Data-driven water volume-water temperature-water quality collaborative intelligent control circulating water waste heat utilization system and method

Abstract

The invention discloses a data-driven water volume-water temperature-water quality collaborative intelligent control circulating water waste heat utilization system and method. The method comprises the following steps: collecting system operation parameters through a sensor network; in the water temperature control link, evaporation loss is estimated based on a mass transfer model, the water feed-forward flow is calculated and adjusted through a discretization dynamic heat balance equation, and PID feedback fine adjustment is combined, so that the water making and inlet temperature is accurately controlled; in the water quality control link, an LSTM neural network is used for predicting the scaling tendency and the microbial risk, feed-forward adding of a scale inhibitor and a bactericide is achieved, real-time parameter feedback correction is combined, meanwhile, a safety interlocking mechanism is arranged, and the corresponding system comprises a circulating water pool module, an integrated monitoring module, a cooperative execution module and a system control module. According to the invention, the problem that the water temperature control of the circulating water waste heat utilization system lags behind and the dosing strategy is not coordinated is solved, and the collaborative optimization of energy consumption and chemical consumption is realized.

Description

Technical Field

[0001] This invention belongs to the field of power plant water treatment technology, specifically relating to a data-driven intelligent control system and method for recycling waste heat from circulating water, which coordinates water quantity, water temperature, and water quality. Background Technology

[0002] In thermal power plants, condensing steam turbines cool exhaust steam with circulating water, generating a large amount of low-temperature (typically between 27 and 35°C) recycled water. This low-grade waste heat is usually discharged directly into the atmosphere through cooling towers, resulting in energy waste. To recover this waste heat, existing technologies propose introducing the recycled water into the water treatment system as feedwater for the demineralized water treatment system (ultrafiltration-reverse osmosis-deep desalination unit). This reduces the amount of steam used to improve reverse osmosis efficiency, thereby achieving energy conservation and consumption reduction.

[0003] However, this technical approach has two prominent technical problems that have not been effectively resolved for a long time: First, there is the problem of lagging water temperature control. Existing technologies typically employ simple pipe mixing or direct intake, lacking effective temperature control methods. Circulating water temperature fluctuates drastically due to factors such as season, ambient temperature, and unit load, while reverse osmosis systems require a relatively constant influent temperature to ensure stable operating efficiency and membrane lifespan. Especially when the mixing point is located in the circulating water tank, the system has high thermal inertia, and traditional feedback control is slow to respond, failing to achieve precise constant temperature control of the influent temperature leading to the water treatment system.

[0004] Second, there is a conflict between water quality and chemical consumption. Due to extensive reuse, the concentration ratio of the circulating water system decreases, reducing the tendency to scale and decreasing the demand for scale inhibitors. However, the use of circulating water with poor quality (rich in microorganisms and organic matter) as makeup water for the water purification system significantly increases the demand for bactericides (such as sodium hypochlorite) and introduces potential risks of membrane fouling and oxidative damage. Current technologies lack synergistic intelligent optimization of the dosing strategies for both systems, making it impossible to minimize chemical consumption while ensuring system safety.

[0005] In summary, existing technologies cannot simultaneously address the three coupled issues of water volume matching, constant water temperature, and coordinated water quality optimization, leading to unstable operation, limited energy-saving benefits, and high operating costs in waste heat recovery systems. There is an urgent need in this field for an intelligent system capable of coordinated water quality optimization, proactive prediction, and precise control of water temperature, in order to achieve global optimization of energy and chemical consumption while ensuring system safety. Summary of the Invention

[0006] To address the aforementioned problems, the present invention aims to provide a data-driven intelligent control system and method for the recycling of waste heat from circulating water, encompassing water quantity, temperature, and quality.

[0007] The specific technical solution is as follows:

[0008] A data-driven intelligent control system for water quantity, temperature, and quality collaborative management of circulating water waste heat utilization is installed inside a cooling tower. The system inlet is connected to a condenser, the system outlet is connected to a water treatment system, and the system regulating water supply point is connected to a raw water pretreatment system. The circulating water waste heat utilization system includes: The system includes a circulating water tank module, an integrated monitoring module, a collaborative execution module, and a system control module. The circulating water tank module includes a circulating water tank body, which is divided into multiple sequentially connected chambers by vertical partitions, including an inlet chamber, a mixing chamber and a water intake chamber. The mixing chamber is equipped with a mechanical stirring device. The integrated monitoring module includes a flow monitoring unit, a temperature monitoring unit, and a water quality monitoring unit. The flow monitoring unit and the temperature monitoring unit are respectively located at the inlet, mixing point, and outlet of the circulating water waste heat utilization system, while the water quality monitoring unit is located inside the water intake chamber. The collaborative execution module includes a regulating water supply pipeline extending into the mixing chamber and a dosing device for water quality control; The system control module is a programmable controller, whose signal input terminal is connected to the integrated monitoring module and its control output terminal is connected to the collaborative execution module.

[0009] The raw water pretreatment system includes a coagulation sedimentation unit and a clear water tank. The water production system includes a deep filter and a chemical-grade water production tank. The water production system is connected to the desalination system, which includes a pre-desalination system, a deep desalination system, and a desalination tank.

[0010] Furthermore, the water inlet chamber is the condenser return water injection chamber, the mixing chamber includes a pre-mixing buffer chamber, a mechanically forced mixing chamber, and a post-mixing buffer chamber connected in sequence, and the water intake chamber is the water intake chamber of the water production system. The condenser return water injection chamber, the pre-mixing buffer chamber, and the mechanically forced mixing chamber are connected by an overflow weir at the top, and the post-mixing buffer chamber and the water intake chamber of the water production system are connected by a flow passage at the bottom.

[0011] Furthermore, the mechanical stirring device is a submersible mixer, located at the bottom of the mixing chamber.

[0012] Furthermore, there are two submersible mixers, which are respectively installed in the pre-mixing buffer chamber and the mechanically forced mixing chamber.

[0013] Furthermore, the water inlet at the outlet of the circulating water waste heat utilization system has a funnel-shaped structure and is fitted with a cylindrical anti-vortex grid on the outside.

[0014] Furthermore, a condenser return water injection pipe is provided at the condenser return water injection chamber. The outlet end of the condenser return water injection pipe extends below the water surface in the condenser return water injection chamber and is equipped with an upwardly inclined diffuser. The inlet end of the condenser return water injection pipe is connected to the condenser. A spray diffuser is installed at the outlet end of the regulating water makeup pipeline. An electric regulating valve is provided on the regulating water makeup pipeline. The inlet end of the regulating water makeup pipeline is connected to the raw water pretreatment system.

[0015] Furthermore, the water quality monitoring unit includes an online water quality analyzer, which integrates at least a pH sensor, a conductivity sensor, and a residual chlorine sensor.

[0016] Furthermore, the dosing device includes a scale inhibitor dosing device connected to the condenser return water injection chamber and a bactericide dosing device installed at the water intake at the outlet of the circulating water waste heat utilization system.

[0017] Furthermore, the flow monitoring unit includes flow sensors installed at the inlet, mixing point and outlet of the circulating water waste heat utilization system, and the temperature monitoring unit includes temperature sensors installed at the inlet, mixing point and outlet of the circulating water waste heat utilization system, as well as a multi-point temperature sensor array installed inside the mechanically forced mixing chamber.

[0018] A method for coordinated intelligent control of water quantity, water temperature, and water quality using the above-mentioned circulating water waste heat recovery system includes the following steps:

[0019] 1) Data Acquisition and Preprocessing: Real-time acquisition of operating parameters of the circulating water waste heat recovery system, followed by first-order low-pass digital filtering and effectiveness checks. Operating parameters must include at least the condenser return water flow rate Q. c Condenser return water temperature T c , Adjust water supply flow rate Q r Adjusting the water supply temperature T r Water intake flow rate Q of the water treatment system m Water intake temperature T of the water treatment system m Temperature T of the circulating water tank pool Sewage discharge flow rate Q b Circulating water conductivity EC c And ambient temperature, humidity, and wind speed;

[0020] 2) Intelligent water temperature control: Based on the mass transfer model and real-time data collection of ambient wind speed, ambient humidity, and circulating water tank temperature, the evaporation loss flow rate Q of the circulating water tank is estimated online. e Based on the real-time data collected in step 1), the estimated evaporation loss flow rate Q e and the preset target water temperature T entering the water purification system set By solving the discretized dynamic heat balance equation, the setpoint Q of the regulating water feedforward flow required to maintain the target water temperature is calculated.rff Measure the actual temperature T at the water intake of the current water treatment system. m Calculate its relationship with the target water temperature T. set The deviation is calculated using an incremental PID algorithm to determine the feedback correction value ΔQ for regulating water flow. r Adjust the water feedforward flow setpoint Q rff With feedback correction value ΔQ r Adding them together, we get the final set value Q for regulating the water flow rate. rset It also outputs a signal to control and adjust the opening degree of the electric water valve;

[0021] 3) Intelligent Optimization Control for Water Quality: This includes scale inhibitor dosing control and bactericide dosing control. The scale inhibitor dosing control process is as follows: The current Langerile Saturation Index (LSI) is calculated based on real-time collected water quality parameters. A pre-trained Long Short-Term Memory (LSTM) neural network model is used to predict the LSI change trend. Based on the range of the predicted maximum LSI value, the feedforward baseline amount D for scale inhibitor dosing is determined. base The process involves PI feedback correction based on the deviation between the current measured LSI value and the target value, synthesizing the scale inhibitor dosage setpoint, and controlling the metering pump dosing. The bactericide dosing control process is as follows: based on the real-time collected water intake temperature T of the water treatment system... m Based on the total organic carbon (TOC) and the historical dosage of non-oxidizing bactericides, the current microbial risk index R is calculated. An LSTM model is then used to predict the trend of the risk index. Based on the range of the predicted maximum risk index value, the feedforward baseline amount C for sodium hypochlorite dosage is determined. base The system combines the deviation between the current measured residual chlorine value and the target value in the influent with PI feedback correction to synthesize the sodium hypochlorite dosage setpoint and control the metering pump dosing. At the same time, a safety protection mechanism based on high residual chlorine interlock and non-oxidizing bactericide shock dosing procedure is set up.

[0022] Furthermore, the discretized dynamic heat balance equation in step 2) is specifically as follows: , Where k is the current control cycle number, ρ is the density of water (kg / m³), Cp is the specific heat capacity of water (kJ / (kg·℃)), λ is the latent heat of vaporization of water (kJ / kg), V is the effective volume of the circulating water tank (m³), and Δt is the control cycle duration (Q). rff To adjust the water feedforward flow setpoint, m 3 / h;Q c The condenser return water flow rate is given in m. 3 / h;Q r To regulate the water supply flow rate, m 3 / h;Q m The water intake flow rate of the water treatment system is m. 3 / h;Qe The evaporation loss flow rate of the circulating water tank is m. 3 / h.

[0023] Furthermore, in step 3), the scale inhibitor dosing control cycle is 15 minutes. The input to the LSTM model is a historical data sequence of LSI value, calcium hardness, total alkalinity, pH value, water temperature and concentration ratio collected every 15 minutes over the past 24 hours. The output is a predicted LSI value sequence for the next 4 hours. The safety protection mechanism in step 3) includes: when the residual chlorine value of the influent continuously exceeds the safety threshold, the sodium hypochlorite dosing pump is forcibly stopped. Based on the preset cycle or the risk index continuously exceeding the standard, the shock dosing procedure of the non-oxidizing bactericide is automatically triggered.

[0024] Furthermore, in step 3), the process of controlling the scale inhibitor dosage also includes a periodic evaluation and update mechanism for the scale inhibitor dosage LSTM model: periodically calculate the mean absolute error (MAE) between the model's predicted value and the actual value, and automatically retrain and update the model using recent historical data when the MAE continuously exceeds the set threshold.

[0025] Further, in step 3), the formula for calculating the microbial risk index R is: , Where k is the current control cycle number, T is the influent temperature (°C), TOC is the total organic carbon value (mg / L), and Δt is the total organic carbon value (mg / L). shock The time elapsed since the last non-oxidizing bactericide shock treatment is measured in hours (h).

[0026] The beneficial effects of this invention are as follows:

[0027] 1) This invention establishes a dynamic thermal balance model to calculate in real time the regulating water flow required to offset all known disturbances and maintain the target water temperature, and takes action in advance; at the same time, it is supplemented by high-precision feedback to correct model errors; and it ensures the representativeness of temperature measurement and control through forced mixing.

[0028] 2) This invention constructs a water quality collaborative intelligent optimization control system of "perception-prediction-optimization-execution", establishes a multi-step water quality prediction model and a multi-objective optimization decision model, and realizes the advanced, precise and synergistic addition of scale inhibitors and bactericides. Attached Figure Description

[0029] Figure 1 A schematic diagram of the original water treatment system before its modification;

[0030] Figure 2 This is a schematic diagram of the process after the water treatment system and the circulating water waste heat utilization system are coupled.

[0031] Figure 3A schematic diagram of a circulating water waste heat recovery system controlled by the water volume, water temperature and water quality system in a cooling tower;

[0032] Figure 4 This is the control logic diagram of the present invention.

[0033] In the diagram: 1. Coagulation and sedimentation device; 2. Deep filter; 3. Chemical-treated water tank; 4. Steam heater; 5. Pre-desalination system; 6. Deep desalination system; 7. Desalinated water tank; 8. Clear water tank; 9. Cooling tower; 10. Condenser; 11. Circulating water tank body; 12. Adjusting water supply pipeline; 13. Condenser return water injection chamber; 14. Pre-mixing buffer chamber; 15. Mechanically forced mixing chamber; 16. Post-mixing buffer chamber; 17. Water intake chamber of water treatment system; 18. Overflow weir; 19. Flow hole; 20. Submersible agitator; 21. Anti-vortex grid; 22. Condenser return water injection pipe; 23. Diffusion distributor; 24. Spray distributor; 25. Electric regulating valve; 26. Online water quality analyzer; 27. Scale inhibitor dosing device; 28. Bactericide dosing device; 29. ​​Multi-point temperature sensor array. Detailed Implementation

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of protection of the present invention is not limited thereto.

[0035] like Figure 1 The diagram shown is a schematic of the process before the water treatment system was modified. It operates independently from the circulating water system. After the raw water is settled by the coagulation sedimentation device 1, it enters the deep filter tank 2 for filtration. After being treated by the chemical water tank 3, the effluent needs to be heated by the steam heater 4 and then enters the pre-desalination system 5 (ultrafiltration-reverse osmosis), the deep desalination system 6 (ion exchange / EDI), and finally collected by the desalinated water tank 7.

[0036] like Figure 2 As shown, after the modification, the raw water treatment system includes a raw water pretreatment system, a circulating water waste heat utilization system, a water production system, and a desalination system. The raw water pretreatment system includes a coagulation sedimentation device 1 and a clear water tank 8. The circulating water waste heat utilization system includes a cooling tower 9 and a condenser 10. The water production system includes a deep filter 2 and a chemical-grade water tank 3. The water production system is connected to the desalination system, which includes a pre-desalination system 5, a deep desalination system 6, and a desalinated water tank 7. The water production system is coupled with the circulating water system. After sedimentation in the coagulation sedimentation device 1, the raw water enters the clear water tank 8 and then enters the cooling tower 9. The warm return water from the condenser 10 is directly introduced before the deep filter 2 at the front end of the water production system, using the waste heat to replace the low-temperature raw water. Therefore, the steam heater 4 can be shut down or its load reduced. At the same time, some of the low-temperature raw water is treated by coagulation sedimentation and then used as regulating water to replenish the circulating water tank of the cooling tower 9. By intelligently regulating its flow rate, it is mixed in the tank to maintain a constant water intake temperature and maintain the low concentration ratio of the circulating water.

[0037] like Figure 3 As shown, the circulating water pool inside the cooling tower 9 serves as a circulating water waste heat utilization system controlled by the water volume-water temperature-water quality system. It includes a circulating water pool module, an integrated monitoring module, a collaborative execution module, and a system control module.

[0038] The circulating water tank module includes a circulating water tank body 11. The interior of the circulating water tank body 11 is divided into multiple sequentially connected chambers by vertical partitions, including an inlet chamber, a mixing chamber, and a water intake chamber. The inlet chamber is the condenser return water injection chamber 13. The mixing chamber includes a pre-mixing buffer chamber 14, a mechanically forced mixing chamber 15, and a post-mixing buffer chamber 16, which are sequentially connected. The water intake chamber is the water intake chamber 17 of the water production system. The condenser return water injection chamber 13, the pre-mixing buffer chamber 14, and the mechanically forced mixing chamber 15 are connected by an overflow weir 18 at the top. The post-mixing buffer chamber 16 and the water intake chamber 17 of the water production system are connected by a flow hole 19 at the bottom, forming a tortuous water flow path of "overflow and downward flow", which greatly extends the mixing process. The mixing chamber is equipped with a mechanical stirring device, which is a submersible stirrer 20. There are two submersible stirrers 20, which are respectively set in the bottom of the pre-mixing buffer chamber 14 and the bottom of the mechanically forced mixing chamber 15. They can form a complex three-dimensional circulation in the mixing chamber, making the mixing more thorough. A condenser return water injection pipe 22 is provided at the condenser return water injection chamber 13. The outlet end of the condenser return water injection pipe 22 extends into the water surface of the condenser return water injection chamber 13 and is equipped with an upwardly inclined diffuser 23. The inlet end of the condenser return water injection pipe 22 is connected to the condenser 10. The outer wall of the condenser return water injection pipe 22 is wrapped with a heat insulation layer.

[0039] The integrated monitoring module includes a flow monitoring unit, a temperature monitoring unit, and a water quality monitoring unit. The flow monitoring unit and temperature monitoring unit are respectively located at the system's inlet, mixing point, and outlet, acquiring real-time data on the system's input and output water volume and heat. The water quality monitoring unit is located inside the water intake chamber and includes an online water quality analyzer 26, which integrates at least a pH sensor, a conductivity sensor, and a residual chlorine sensor. The flow monitoring unit includes flow sensors located at the system's inlet, mixing point, and outlet. The temperature monitoring unit includes temperature sensors located at the system's inlet, mixing point, and outlet, as well as a multi-point temperature sensor array 29 located inside the mechanically forced mixing chamber 15. The multi-point temperature sensor array 29 is used to verify the mixing uniformity.

[0040] The collaborative execution module includes a regulating water supply pipeline 12 extending into the mixing chamber and a dosing device for water quality control. A spray distributor 24 is installed at the outlet end of the regulating water supply pipeline 12. An electric regulating valve 25 is provided on the regulating water supply pipeline 12. The inlet end of the regulating water supply pipeline 12 is connected to the raw water pretreatment system. The dosing device includes a scale inhibitor dosing device 27 connected to the condenser return water injection chamber 13 and a bactericide dosing device 28 located at the water intake at the system outlet.

[0041] The system control module is a programmable logic controller (PLC). Its signal input terminals are connected to all sensors in the integrated monitoring module, and its control output terminals are connected to the actuators in the collaborative execution module. Direct physical and electrical connections are established with all sensors and actuators in the integrated monitoring module via cables, ensuring real-time aggregation of water volume, temperature, and quality data, and rapid distribution of control commands. This lays the foundation for collaborative control at the hardware level. The system's water intake is a funnel-shaped structure, externally fitted with a cylindrical anti-vortex mesh 21. The system's water intake is connected to the water production system via an electric water intake valve and a water intake pump, ensuring a stable water flow that does not affect the flow field and mixing effect within the water tank.

[0042] Sewage outlet: Located at the lowest point of the circulating water tank body 11, a sewage discharge electric regulating valve is installed.

[0043] Flow monitoring unit: A flow sensor is installed at condenser recovery injection pipe 22 to measure flow rate Qc; a flow sensor is installed at regulating water makeup pipe 12 to measure flow rate Qr; a flow sensor is installed at water intake pipe of water treatment system to measure flow rate Qm; and a flow sensor is installed at sewage discharge pipe to measure flow rate Qb.

[0044] Temperature monitoring unit: Pt100 platinum resistance thermometers (accuracy ±0.1℃, response time <5s) are installed on the condenser recovery injection pipe 22, the regulating water supply pipe 12, and the water production and intake pipe to measure the temperature T respectively. c T r T m .

[0045] A multi-point temperature sensor array is arranged inside the mechanically forced mixing chamber 15. The central measuring point is located 1.5 meters underwater, and the other two points are located 1 meter and 2 meters underwater, respectively. The average value of the three measurements is taken as the temperature T of the circulating water pool. pool。

[0046] Water quality monitoring unit: An online conductivity meter (measuring range 0-2000 μS / cm) is installed in the circulating water tank to measure the EC conductivity of the circulating water. cAn online hardness analyzer (measuring range 0-500 mg / L CaCO3), an online alkalinity analyzer (measuring range 0-500 mg / L CaCO3), and an online pH meter are installed in parallel on the outlet pipeline from the circulating water tank to the condenser; an online conductivity meter is installed on the conditioning water pipeline to measure the EC conductivity of the conditioning water. r An online residual chlorine analyzer (measurement range 0-5 mg / L, accuracy ±0.01 mg / L) and an online total organic carbon analyzer (measurement range 0-50 mg / L) are installed on the water intake pipeline of the water treatment system.

[0047] Environmental parameter monitoring unit: Install an ambient temperature and humidity sensor (temperature range -40~85℃, humidity range 0~100%RH) and an anemometer (measurement range 0-30m / s) near cooling tower 9.

[0048] The collaborative intelligent controller adopts a three-layer architecture of "PLC+DCS+host computer".

[0049] Lower-level actuators: A PLC is used to collect all sensor signals and directly control actuators such as electric valves, metering pumps, and agitators. The PLC communicates with the actuators via a PROFIBUS-DP bus. Intermediate control layer: Employs a distributed control system (DCS) with built-in conventional PID control modules and logic control modules, responsible for controlling the basic loops; Upper-level optimization layer: An industrial computer is used as the host computer, with an operating system and configuration software installed. Software such as water quantity prediction models, water temperature feedforward-feedback composite control algorithms, and water quality collaborative optimization algorithms are also deployed. The host computer communicates with the PLC and DCS via industrial Ethernet.

[0050] like Figure 4 As shown, a data-driven intelligent control method for water quantity-temperature-quality collaborative control of a circulating water waste heat utilization system includes the following steps: Step S100: Data Acquisition and Preprocessing

[0051] 1) At the beginning of each control cycle, the PLC collects raw data from all sensors;

[0052] 2) Perform first-order low-pass digital filtering on the flow and temperature data (the filtering time constant is set according to the instrument response time, usually 3-5 control cycles) to eliminate high-frequency noise;

[0053] 3) Check the validity of the data. If the data of a certain measuring point exceeds the range or the sudden change is beyond the reasonable range, the valid value of the previous cycle or the average value of adjacent measuring points shall be used as a substitute and an alarm shall be issued.

[0054] Step S200: Real-time online estimation of evaporation (Qe) Since direct measurement of evaporation is difficult, this invention uses a mass transfer model calibrated with field data for real-time estimation, the formula of which is: , in: ρ air Air density (kg / m³), calculated based on local atmospheric pressure and ambient temperature Ta.

[0055] A: The surface area (m²) of the circulating water tank is a fixed design parameter.

[0056] V w Wind speed (m / s), from an anemometer.

[0057] W s At a temperature T in the circulating water tank pool Moisture content of saturated air (kg / kg dry air).

[0058] W a The actual moisture content of the current ambient air (kg / kg dry air) is calculated from the ambient temperature Ta and relative humidity RH.

[0059] ρ: Density of water (kg / m³).

[0060] The empirical coefficients 0.00073 and 0.00021 in the formula need to be back-calculated and calibrated using on-site water balance test data over a period of one week during the initial stage of system commissioning.

[0061] Step S300: Calculation of feedforward control quantity based on dynamic thermal balance The goal of this step is to calculate in advance the amount of water needed to maintain the target water temperature T. set (Set at 26℃) Required regulating water flow rate Qr ff .

[0062] 1) Model Establishment: A dynamic thermal balance differential equation is established for the circulating water tank, and discretized and approximated within one control cycle Δt to obtain the algebraic equation for real-time control: , Where ρ is the density of water, kg / m³ 3 Cp is the specific heat capacity of water, kJ / (kg·℃); V is the effective volume of the pool, m³. 3 λ is the latent heat of vaporization of water, taken as 2260 kJ / kg; This is the sensible heat lost from the pool to the environment through the pool walls, in kJ (this term is relatively small and can be ignored or used as a correction term in the initial model).

[0063] 2) Model Discretization and Controller Design To achieve real-time calculations in a digital logic controller (PLC / DCS), the aforementioned continuous equations need to be discretized over short time intervals (i.e., the control period Δt, such as 30 seconds). A backward difference approximation is used to discretize the differential term d(T). pool ) / d t Approximately [T] pool _ set (k) - T pool (k-1)] / Δt, Wherein: T pool_set (k) is the target temperature (T) that the pool hopes to reach in the current period k. set (e.g., 26℃).

[0064] T pool (k-1) represents the temperature of the pool as measured in the previous period (k-1). Substituting the discretized form into the original equation, and using the unknown quantity to be solved in the current period—the regulating water flow rate Q—[the solution is needed]. rff (k)——Simplifying the variables yields the algebraic equations used for real-time feedforward computation.

[0065] 3) Calculation formula for feedforward control quantity The final formula used for calculation is as follows: ,

[0066] Wherein: T set Target water temperature is set to 26 (unit: °C).

[0067] Q c (k): Current condenser return water flow rate (unit: m³ / h).

[0068] T c (k): Current condenser return water temperature (unit: °C).

[0069] T r (k): Current regulated water temperature (unit: °C).

[0070] Q m (k): Current water intake flow rate of the water treatment system (unit: m³ / h).

[0071] Q b (k): Current sewage flow rate (unit: m³ / h). If the sewage flow rate is controlled independently, the stable value or set value of the previous cycle shall be taken.

[0072] T pool (k-1): The temperature of the circulating water tank measured in the previous control cycle (unit: °C), which is obtained by averaging multiple temperature sensors.

[0073] Q e(k): Current evaporation loss flow rate (unit: m³ / h) estimated by step S200.

[0074] V: Effective volume of the circulating water tank (unit: m³), ​​a fixed design parameter.

[0075] ρ: Density of water, taken as 1000 kg / m³.

[0076] Cp: ​​The specific heat capacity of water at constant pressure, taken as 4.18 kJ / (kg·℃).

[0077] λ: Latent heat of vaporization of water, taken as 2260 kJ / kg.

[0078] Δt: Control period, taken as 30 seconds (i.e. 0.008333 hours).

[0079] 4) Constraint and Validity Handling

[0080] Calculated Q rff Must satisfy: 0 ≤ Q rff ≤ Q rmax Q rmax This is used to regulate the maximum supply capacity of the water pipeline. If the limit is exceeded, the output will be based on the limit value, triggering an advanced optimization algorithm or an alarm.

[0081] Step S400: Feedback fine-tuning based on temperature deviation This step employs an incremental digital PID algorithm to precisely correct the feedforward control output, thereby eliminating model errors and unmeasured disturbances. The specific process is as follows: 1) Calculate the current temperature deviation e(k) Read the actual temperature measurement value T at the current water intake. m (k) (unit: °C), calculated relative to the target temperature T set Deviation at (26℃): e(k) = T set - T m (k), e(k) is the temperature deviation (unit: °C) of the current control cycle, and its value can be positive or negative.

[0082] 2) Perform incremental PID calculation Based on the deviations of the current cycle and the previous two cycles, calculate the correction value ΔQ for the regulating water flow. r (k) (unit: m³ / h): , Where: K p Proportional gain (unit: m³ / h / ℃) determines the strength of the response to changes in the current deviation.

[0083] K iIntegral gain (unit: m³ / h / ℃ / cycle) is used to eliminate steady-state deviations, and its effect is related to the cumulative value of the deviation.

[0084] K d Differential gain (unit: m³ / h·period / ℃) reflects the trend of deviation change and plays a role in advance adjustment.

[0085] e(k-1), e(k-2): The temperature deviation values ​​of the previous one and the previous two control cycles, respectively.

[0086] 3) Parameter tuning and output limiting Parameter range: Based on the feedforward-dominated control structure, it is recommended that K... p =0.5~1.2, K i =0.02~0.08, K d =0.1~0.4. The specific value needs to be adjusted on-site based on the system response.

[0087] Output limiting: To prevent integral saturation and overcorrection, ΔQ is limited. r (k) Apply amplitude limiting, for example |ΔQ r (k)| ≤50 m³ / h.

[0088] Output: The calculated and limited correction value ΔQ r (k) Output, used for synthesis in step S500.

[0089] Step S500: Control Variable Synthesis and Execution 1) The final flow setting value sent to the regulating water electric valve XV-01 is: , 2) Q rset (k) Convert to a 4-20mA analog signal or valve position percentage signal and output to the control valve. The valve positioner should be configured with equal percentage characteristics to ensure good linearity of regulation across the full opening range.

[0090] Intelligent dosing control method for scale inhibitors: The scale inhibitor dosing control is based on real-time scaling tendency prediction and adopts a strategy of "model prediction feedforward + feedback fine-tuning". The specific steps are as follows, and the control cycle is 15 minutes.

[0091] Step A100: Real-time scaling tendency index calculation In each control cycle k, the PLC collects and uploads the latest water quality data to the edge computing module. The module calculates the current Langerile Saturation Index (LSI(k)): The calculation formula is as follows: , Where pH(k) is the real-time pH value of the circulating water.

[0092] pCa(k) is the negative logarithm of the calcium ion concentration (calculated as CaCO3, in mg / L), pCa = -log 10 ([Ca 2+ ] / 40.08 * 100).

[0093] pAlk(k) is the negative logarithm of the total alkalinity (in CaCO3, mg / L), pAlk = -log 10 (Alk / 100).

[0094] C(k) is a constant related to the total dissolved solids (TDS) and temperature (T). , TDS can be estimated from conductivity (empirical coefficient 0.65).

[0095] Step A200: LSTM-based fouling trend prediction

[0096] This step is executed by the Long Short-Term Memory (LSI) network model within the edge computing module. Its input consists of 96 consecutive historical data points containing six parameters: LSI value, calcium hardness (CaH), total alkalinity (Alk), pH value, water temperature (T), and concentration ratio (K), collected every 15 minutes over the past 24 hours. In each control cycle, new data points replace the oldest data in the queue, forming a new 96×6 dimensional input matrix X. input The standardized matrix is ​​then input into a pre-trained LSTM model. The model has a fixed structure, consisting of two LSTM layers, each with 128 units, followed by a fully connected output layer with 16 neurons. After inference, the model outputs a vector of length 16, representing the continuous prediction sequence of LSI values ​​for the next 4 hours (16 periods). pre The output sequence is smoothed by a three-point moving average filter and then stored in memory for subsequent decision-making.

[0097] Step A300: Decision on scale inhibitor dosage

[0098] This step calculates the final dosing instruction based on the prediction results. First, from the predicted sequence LSI pre Extract the maximum LSI value within the next 4 hours. pre_max According to LSI pre_max If the value falls within the preset range, the baseline dosing concentration D is determined by consulting the feedforward dosing table. base If LSI pre_max If D < -0.5, then D base = 1.0 ppm; if -0.5 ≤ LSI pre_max If D < 0, then D base = 1.5ppm; if LSI pre_max If ≥0, then Dbase = 2.0 ppm. This is the feedforward control value.

[0099] Secondly, calculate the feedback correction amount. Obtain the measured LSI(k) value for the current period and calculate its relationship with the target LSI value. target (Set as -0.3) Deviation e LSI(k) = LSI target - LSI(k). A discrete PI control algorithm is used, based on the formula... , Calculate the correction amount, where the proportionality coefficient Kp LSI Take 0.2, integration coefficient Ki LSI Take 0.02, the integral term Σe LSI(i) Accumulation is only performed when the absolute value of the deviation is less than 0.5, and is limited to ±0.5 ppm to prevent integral saturation.

[0100] Finally, the setpoint is synthesized. The feedforward quantity and the feedback correction quantity are added together to obtain the scale inhibitor dosage concentration setpoint D for this cycle. scale_set (k) = D base + ΔD(k). Constrain this value to ensure it falls between 1.0 ppm and 3.0 ppm. Based on the metering pump's calibration curve (e.g., 2.0 ppm for a frequency of 50 Hz), linearly convert the concentration setpoint to the metering pump frequency command F. scale_set (k) is sent to the PLC for execution.

[0101] Step A400: Periodic Model Evaluation and Update To maintain prediction accuracy, the system employs an automatic evaluation and update mechanism. Every Sunday at midnight, the system initiates an evaluation process, extracting all LSI predictions made by the model one cycle in advance over the past week, along with their corresponding actual measurements, and calculating the Mean Absolute Error (MAE). If the MAE exceeds a set threshold of 0.2 for two consecutive weeks, the model is deemed to have degraded, and the model update process is automatically triggered.

[0102] During the update, the system uses the most recent three months of historical data (after cleaning and validity verification) as the new training set. In the edge computing module, this dataset is used to retrain the LSTM model for 100 epochs using the Adam optimizer and a learning rate of 0.001. After training, the new model is tested on an independent validation set. If its MAE is below 0.18, the old online model is automatically replaced; otherwise, the old model is retained with an alert indicating that manual inspection of the data or model structure is required.

[0103] Intelligent dosing control method for fungicides: The control of bactericides is based on microbial risk prediction and precise control of residual chlorine. It adopts a strategy of "risk prediction feedforward + residual chlorine feedback + safety interlock" with a control cycle of 5 minutes.

[0104] Step B100: Calculation and Prediction of Microbial Risk Index This step calculates and predicts the risk of microbial growth in real time. During each control cycle k (5 minutes), the edge computing module performs the following operations: First, it collects the influent water temperature T(k) (°C) and total organic carbon TOC(k) (mg / L), and reads the number of hours Δt since the last non-oxidizing bactericide shock treatment. shock (k). Using the formula , Calculate the current microbial risk index R(k), which ranges from 0 to 3. Simultaneously, the system maintains a value including R, T, TOC, and water production flow rate Q. m The historical data window (length 576, corresponding to 48 hours) is used. After each periodic update, this data is fed into a dedicated LSTM prediction model (structure: input layer, two 64-unit LSTM layers, 24-unit output layer). The model outputs a predicted R-value sequence for the next 2 hours (24 periods). pre The sequence is filtered using a three-point moving average and then stored in memory. This LSTM model is incrementally trained every Sunday using data from the past month to maintain prediction accuracy.

[0105] Step B200: Decision on fungicide dosage This step determines the sodium hypochlorite dosage based on risk prediction and residual chlorine feedback. First, from the aforementioned R... pre Extracting the maximum value R from the sequence for the next 2 hours pre_max According to R pre_max The preset interval falls into the feedforward dosing table: if R pre_max <1.0, basic available chlorine dosage concentration C base = 0.8 mg / L; if 1.0 ≤ R pre_max If C < 2.0, then C base = 1.0 mg / L; if R pre_max If ≥ 2.0, then C base = 1.2 mg / L. Based on this, the feedforward dosage D is calculated. naocl_ff (k) = C base ×Q m (k) / (1.1 × 0.1) (assuming the drug solution density is 1.1 kg / L and the available chlorine content is 10%).

[0106] Secondly, residual chlorine feedback correction is performed: the current measured residual chlorine value Cl(k) is read, and the result is calculated relative to the target value Cl. target =0.10 mg / L deviation eCl (k). A discrete PI controller is used, according to the formula ΔD naocl (k) = 0.5×e Cl (k) + 0.05 ×Σe Cl (i) Calculate the correction, the integral term is only in |e Cl When (k)| < 0.05, the cumulative effect is limited to ±10 L / h. Finally, the final synthetic dosage setpoint D... naocl_set (k) = D naocl_ff (k) +ΔD naocl (k), and constrained to between 0 and the maximum flow rate of the metering pump (e.g., 125 L / h), and converted into a frequency command for execution.

[0107] Step B300: Safety Interlock and Impact Dosing Procedure

[0108] This step aims to ensure membrane safety and implement enhanced sterilization. A two-level safety interlock is set up: Level 1: If the residual chlorine value Cl(k) exceeds 0.25 mg / L for three consecutive cycles (15 minutes), the PLC will immediately and forcibly stop the sodium hypochlorite dosing pump, triggering a "high residual chlorine alarm" which requires manual reset. Level 2: The system automatically executes a non-oxidizing bactericide shock dosing procedure every two weeks; this procedure is also immediately triggered when R(k) is detected to be greater than 2.5 for 4 consecutive hours. The procedure execution flow is as follows: First, the sodium hypochlorite dosage is reduced to the maintenance level (corresponding to C... base =0.5 mg / L), and simultaneously start the non-oxidizing bactericide metering pump at a concentration of 20 ppm (relative to the water production flow rate Q). m Continuous dosing for 4 hours. During the initial dosing, residual chlorine feedback adjustment is suspended. After the procedure ends, the system automatically returns to normal control mode. All interlock actions and procedure executions are fully recorded.

Claims

1. A data-driven intelligent control system for water quantity, temperature, and quality collaborative management of circulating water waste heat utilization, characterized in that, Located inside the cooling tower, the system inlet is connected to the condenser (10), the system outlet is connected to the water treatment system, and the system regulating water supply point is connected to the raw water pretreatment system. The circulating water waste heat utilization system includes: The system includes a circulating water tank module, an integrated monitoring module, a collaborative execution module, and a system control module. The circulating water tank module includes a circulating water tank body (11). The interior of the circulating water tank body (11) is divided into multiple sequentially connected chambers by vertical partitions, including an inlet chamber, a mixing chamber and a water intake chamber. A mechanical stirring device is provided in the mixing chamber. The integrated monitoring module includes a flow monitoring unit, a temperature monitoring unit, and a water quality monitoring unit. The flow monitoring unit and the temperature monitoring unit are respectively located at the inlet, mixing point, and outlet of the circulating water waste heat utilization system, while the water quality monitoring unit is located inside the water intake chamber. The collaborative execution module includes a regulating water supply pipeline (12) extending into the mixing chamber and a dosing device for water quality control; The system control module is a programmable controller, whose signal input terminal is connected to the integrated monitoring module and its control output terminal is connected to the collaborative execution module.

2. The circulating water waste heat utilization system with data-driven intelligent control of water quantity, water temperature, and water quality as described in claim 1, characterized in that, The water inlet chamber is the condenser return water injection chamber (13). The mixing chamber includes a pre-mixing buffer chamber (14), a mechanically forced mixing chamber (15), and a post-mixing buffer chamber (16) connected in sequence. The water intake chamber is the water intake chamber of the water production system (17). The condenser return water injection chamber (13), the pre-mixing buffer chamber (14), and the mechanically forced mixing chamber (15) are connected by an overflow weir (18) at the top. The post-mixing buffer chamber (16) and the water intake chamber of the water production system (17) are connected by a flow hole (19) at the bottom. The mechanical stirring device is a submersible stirrer (20) located at the bottom of the mixing chamber.

3. The circulating water waste heat utilization system with data-driven intelligent control of water quantity, water temperature, and water quality as described in claim 2, characterized in that, There are two submersible mixers (20), which are respectively installed in the pre-mixing buffer chamber (14) and the mechanically forced mixing chamber (15).

4. The circulating water waste heat utilization system with data-driven intelligent control of water quantity, water temperature, and water quality as described in claim 2, characterized in that, The water inlet of the circulating water waste heat utilization system is a funnel-shaped structure and is fitted with a cylindrical anti-vortex grid (21). A condenser return water injection pipe (22) is provided at the condenser return water injection chamber (13). The water outlet of the condenser return water injection pipe (22) extends into the water surface of the condenser return water injection chamber (13) and is equipped with an upwardly inclined diffuser (23). The water inlet of the condenser return water injection pipe (22) is connected to the condenser (10). A spray distributor (24) is installed at the water outlet of the regulating water supply pipeline (12). An electric regulating valve (25) is provided on the regulating water supply pipeline (12). The water inlet of the regulating water supply pipeline (12) is connected to the raw water pretreatment system.

5. The circulating water waste heat utilization system with data-driven intelligent control of water quantity, water temperature, and water quality as described in claim 4, characterized in that, The water quality monitoring unit includes an online water quality analyzer (26), which integrates at least a pH sensor, a conductivity sensor and a residual chlorine sensor. The dosing device includes a scale inhibitor dosing device (27) connected to the condenser return water injection chamber (13) and a bactericide dosing device (28) set at the water intake at the outlet of the circulating water waste heat utilization system. The flow monitoring unit includes flow sensors set at the inlet, mixing point and outlet of the circulating water waste heat utilization system. The temperature monitoring unit includes temperature sensors set at the inlet, mixing point and outlet of the circulating water waste heat utilization system and a multi-point temperature sensor array (29) set inside the mechanical forced mixing chamber (15).

6. A method for coordinated intelligent control of water quantity, water temperature, and water quality using a circulating water waste heat utilization system as described in claim 5, characterized in that, Includes the following steps: 1) Data Acquisition and Preprocessing: Real-time acquisition of operating parameters of the circulating water waste heat recovery system, followed by first-order low-pass digital filtering and effectiveness checks. Operating parameters must include at least the condenser return water flow rate Q. c Condenser return water temperature T c , Adjust water supply flow rate Q r Adjusting the water supply temperature T r Water intake flow rate Q of the water treatment system m Water intake temperature T of the water treatment system m Temperature T of the circulating water tank pool Sewage discharge flow rate Q b Circulating water conductivity EC c And ambient temperature, humidity, and wind speed; 2) Intelligent water temperature control: Based on the mass transfer model and real-time data collection of ambient wind speed, ambient humidity, and circulating water tank temperature, the evaporation loss flow rate Q of the circulating water tank is estimated online. e Based on the real-time data collected in step 1), the estimated evaporation loss flow rate Q e and the preset target water temperature T entering the water purification system set By solving the discretized dynamic heat balance equation, the setpoint Q of the regulating water feedforward flow required to maintain the target water temperature is calculated. rff Measure the actual temperature T at the water intake of the current water treatment system. m Calculate its relationship with the target water temperature T. set The deviation is calculated using an incremental PID algorithm to determine the feedback correction value ΔQ for regulating water flow. r Adjust the water feedforward flow setpoint Q rff With feedback correction value ΔQ r Adding them together, we get the final set value Q for regulating the water flow rate. rset It also outputs a signal to control and regulate the opening of the electric water valve; 3) Intelligent Optimization Control for Water Quality: This includes scale inhibitor dosing control and bactericide dosing control. The scale inhibitor dosing control process is as follows: The current Langerile Saturation Index (LSI) is calculated based on real-time collected water quality parameters. A pre-trained Long Short-Term Memory (LSTM) neural network model is used to predict the LSI change trend. Based on the range of the predicted maximum LSI value, the feedforward baseline amount D for scale inhibitor dosing is determined. base The process involves PI feedback correction based on the deviation between the current measured LSI value and the target value, synthesizing the scale inhibitor dosage setpoint, and controlling the metering pump dosing. The bactericide dosing control process is as follows: based on the real-time collected water intake temperature T of the water treatment system... m Based on the total organic carbon (TOC) and the historical dosage of non-oxidizing bactericides, the current microbial risk index R is calculated. An LSTM model is then used to predict the trend of the risk index. Based on the range of the predicted maximum risk index value, the feedforward baseline amount C for sodium hypochlorite dosage is determined. base The system combines the deviation between the current measured residual chlorine value and the target value in the influent with PI feedback correction to synthesize the sodium hypochlorite dosage setpoint and control the metering pump dosing; at the same time, a safety protection mechanism based on high residual chlorine interlock and non-oxidizing bactericide shock dosing procedure is set up.

7. The method as described in claim 6, characterized in that, The discretized dynamic heat balance equation in step 2) is specifically as follows: ， Where k is the current control cycle number, ρ is the density of water (kg / m³), Cp is the specific heat capacity of water (kJ / (kg·℃)), λ is the latent heat of vaporization of water (kJ / kg), V is the effective volume of the circulating water tank (m³), and Δt is the control cycle duration (Q). rff To adjust the water feedforward flow setpoint, m 3 / h;Q c The condenser return water flow rate is given in m. 3 / h;Q r To regulate the water supply flow rate, m 3 / h;Q m The water intake flow rate of the water treatment system is m. 3 / h;Q e The evaporation loss flow rate of the circulating water tank is m. 3 / h.

8. The method as described in claim 6, characterized in that, In step 3), the scale inhibitor dosing control cycle is 15 minutes. The input of the LSTM model is the historical data sequence of LSI value, calcium hardness, total alkalinity, pH value, water temperature and concentration ratio collected every 15 minutes in the past 24 hours. The output is the LSI value prediction sequence for the next 4 hours. The safety protection mechanism in step 3) includes: when the residual chlorine value of the influent is detected to continuously exceed the safety threshold, the sodium hypochlorite dosing pump is forcibly stopped. According to the preset cycle or the risk index continues to exceed the standard, the shock dosing program of non-oxidizing bactericide is automatically triggered.

9. The method as described in claim 6, characterized in that, In step 3), the scale inhibitor dosing control process also includes a periodic evaluation and update mechanism for the scale inhibitor dosing LSTM model: the mean absolute error (MAE) between the model's predicted value and the actual value is calculated periodically, and when the MAE continuously exceeds the set threshold, the model is automatically retrained and updated using recent historical data.

10. The method as described in claim 6, characterized in that, In step 3), the formula for calculating the microbial risk index R is: ， Where k is the current control cycle number, T is the influent temperature (°C), TOC is the total organic carbon value (mg / L), and Δt is the total organic carbon value (mg / L). shock The time elapsed since the last non-oxidizing bactericide shock treatment is measured in hours (h).

Images