 / h) sequences of the reverse osmosis unit in the past 3 days, the model outputs a predicted fouling degree value of 0.87 for the next 12 hours.

[0065] Then, the edge server continues to generate dynamic control instructions according to the following situations: If the predicted fouling degree < 0.3 and the turbidity < 50 mg / L, close 50% of the modules (3 modules are reserved); if the fouling > 0.6 or the turbidity > 100 mg / L, start all modules. The control instructions include the valve action sequence:

[0066] Close the electric valve of the sub-outlet pipe of the target module (response time < 0.5 seconds);

[0067] After a 2-second delay, close the pneumatic valve of the branch inlet pipe (to prevent water hammer effect);

[0068] Gradually open the valves of the redundant modules to the target opening degree (such as from 0% to 80%) according to the PID algorithm (Kp = 0.8, Ti = 120 s).

[0069] Pressure regulation: Total inlet pipe pressure setting value = base pressure (0.5MPa) + AI compensation value (0-0.3MPa). For example, when the predicted membrane flux drops to 60% of the design value, the pressure is increased from 0.6MPa to 0.8MPa, and the flow valve opening is adjusted from 60% to 75%.

[0070] Buffer storage tank management: Based on the AI-predicted flow peak (e.g., +15% in the next 3 hours), the upper limit of the first buffer storage tank liquid level is adjusted from 70% to 90%, and the opening of the water inlet pipe electric valve is increased from 50% to 80%, giving priority to storing wastewater with high conductivity (>8000μS / cm).

[0071] When the real-time or predicted contamination level is ≥0.8, the PLC closes the valve on the target module's branch inlet pipe, switches the three-way valve to the backwash path, and starts the backwash pump (pressure 0.6 MPa) for a 5-minute reverse flush. A pressure sensor on the backwash branch pipe monitors the flushing effect. If the pressure difference drops by less than 0.1 MPa within 5 minutes (for example, from 0.6 MPa to 0.55 MPa), the flushing time is extended to 8 minutes and chemical cleaning is prepared.

[0072] When contamination exceeds 1.0 or conductivity exceeds 8000 μS / cm, a metering pump injects cleaning agent (pH <2 or >12) at a rate of 3 L / min and circulates for 20 minutes. The cleaning agent temperature is maintained at 25-40°C by a plate heat exchanger to prevent membrane damage. The cleaning wastewater is discharged to a collection tank within the plant via a collection branch pipe (DN40 PP).

[0073] The dose of the UV disinfection component is calculated according to the formula Dose = Intensity × t × 0.001 (intensity 300μW / cm 2 , exposure time 60 seconds, dose 18mJ / cm 2 The ozone dosage is linked to the COD value: when the COD is 30mg / L, add 10mg / L, and for every 5mg / L increase in COD, add 5mg / L of ozone. If the outlet water sensor detects COD > 30mg / L three times in a row, the PLC sends a pressure increase command (e.g., 1.2MPa to 1.4MPa) to the upstream reverse osmosis unit and increases the speed of the spray mechanism's variable frequency water pump (50Hz to 60Hz).

[0074] In the application scenario of electronic electroplating plant, the system treats nickel-containing wastewater (COD 800-1200mg / L, Ni 2 +50mg / L), AI prediction can extend the reverse osmosis membrane cleaning cycle from 24 hours to 40 hours, reduce the amount of chemical cleaning agents by 35%, and reduce the effluent Ni 2+ Concentration ≤ 0.1 mg / L. When the influent turbidity suddenly rises to 800 NTU, the system enables all filtration modules 15 minutes in advance, the liquid level of the buffer storage tank rises from 70% to 95%, and the membrane flux fluctuation is controlled within ±5%. The PLC communicates with the edge server using the OPC UA protocol, the instruction transmission delay < 50 ms, and the model prediction error < 8%.

[0075] In this embodiment, the parts not detailed, such as sensor calibration (calibrating the turbidimeter with a formazin standard solution every month) and valve sealing materials (EPDM rubber), are implemented according to industry standards and are well-known technologies to those skilled in the art.

[0076] Although the exemplary embodiments of the present disclosure have been described, those skilled in the art should understand that various changes and modifications can be made to the exemplary embodiments of the present disclosure without substantially departing from the spirit and scope of the present disclosure. Therefore, all changes and modifications are included within the protection scope of the present disclosure defined by the claims. The present disclosure is defined by the appended claims, and equivalents of these claims are also included.

Claims

1. An industrial wastewater advanced reuse system based on multi-scale membrane separation technology, characterized in that, It includes a control unit, and a pretreatment unit, a low-molecular-weight pollutant separation unit, a nanofiltration membrane unit, a reverse osmosis membrane unit, and a post-treatment unit arranged in sequence from the wastewater inlet end to the wastewater outlet end; First buffer storage tanks, second buffer storage tanks, third buffer storage tanks, and fourth buffer storage tanks are respectively arranged between adjacent units among the pretreatment unit, the low-molecular-weight pollutant separation unit, the nanofiltration membrane unit, the reverse osmosis membrane unit, and the post-treatment unit, and liquid buffer transmission between adjacent units is achieved through the first buffer storage tank, the second buffer storage tank, the third buffer storage tank, and the fourth buffer storage tank; The pretreatment unit includes a pretreatment tank. An inlet is arranged at the left end of the pretreatment tank, and an outlet is arranged at the opposite end. Between the inlet and the outlet, a coarse grid, a sedimentation area, and a chemical reaction area are horizontally arranged in sequence; The low-molecular-weight pollutant separation unit, the nanofiltration membrane unit, and the reverse osmosis membrane unit have the same structure, and each includes a total inlet pipe and a total outlet pipe with pumps, and a number of filtration modules connected in parallel between the total inlet pipe and the total outlet pipe; A filtration membrane for filtration is arranged in the filtration module. The inlet of each filtration module is connected to the total inlet pipe through a branch inlet pipe with a valve, and the outlet is connected to the total outlet pipe through a sub-outlet pipe with a valve; The valves of the branch inlet pipe and the sub-outlet pipe are adjusted by the control unit; The pore size of the filtration membrane arranged in the filtration module varies according to the unit function. Among them, the pore size of the filtration membrane in the low-molecular-weight pollutant separation unit is 0.1 to 10 micrometers, the pore size of the filtration membrane in the nanofiltration membrane unit is 1 to 10 nanometers, and the pore size of the filtration membrane in the reverse osmosis membrane unit is 0.1 to 0.5 nanometers; A hydraulic sensor, a turbidity sensor, and a conductivity sensor are configured at the inlet of each filtration module, and a membrane surface pressure sensor and a temperature sensor matched with the membrane are also installed in each filtration module; A flow regulating valve and a pressure sensor are configured on the total inlet pipe; Data collected by all sensors are transmitted to the control unit through communication lines or modules; The system also includes a backwash water storage tank and a backwash pipeline; the backwash water storage tank is connected to the filtration modules of each unit through a backwash main pipeline, and a backwash pump and a backwash main valve are configured on the backwash main pipeline; a backwash branch pipe is arranged at the outlet of each filtration module, and the backwash branch pipe is connected in parallel with the sub-outlet pipe through a connecting piece with a valve; The system also includes a chemical cleaning agent storage tank. The chemical cleaning agent storage tank is connected to the filtration modules of each unit through a cleaning agent main pipeline, and a metering pump and a cleaning agent main valve are configured on the cleaning agent main pipeline; a cleaning agent branch pipe is arranged at the inlet of each filtration module, and the cleaning agent branch pipe is connected in parallel with the branch inlet pipe through a connecting piece with a valve; A cleaning waste liquid collection branch pipe is arranged on the branch inlet pipe of each filtration module. The collection branch pipe is connected to the branch inlet pipe through a connecting piece with a valve and converges into the cleaning waste liquid main pipeline; The cleaning waste liquid main pipeline is connected to a waste liquid collection tank or a discharge port outside the system, and a waste liquid discharge valve is configured on the waste liquid main pipeline; Three-way valves are arranged on the branch inlet pipe and the sub-outlet pipe of each filtration module; The post-treatment unit includes a working tank, with a spraying mechanism arranged above the interior of the working tank. An ultraviolet disinfection component and an ozone generator are configured inside the working tank. The water outlet of the post-treatment unit is connected to the system water outlet through a pipeline, and a water quality sensor is configured at the water outlet.

2. A control method for the advanced reuse of industrial wastewater based on multi-scale membrane separation technology, characterized in that it is applied to the industrial wastewater advanced reuse system based on multi-scale membrane separation technology as described in claim 1, and is characterized in that, It includes the following steps: Step 1: Parameter collection and data packet generation. Through the hydraulic sensors, turbidity sensors, conductivity sensors, membrane surface pressure sensors, and temperature sensors configured at the liquid inlets of each filtration module, the pressure, suspended solid concentration, pollution degree, membrane surface pressure, and temperature data of the wastewater are collected in real time. At the same time, the wastewater flow rate and total pressure data are collected through the flow regulating valve and pressure sensor at the total liquid inlet pipe, and the bacterial content and chemical oxygen demand value of the reclaimed water are collected through the water quality sensor at the water outlet. The control unit receives the data collected by the above sensors, generates a data packet containing the unit identifier, time stamp, and corresponding parameter values, and transmits it to the control unit through the communication system for subsequent analysis and processing. The generated data packet is also stored in the historical database of the control unit as the training data for the AI deep learning model. Step 2: Data processing, membrane pollution assessment, and prediction. The control unit preprocesses the received data packet, including outlier filtering, denoising, and standardization operations to adjust the data to a unified scale. Based on the preprocessed data, the control unit calculates the current membrane pollution degree of each filtration module using the membrane surface pressure sensor data and flow rate data. The calculation formula is: Among them, ΔR m (t) represents the current membrane fouling degree, ΔR m (t) is the initial membrane surface pressure, P(t) is the current membrane surface pressure, J(t) is the current membrane flux, and the membrane flux is calculated from the total inlet pipe flow rate and the effective membrane area of the filtration module; the control unit further predicts the membrane fouling trend through a built-in AI deep learning model, which is based on historical data and adopts a convolutional neural network combined with a long short-term memory network architecture. After training, it can predict the membrane fouling degree ΔR m (t + Δt) in a future period of time; the prediction input includes the current data packet and the historical data sequence of the past 72 hours, and the output is the time series curve of the membrane fouling degree; the control unit combines the current evaluation result and the predicted value, analyzes the change trends of the wastewater suspended solid concentration, fouling degree and the treatment load of each unit, and provides a basis for subsequent control decisions; Step 3: Dynamic regulation of filtration modules and AI optimization control decision-making. The control unit dynamically regulates the operating states of the filtration modules of each unit according to the current membrane pollution assessment results in Step 2 and the predicted membrane pollution trend by the AI. The control unit determines the number of active filtration modules in each unit based on the membrane fouling degree ΔR m (t+Δt) predicted by AI and the real-time turbidity data: when the prediction shows that ΔR m (t+tΔt) is less than 0.3 within the next 24 hours and the turbidity is lower than 50 mg / L, the control unit closes the sub-outlet pipe valves of some filtration modules to reduce the number of active modules to the minimum operating requirement; when the prediction shows that ΔR m (t+tΔt) is greater than 0.6 or the turbidity exceeds 100 mg / L, the control unit turns on all filtration modules in advance to improve the processing capacity. The adjustment process refers to the pollution peak time predicted by the AI model to ensure that the change in the number of modules matches the dynamic wastewater load; The control unit dynamically adjusts the operating pressure and the influent flow rate according to the real-time hydraulic data and the predicted membrane flux change trend by AI: when the prediction shows that ΔR m (t + tΔt) is less than 0.2 and the membrane flux is maintained above 80% of the design value, the control unit reduces the pressure of the total influent pipe and decreases the opening of the flow valve to optimize the energy consumption; when the prediction shows that ΔR m (t + tΔt) is between 0.4 and 0.6 and the membrane flux will drop below 60% of the design value, the control unit increases the operating pressure and the flow rate in advance to avoid the decrease in treatment efficiency. The adjustment parameters are determined by the optimization curve trained by the AI model based on historical data and corrected according to the temperature data to ensure that the operation is within the working range of the membrane; The control unit uses the predicted wastewater flow rate and pollution load trend by the AI model to adjust the storage capacities of the first buffer storage tank to the fourth buffer storage tank. When the predicted flow rate will exceed 110% of the design capacity of each unit within the next 6 hours, the control unit increases the storage volume of the corresponding buffer storage tank and preferentially allocates it to the upstream storage tank to relieve the pressure on the downstream membrane units. When the predicted flow rate drops below the design value, the wastewater in the storage tank is gradually released, and the distribution decision combines the predicted conductivity value. If the pollution degree is expected to increase, the control unit adjusts the storage strategy in advance. The control unit switches the state of the filtration module in advance according to the ΔR m (t + tΔt) value predicted by AI: when it is predicted that a certain module will reach the cleaning threshold within the next 12 hours, the control unit closes the valve of its sub-outlet pipe, switches to the standby state, and enables the redundant module to take over the operation. The switching time window is optimized by the AI model according to the pollution accumulation rate to ensure stable flow; Through the above regulation, the AI deep learning and prediction technology optimize the control decision-making, enabling the system to proactively respond to changes in wastewater load and membrane pollution and improving the operating efficiency. Step 4: Execute the membrane cleaning procedure. When the current membrane fouling degree ΔR m (t) or the predicted value ΔR m (t + Δt) reaches the preset threshold, the control unit starts the cleaning procedure for the corresponding filtration module; the cleaning method is determined according to the fouling degree, membrane material, and conductivity data: when the fouling degree is between 0.8 and 1.0, the control unit closes the connection valve between the branch inlet pipe and the main inlet pipe, adjusts the three-way valve to the backwash path, starts the backwash pump to transport clean water, and each time lasts for 5 minutes; when the fouling degree exceeds 1.0 or the conductivity exceeds 8000 μS / cm, the control unit adjusts the three-way valve to the chemical cleaning path, starts the metering pump to inject the cleaning agent, and the cleaning time is 20 minutes; the cleaning waste liquid is discharged through the cleaning waste liquid collection branch pipe; the cleaning cycle is adjusted by the pollution accumulation rate predicted by the AI model; Step 5: Post-treatment and water quality assurance. After the wastewater is treated by multi-scale membrane separation, it enters the post-treatment stage. The control unit drives the spraying mechanism to distribute the wastewater, activates the ultraviolet disinfection component and the ozone generator to remove microorganisms and organic pollutants. The water quality sensor at the water outlet monitors the bacterial content and COD value. If it exceeds the standard, the control unit adjusts the ozone dosage or the pressure of the previous membrane unit according to the AI prediction until the water quality reaches the standard. Step 6: Real-time feedback and AI adaptive optimization. After each treatment cycle, the control unit evaluates the system performance according to the sensor feedback data. If the membrane pollution or water quality fluctuation exceeds the preset range, the operating parameters are adjusted. The control unit updates the AI model using the real-time data to optimize the prediction accuracy and control strategy through online learning.

Images