Fly ash water washing water evaporation salt separation system and salt separation method thereof

Abstract

The present application relates to the technical field of fly ash washing water treatment, and particularly relates to a fly ash washing water evaporation and salt separation system and a salt separation method thereof. The fly ash washing water evaporation and salt separation system is additionally provided with an intelligent electrochemical purification unit between a traditional evaporation and crystallization unit and a potassium chloride crystallization unit, the intelligent electrochemical purification unit is provided with an inclined electrode plate group, and is additionally provided with an intelligent controller based on chemical oxygen demand and water turbidity double-loop feedback. The intelligent controller adopts a double-layer optimization algorithm including model predictive control and deep reinforcement learning, dynamically calculates and cooperatively determines the duty cycle, frequency and peak current density of the pulse power source after safety constraint, and the final execution values of the duty cycle and the frequency are the smaller values of two sets of target values calculated based on COD and turbidity respectively. The present application solves the problems of low potassium salt purity, organic matter accumulation and system fouling in the traditional process, and realizes efficient and stable salt separation of fly ash washing water.

Description

Technical Field

[0001] This invention relates to the field of fly ash wash water treatment technology, and in particular to a fly ash wash water evaporation and salt separation system and its salt separation method. Background Technology

[0002] In recent years, with increasingly stringent environmental protection requirements, the resource utilization and harmless treatment of waste incineration fly ash has become a focus of the industry. Fly ash washing liquid contains high concentrations of sodium chloride (NaCl), potassium chloride (KCl), and small amounts of organic matter and hardness ions. Achieving sodium-potassium salt separation and recovering high-quality salt products through evaporation and crystallization is a key pathway to realize its resource utilization. Currently, the typical fly ash washing water treatment process in the industry is as follows: first, the fly ash washing liquid is evaporated and crystallized to preferentially separate sodium chloride; then, the remaining mother liquor after sodium chloride separation is cooled and crystallized to extract potassium chloride. This technical solution can, in principle, achieve preliminary separation of sodium and potassium salts. However, the above-mentioned existing technology has revealed the following technical defects and problems in long-term industrial operation: During the evaporation and crystallization process of separating sodium chloride, fine sodium chloride crystals that are not completely separated are carried into the subsequent cooling crystallization mother liquor. These fine sodium chloride crystals are difficult to completely separate during the potassium chloride crystallization process, resulting in the final potassium chloride by-product containing sodium chloride, and the salt purity is difficult to meet the product standards.

[0003] Fly ash washing liquid and its concentrated mother liquor typically contain small amounts of recalcitrant organic matter (COD). Traditional evaporation-cooling salt separation processes lack effective online removal methods for this organic matter. With prolonged continuous system operation, organic matter accumulates in the circulating mother liquor, potentially affecting the color and quality of the crystalline salt. Furthermore, it forces the system to periodically discharge large quantities of high-COD mother liquor to control concentration, resulting in material loss and additional hazardous waste treatment costs.

[0004] The calcium and magnesium ions remaining in the mother liquor, such as hardness ions, readily combine with anions like carbonate and sulfate in the solution during high-temperature, concentrated processes like evaporation and crystallization, forming insoluble inorganic salt scale layers such as calcium carbonate and calcium sulfate. These scale layers adhere firmly to critical components like the heat exchanger tube walls and evaporator inner walls. Scale buildup significantly reduces heat exchange efficiency, increases energy consumption, and in severe cases, can even lead to equipment blockage and deterioration of heat transfer, necessitating frequent shutdowns for chemical or mechanical cleaning. This seriously affects the long-term, continuous, and stable operation of the entire salt separation system.

[0005] Therefore, the industry urgently needs a new system integration solution and intelligent control method that can simultaneously remove organic matter, inhibit scaling, and effectively improve the purity of potassium salt products during the salt separation process, so as to achieve efficient, stable, and high-value resource treatment of fly ash washing water. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a fly ash washing water evaporation and salt separation system and its salt separation method in order to solve the problems of low product purity, organic matter accumulation and easy scaling of equipment in the prior art mentioned above.

[0007] The technical solution adopted by this invention to solve its technical problem is: a fly ash washing water evaporation and salt separation system, comprising: An evaporation and crystallization unit is used for evaporating and crystallizing fly ash washing liquid. A sodium chloride separation system is used to separate sodium chloride and the first mother liquor; The intelligent electrochemical purification unit has its inlet connected to the sodium chloride separation system to receive the first mother liquor. The intelligent electrochemical purification unit includes a reactor shell, an inclined electrode plate group, a pulse power supply, a first sensor for monitoring chemical oxygen demand, and a second sensor for monitoring the turbidity of the effluent. A potassium chloride crystallization unit is connected to the supernatant outlet of the intelligent electrochemical purification unit; The intelligent controller is connected to the pulse power supply, the first sensor, and the second sensor signals, and performs dual-loop coordinated control, including: (a) Real-time COD value obtained based on the first sensor and its rate of change And electrode parameters, to calculate the duty cycle of the first target pulse. and the first target pulse frequency And calculate the target peak current density. ; (b) Real-time turbidity value obtained based on the second sensor and its rate of change Calculate the duty cycle of the second target pulse. Second target pulse frequency ; (c) Based on the preset collaborative decision-making rules, determine the final pulse parameters to be executed, wherein the final pulse duty cycle is... Pick and The smaller value in the middle, the final pulse frequency Pick and The smaller value in; (d) Target peak current density With a dynamically calculated upper limit of safe current density The values ​​are compared, and the smaller value is taken as the peak current density for controlling the pulse power supply. (e) According to , The peak current density determined in step (d) controls the operation of the pulse power supply.

[0008] An intelligent electrochemical purification unit was added between the existing evaporation crystallization unit and the potassium chloride crystallization unit. The intelligent electrochemical purification unit integrates tilted electrodes, pulse power supply and multi-parameter sensors, and is controlled by an intelligent controller. The controller simultaneously receives chemical oxygen demand signals from wastewater and effluent turbidity signals, and independently calculates two sets of optimized pulse parameters. Then, through a unique collaborative decision-making rule that takes the smaller value, the final execution command is generated to achieve organic matter degradation, scale control and fine crystal separation.

[0009] Furthermore, the target peak current density Determined using the following feedback-feedforward composite algorithm: , in, To preset the reference current density, This is the real-time COD value. For the target COD value, The initial COD concentration. This represents the real-time COD change rate. The target COD removal rate, and This is an adaptively adjustable gain coefficient.

[0010] This algorithm integrates feedback control, which is based on the deviation between the current COD and the target, and feedforward control, which is based on the prediction of the COD degradation rate. and The setting of two gain coefficients enables the controller to simultaneously correct current errors and suppress future fluctuations in advance, achieving more precise and faster closed-loop control of the organic oxidation rate, optimizing COD removal efficiency, minimizing the accumulation trend of organic matter from the process control level, and solving the problem of large discharge of mother liquor.

[0011] According to an embodiment of the present invention, the duty cycle of the first target pulse The algorithm, which includes a nonlinear concentration feedback term, a current efficiency compensation term, and a cleaning compensation term, is used to determine the specific terms: , in, The preset base duty cycle is η, which is an adaptively adjustable weighting coefficient with a value range of 1.0 to 2.0. and The gain coefficient is adaptively adjustable. For real-time current efficiency, For reference current efficiency, This is a cleaning compensation amount triggered based on the electrode state. When it is determined that the electrode needs to be cleaned, a negative compensation is added.

[0012] In addition to considering COD removal targets and enhancing its capabilities by introducing a nonlinear exponent, the algorithm also incorporates a current efficiency compensation term. When the real-time current efficiency falls below the reference value, the algorithm proactively adjusts the duty cycle to optimize the economics of the reaction process. Furthermore, a cleaning compensation factor is introduced. This creates a time window for electrochemical cleaning of the electrode surface, integrates electrode maintenance requirements into daily control logic, and achieves initial linkage between process optimization and equipment maintenance. While ensuring COD removal efficiency, it also considers the energy economy of the treatment process and begins to address preventive control of scaling, which helps reduce system operating costs, extend the effective working time of the electrodes, and indirectly improve system stability.

[0013] According to an embodiment of the present invention, in step (a), the first target pulse frequency The algorithm, which includes a current efficiency compensation term, a removal rate feedback term, and a cleaning compensation term, is used to determine the specific terms: , in, Based on the base frequency, For real-time current efficiency, For reference current efficiency, This represents the real-time COD change rate. The target COD removal rate, and The gain coefficient is adaptively adjustable. For cleaning compensation frequency, positive compensation is superimposed when it is determined that the electrode needs to be cleaned.

[0014] Saturated nonlinearity processing is applied to the rate deviation to prevent sudden changes or divergence in frequency commands under abnormal operating conditions, thus enhancing system robustness. A current efficiency compensation term is introduced; when the real-time current efficiency is lower than the reference value, or the actual degradation rate is lower than expected, the formula will automatically output a frequency higher than the base value. By increasing the frequency, mass transfer is improved, thereby increasing efficiency. Simultaneously, a cleaning compensation amount is introduced. Electrode maintenance requirements are integrated into daily control logic, achieving initial linkage between process optimization and equipment maintenance.

[0015] According to an embodiment of the present invention, in step (b), the second target pulse frequency Determined using the following algorithm, which is nonlinearly related to turbidity deviation: , in, The base frequency, ΔNTU represents the real-time turbidity value and the set turbidity value. deviation, The gain coefficient is adaptively adjustable. This is the cleaning compensation frequency.

[0016] The hyperbolic tangent function is used to describe the nonlinear saturation relationship between turbidity deviation and frequency adjustment. When the turbidity deviation is small, the adjustment is gradual to maintain stability; as the deviation increases, the adjustment becomes stronger; however, there is an upper limit to prevent over-adjustment. Through smooth and powerful frequency adjustment, the electric field coalescence of fine crystals and the sedimentation of the tilted plate can be more effectively promoted, thus ensuring the control of the purity of potassium chloride in the downstream process.

[0017] According to one embodiment of the present invention, the duty cycle of the second target pulse Determined using the following algorithm: , in, Based on duty cycle, The gain coefficient is adaptively adjustable. The coupling coefficient is... To compensate for the duty cycle during cleaning, negative compensation is added when it is determined that the electrode needs cleaning.

[0018] This algorithm not only directly responds to turbidity deviations, but also introduces a linkage with frequency through a coupling coefficient λ. That is, the duty cycle and frequency, two pulse parameters, are not independent in affecting the sedimentation effect, but are internally coupled. The algorithm design actively considers this coupling and performs collaborative calculations, making the control of turbidity more coordinated and precise. The linkage adjustment of duty cycle and frequency can generate better electric field shape and hydrodynamic conditions, thereby maximizing the sedimentation and separation efficiency of the tilted electrode plate and more thoroughly removing sodium chloride fine crystals.

[0019] According to one embodiment of the present invention, the upper limit of the safe current density Determined using the following algorithm: Real-time current density limit The minimum values ​​of the limiting diffusion current density of the electrode and the upper limit current density of the potential window are taken. , Among them, the upper limit current density of the potential window Determined by the electrode material itself, The limiting diffusion current density of the electrode is , , The average number of electrons transferred during the complete oxidation of organic molecules. for Faraday constant, The mass transfer coefficient of organic matter is... This represents the concentration of organic matter in the bulk.

[0020] According to one embodiment of the present invention, the intelligent controller adopts a two-layer intelligent optimization architecture, specifically including: The first layer is a real-time dynamic optimization layer. During a single processing step, it fine-tunes parameters based on real-time feedback. It employs a model predictive control algorithm to monitor parameters in real time. , and As input, pulse parameters are rapidly optimized and adjusted at the minute or second level; The second layer is the batch learning layer. After a complete batch is processed, deep learning and policy updates are performed. Its reward function, Reward, is as follows: , Among them, α, β, γ, , These are dynamically configurable weighting coefficients, dynamically adjusted by a higher-level meta-controller based on the phased macro-level objectives of the system's operation. This represents the amount of COD degraded per unit time. The amount of COD removed per unit of energy consumption. This is a penalty term, representing the ratio of real-time turbidity to a set value. This is a penalty term for oxygen evolution potential. This is the initial oxygen evolution potential. This represents the change in oxygen evolution potential. An increase indicates that the anodic catalytic oxygen evolution reaction has become difficult, and the coating may be deactivated or the surface may be covered by inactive substances. This is a penalty term for the electrode active area. This represents the change in differential capacitance. As the initial differential capacitance decreases, the effective active area of ​​the surface electrode also decreases. After a complete batch of processing is completed, this layer uses the data from this operation to train the strategy network and the value evaluation network, and updates the control strategy. The reward function is a weighted sum of COD removal rate, processing energy consumption, turbidity and electrode loss. The pulse parameter control strategy is then iteratively optimized and learned offline. The batch learning layer also includes an expert knowledge base for storing the trained policy network parameters; the deep reinforcement learning algorithm is a deep deterministic policy gradient algorithm.

[0021] The first layer employs Model Predictive Control (MPC), responsible for rapid, rolling optimization based on real-time models to cope with transient disturbances. The second layer employs Deep Reinforcement Learning (DRL), a multi-objective reward function that simultaneously encourages greater COD removal per unit time and greater COD removal per unit of energy consumption. DRL learns from historical data and continuously seeks a global strategy that maximizes the reward in the long term. This is a combination of real-time precise control and long-term empirical learning and evolution. The system can not only respond quickly but also learn from historical operations, continuously approaching the globally optimal operating point under specific water quality conditions.

[0022] The Deep Deterministic Policy Gradient (DDPG) algorithm is adopted and equipped with an expert knowledge base. DDPG is particularly suitable for action space optimization of continuous parameters such as current and duty cycle. The expert knowledge base is used to store and call up the parameters of the well-trained control policy network, realizing the accumulation, solidification and reuse of knowledge. The DDPG algorithm ensures the effectiveness of continuous parameter optimization, while the expert knowledge base allows the optimization experience gained by the system to be preserved and passed on, avoiding repeated learning and improving the engineering practical value and intelligence level of the system.

[0023] Furthermore, it also includes an electrode condition monitoring unit for monitoring the slot voltage at both ends of the electrode plate assembly; The intelligent controller executes an adaptive electrode cleaning procedure, including: When the rise in the slot voltage exceeds the first preset threshold, the control pulse power supply performs a polarity reversal operation. After the polarity reversal operation, if the rate of decrease of the cell voltage is lower than the second preset threshold, the pulse power supply is controlled to switch to the descaling pulse mode. The frequency of the descaling pulse mode is 500Hz to 5000Hz, and the duty cycle is 0% to 10%.

[0024] The voltage between the electrode plates is monitored online and used as an indirect and sensitive indicator of the degree of scaling. When the voltage rises, indicating that scaling is intensifying, the polarity is first reversed to loosen the scale layer through electrochemical action. If the effect is not good, it is automatically upgraded to a high-frequency, low-duty-cycle powerful descaling pulse, which generates dense microbubbles for physical flushing. This achieves a transition from passive shutdown cleaning after scaling to online monitoring, active intervention, and graded cleaning during the scaling process.

[0025] Furthermore, during the polarity reversal operation, the intelligent controller controls the current density applied by the pulse power supply. With normal operating current density satisfy: , where β ranges from 0.5 to 1.2.

[0026] The current intensity relationship during the polarity reversal operation is limited, and the polarity reversal current density can be adjusted within the range of 0.5-1.2 times the normal operating current density. This ensures sufficient cleaning force while avoiding damage to the electrodes due to excessive current. This makes the self-cleaning process more controllable and safe, optimizes the balance between cleaning effect and electrode life, and further improves the intelligence and economy of maintenance operations.

[0027] According to one embodiment of the present invention, the intelligent controller performs the following safety interlock protection: When the liquid level in the intelligent electrochemical purification unit is lower than the safe liquid level threshold, the output of the pulse power supply is cut off. When the bottom solid material level is higher than the discharge material level threshold, the discharge valve will automatically open. Based on the start / stop signals from the evaporation and crystallization unit, the output of the pulse power supply is enabled or cut off accordingly.

[0028] This setup prevents equipment damage or safety accidents caused by abnormal operation, equipment failure, or process misalignment, thereby improving the safety and stability of the entire complex system from an automation perspective.

[0029] According to one embodiment of the present invention, the tilted electrode plate assembly includes alternating anode plates and cathode plates, wherein the tilt angle between the anode plates and cathode plates is 50° to 70°. The lower part of the inclined electrode plate assembly is equipped with a collection hopper. The solid discharge port of the collection hopper is connected to the feed end of the evaporation crystallization unit through a pipeline, which is used to reflux the sodium chloride crystals that have settled and separated in the electrochemical purification unit as seed crystals.

[0030] The tilted electrode plate serves the dual function of an electric field reaction interface and a gravity settling slope. The tilted design enhances solid-liquid separation, directly intercepting fine crystals and improving the purity of potassium chloride. The seed crystals are returned to the front-end evaporator crystallizer as seed crystals, forming an internal material cycle. This not only reduces solid waste emissions but also promotes the production of larger and purer sodium chloride crystals in the front-end evaporator crystallizer, improving the overall resource recovery rate and economic efficiency of the process.

[0031] A method for salt separation using the fly ash wash water evaporation and salt separation system described above is also provided, comprising the following steps: S1. The fly ash washing liquid is evaporated and crystallized, and then solid sodium chloride and the first mother liquor are separated by a sodium chloride separation system. S2. The first mother liquor is introduced into the intelligent electrochemical purification unit. While applying a pulsed electric field for electrochemical treatment, dual-loop coordinated control is executed, including: S21. Real-time monitoring of electrode parameters and chemical oxygen demand (COD) values ​​during the treatment process. and its rate of change and effluent turbidity value ; S22, based on and Calculate the duty cycle of the first target pulse and the first target pulse frequency And calculate the target peak current density. ; S23, based on Calculate the duty cycle of the second target pulse Second target pulse frequency ; S24. The duty cycle of the final executed pulse. Determined as and The smaller value in the value will determine the final pulse frequency. Determined as and The smaller value in; S25, Target peak current density With a dynamically calculated upper limit of safe current density The values ​​are compared, and the smaller value is taken as the peak current density for controlling the pulse power supply. S26, according to , And the parameters of the pulse electric field determined by S25 for adjusting the peak current density; S3. Cool the supernatant after S2 treatment to crystallize and separate solid potassium chloride.

[0032] Furthermore, S2 also includes: The cell voltage of the electrodes in the electrochemical purification unit is monitored in real time. When the cell voltage rises abnormally, an adaptive cleaning program is automatically executed, which includes polarity reversal operation and / or high-frequency, low-duty-cycle descaling pulse mode.

[0033] By integrating the adaptive electrode cleaning procedure into the process steps, it is clarified that cleaning and maintenance are automatically triggered and seamlessly embedded based on monitoring data during the salt separation process, rather than being an independent external operation, thus ensuring the continuity of the entire salt separation process and the ability to operate unmanned or with minimal manpower.

[0034] Furthermore, in S2, sodium chloride crystals are separated from the bottom of the electrochemical purification unit and returned to the evaporation and crystallization step of S1 as seed crystals.

[0035] The methodological approach solidifies and optimizes the internal resource recycling pathway, enabling the goals of improving sodium salt product quality and potassium salt product purity to mutually reinforce each other, thereby enhancing the overall resource recovery efficiency and product value of the process.

[0036] The beneficial effects of this invention are: An integrated electrochemical sedimentation reactor is added between the sodium chloride evaporation and crystallization unit and the potassium chloride cooling and crystallization unit. This reactor integrates an inclined electrode plate group and is equipped with a multi-parameter sensing and intelligent collaborative control system. It highly integrates multiple processing units such as electrochemical advanced oxidation, online descaling, crystal regulation and sedimentation separation into a single device, realizing equipment compactness and process integration. By utilizing the electrochemical oxidation function of the anode in the reactor, organic matter in the mother liquor is degraded online, effectively solving the problem of organic matter accumulating due to the lack of online removal methods in traditional processes, which ultimately leads to the forced discharge of large amounts of mother liquor. This reduces the concentration of organic matter at the source, significantly reducing the amount of mother liquor that the system needs to discharge to control the concentration of organic matter, improving resource recovery rate, and reducing the burden of hazardous waste treatment. Through electrochemically induced precipitation, hardness ions are removed, fundamentally alleviating the problems of system scaling, reduced heat exchange efficiency, and unstable operation caused by the deposition of hardness ions in heat exchange parts. This improves the inherent stability of the system, reduces unplanned downtime for cleaning due to scaling, and ensures long-term continuous operation. By applying a specific pulsed electric field, energy and induction are provided for the further precipitation and grain growth of sodium chloride fine crystals. At the same time, the inclined electrode plate itself constitutes a highly efficient inclined plate settler, solving the key quality problem of sodium chloride fine crystals being carried into the downstream potassium salt crystallization system, which leads to a decrease in the purity of potassium chloride products. This achieves the synergy of electric field-induced growth and structural sedimentation, which not only more thoroughly separates sodium chloride fine crystals and uses them as seed crystals for reflux to improve the quality of upstream sodium salts, but also directly ensures the high purity of downstream potassium chloride products. The tilted electrode plate design allows the flowing crystal slurry to continuously scour its surface; at the same time, the electrochemical reaction process itself generates Joule heat, which alleviates the problem of scale adhering to the cathode surface and is difficult to clean, and compensates for the heat loss of the mother liquor during transportation and treatment, realizing self-cleaning of the electrode during operation and extending the maintenance cycle; at the same time, the heat of reaction is effectively utilized, reducing the additional energy consumption required to maintain the system temperature and achieving energy saving. The electrochemical sedimentation integrated reactor of this invention is equipped with a collaborative control system based on dual-loop feedback and dual-layer intelligent optimization. This solves the problem that traditional control methods are difficult to deal with complex, multi-variable, and strongly coupled wastewater treatment processes, and cannot simultaneously optimize multiple objectives such as organic matter removal, fine crystal separation, and energy consumption control. This enables the system to have advanced intelligence with self-adaptation, self-learning, and self-optimization capabilities. It can dynamically adjust operating parameters according to real-time water quality to ensure that overall energy consumption is optimized while meeting the requirements of product purity, system stability, and treatment efficiency. Attached Figure Description

[0037] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0038] Figure 1This is a schematic diagram of the system of the present invention.

[0039] Figure 2 This is a schematic diagram of the intelligent electrochemical purification unit in the system of this invention.

[0040] Figure 3 This is a flowchart of the salt separation method of the present invention.

[0041] In the diagram: 1. Preheating unit; 11. Preheater; 12. Forced circulation heat exchanger; 13. Forced circulation pump; 2. Evaporation and crystallization unit; 3. Sodium chloride separation unit; 4. Intelligent electrochemical purification unit; 41. Reactor shell; 42. Electrode plate assembly; 43. Pulse power supply; 44. First sensor; 45. Second sensor; 46. Collection hopper; 5. Potassium chloride crystallization unit; 6. Intelligent controller. Detailed Implementation

[0042] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0043] Example 1 like Figure 1 As shown, a fly ash wash water evaporation and salt separation system includes a preheating unit 1, an evaporation and crystallization unit 2, a sodium chloride separation unit 3, an intelligent electrochemical purification unit 4, a potassium chloride crystallization unit 5, and an intelligent controller 6. The preheating unit 1 includes a preheater 11, a forced circulation heat exchanger 12, and a forced circulation pump 13. The fly ash wash water sequentially passes through the preheater 11 and the forced circulation heat exchanger 12 before entering the evaporation and crystallization unit 2. The evaporation and crystallization unit 2 is used to treat the fly ash wash water and separate sodium chloride and a first mother liquor. The sodium chloride is discharged to the sodium chloride separation unit 3, where it is processed into solid sodium chloride, packaged, and sold. The first mother liquor can be returned to the evaporation and crystallization unit 2 or enter the intelligent electrochemical purification unit. 4; The feed inlet of the intelligent electrochemical purification unit 4 receives the first mother liquor. The first mother liquor removes COD and hardness ions in the intelligent electrochemical purification unit 4. At the same time, the intelligent electrochemical purification unit 4 can generate an electric field, which can promote the precipitation and enlargement of sodium chloride crystals. Sodium chloride crystals settle and separate fine sodium chloride crystals. After being discharged from the solid discharge port at the bottom, they flow back to the upstream evaporation crystallization unit 2. As seed crystals, they can promote the precipitation and enlargement of sodium chloride crystals. The supernatant is discharged from the top overflow port and enters the downstream potassium chloride crystallization unit 5. The potassium chloride crystallization unit 5 separates potassium chloride and the second mother liquor. Solid potassium chloride is packaged and sold. The second mother liquor is returned to the upstream evaporation crystallization unit 2 for further processing.

[0044] like Figure 2As shown, the intelligent electrochemical purification unit 4 includes a reactor shell 41, an inclined electrode plate assembly 42, a pulse power supply 43, a first sensor 44 for monitoring chemical oxygen demand (COD), and a second sensor 45 for monitoring effluent turbidity. The reactor shell 41 has an inlet on its lower left side, an overflow outlet on its upper right side, and a collection hopper 46 at the bottom with a solid discharge outlet. The inclined electrode plate assembly 42 includes alternately arranged anode and cathode plates with an inclination angle of 50°–70°. The cathode plates are preferably made of stainless steel, while the anode plates are made of materials such as DSA, graphite, BDD, and lead dioxide. A level gauge at the top detects the liquid level, and a level gauge at the bottom detects the height of the sodium chloride crystal slurry. The first sensor 44 and the second sensor 45 at the outlet are a COD meter and a turbidity meter, respectively, detecting the concentration of organic matter and the concentration of fine particles. A solenoid valve is installed in the solid discharge pipeline.

[0045] The anode generates highly oxidizing free radicals to oxidize and degrade organic matter, while the cathode generates alkaline ions such as OH- to remove hardness ions. The external electric field applied by the pulsed power supply promotes the crystallization of sodium chloride into larger particles. The tilted electrode plates enable efficient solid-liquid separation, separating fine sodium chloride crystals entrained in the liquid and preventing them from entering potassium chloride and affecting product quality. The fine crystals are returned to the evaporation crystallization system as seed crystals to further promote the precipitation and enlargement of sodium chloride crystals. In addition, the crystals in the wastewater can flush and clean the electrode plates, effectively alleviating surface contamination and the problem of scale adhering to the cathode plate surface, thus extending the cleaning cycle of the electrode plates. The electrochemical process generates heat, which helps to keep the mother liquor warm and reduce evaporation energy consumption. The system design fully utilizes the Joule heat generated during the electrochemical process to compensate for heat loss during mother liquor transportation and treatment, thereby reducing the external energy consumption required to maintain the system temperature and achieving internal energy recovery and utilization.

[0046] The intelligent controller 6 adopts a two-layer intelligent optimization architecture, which specifically includes: The first layer is a real-time dynamic optimization layer. During a single processing step, it fine-tunes parameters based on real-time feedback. It employs a model predictive control algorithm to monitor parameters in real time. , and As input, pulse parameters are rapidly optimized and adjusted at the minute or second level; The second layer is the batch learning layer. After a complete batch is processed, deep learning and policy updates are performed. Its reward function, Reward, is as follows: , Among them, α, β, γ, , These are dynamically configurable weighting coefficients, dynamically adjusted by a higher-level meta-controller based on the phased macro-level objectives of the system's operation. This represents the amount of COD degraded per unit time. The amount of COD removed per unit of energy consumption. This is a penalty term, representing the ratio of real-time turbidity to a set value. This is a penalty term for oxygen evolution potential. This is the initial oxygen evolution potential. This represents the change in oxygen evolution potential. An increase indicates that the anodic catalytic oxygen evolution reaction has become difficult, and the coating may be deactivated or the surface may be covered by inactive substances. This is a penalty term for the electrode active area. This represents the change in differential capacitance. As the initial differential capacitance decreases, the effective active area of ​​the surface electrode also decreases. After a complete batch of processing is completed, this layer uses the data from this operation to train the strategy network and the value evaluation network, and updates the control strategy. The reward function is a weighted sum of COD removal rate, processing energy consumption, turbidity and electrode loss. The pulse parameter control strategy is then iteratively optimized and learned offline. The batch learning layer also includes an expert knowledge base for storing the trained policy network parameters; the deep reinforcement learning algorithm is a deep deterministic policy gradient algorithm.

[0047] The intelligent controller 6 is connected to the pulse power supply 43, the first sensor 44, and the second sensor 45, and performs dual-loop coordinated control, including: (a) Real-time COD value acquired based on the first sensor 44 and its rate of change And electrode parameters, to calculate the duty cycle of the first target pulse. and the first target pulse frequency And calculate the target peak current density. ; First target pulse duty cycle The algorithm, which includes current efficiency compensation and cleaning compensation terms, is used to determine the following: , in, The preset base duty cycle is defined by η, which is an adaptively adjustable weighting coefficient ranging from 1.0 to 2.0. and The gain coefficient is adaptively adjustable. For real-time current efficiency, For reference current efficiency, This is the cleaning compensation amount triggered based on electrode state; First target pulse frequency The algorithm, which includes a current efficiency compensation term, a removal rate feedback term, and a cleaning compensation term, is used to determine the specific terms: , in, Based on the base frequency, For real-time current efficiency, For reference current efficiency, This represents the real-time COD change rate. The target COD removal rate, and The gain coefficient is adaptively adjustable. Cleaning compensation frequency; Intelligent controller according to With the set target value deviation and To determine whether the system has entered a high-load impact state, a diffusion control state, or a steady-state optimization state; When a high-load impact condition is detected, the intelligent controller 6 synchronously increases the target peak current density. and When the state is determined to be diffusion control, the intelligent controller 6 increases... and reduce ; Among them, the target peak current density Determined using the following feedback-feedforward composite algorithm: , in, To preset the reference current density, This is the real-time COD value. For the target COD value, The initial COD concentration. This represents the real-time COD change rate. The target COD removal rate, and This is an adjustable gain coefficient.

[0048] (b) Real-time turbidity value acquired based on the second sensor 45 and its rate of change Calculate the duty cycle of the second target pulse. Second target pulse frequency ; Among them, the duty cycle of the second target pulse Determined using the following algorithm: , in, Based on duty cycle, The gain coefficient is adaptively adjustable. The coupling coefficient is... To compensate for the duty cycle during cleaning; Second target pulse frequency Determined using the following algorithm, which is nonlinearly related to turbidity deviation: , in, The base frequency, ΔNTU represents the real-time turbidity value and the set turbidity value. deviation, The gain coefficient is adaptively adjustable. This is the cleaning compensation frequency.

[0049] (c) Based on the preset collaborative decision-making rules, determine the final pulse parameters to be executed, wherein the final pulse duty cycle is... Pick and The smaller value in, i.e. Final pulse frequency Pick and The smaller value in, i.e. ; (d) Target peak current density With a dynamically calculated upper limit of safe current density The smaller value after comparison is taken as the peak current density for controlling the pulse power supply 43. The intelligent controller 6 dynamically calculates the safe upper limit of current density based on real-time water quality parameters. , Based on limiting diffusion current density and upper limit current density of electrode material potential window The smaller value in is determined, that is ; Among them, the limiting diffusion current density n is the average number of electrons transferred during the complete oxidation of an organic molecule, which can be determined experimentally or from literature and is within a range of 4–6; F is the Faraday constant, which is 96485 C / mol. The mass transfer coefficient of organic matter is related to the fluid velocity, electrode plate spacing, temperature, etc. in the reactor and can be estimated by the Sherwood number correlation. This refers to the organic matter concentration, which is the molar concentration of organic matter converted from the real-time monitored COD value.

[0050] (e) according to , The peak current density determined in step (d) controls the operation of the pulse power supply.

[0051] Taking a smaller value for the duty cycle is a conservative strategy that ensures good performance. Larger sizes mean greater oxidation intensity, corresponding to higher energy consumption. A large value may be detrimental to fine-grained sedimentation. Excessive stirring or bubbles can also hinder sedimentation. Taking a small value ensures that basic oxidation requirements are met while prioritizing relatively mild hydrodynamic conditions that promote sedimentation.

[0052] For the frequency, a smaller value was chosen to avoid unnecessary energy consumption and excessive disturbance to the electrode interface that may result from excessively high frequencies. Under the premise of meeting control requirements, a more energy-efficient and stable parameter was selected.

[0053] The fly ash washing water evaporation and salt separation system of this embodiment also includes an electrode status monitoring unit for monitoring the tank voltage at both ends of the electrode plate group 42; the intelligent controller 6 executes an adaptive electrode cleaning program, including: When the rise in the slot voltage exceeds the first preset threshold, the control pulse power supply performs a polarity reversal operation. After the polarity reversal operation, if the rate of decrease of the cell voltage is lower than the second preset threshold, the pulse power supply is controlled to switch to the descaling pulse mode. The frequency of the descaling pulse mode is 500 Hz to 5000 Hz, and the duty cycle is 1% to 10%.

[0054] The intelligent controller 6 also performs the following safety interlock protections: When the liquid level in the intelligent electrochemical purification unit 4 is lower than the safe liquid level threshold, the output of the pulse power supply is cut off. When the bottom solid material level is higher than the discharge material level threshold, the discharge valve will automatically open. Based on the start / stop signal from the evaporation and crystallization unit 2, the output of the pulse power supply 43 is enabled or cut off accordingly.

[0055] Example 2 like Figure 3 As shown, the salt separation method of the fly ash wash water evaporation and salt separation system in Example 1 includes the following steps: Step 1: Evaporate and crystallize the fly ash washing liquid, and then separate solid sodium chloride and the first mother liquor through sodium chloride separation system 3; Step 2: The first mother liquor is introduced into the intelligent electrochemical purification unit 4. While applying a pulsed electric field for electrochemical treatment, dual-loop coordinated control is executed, including: (1) Real-time monitoring of electrode parameters and chemical oxygen demand (COD) values ​​during the treatment process. and its rate of change and effluent turbidity value ; (2) Based on and Calculate the duty cycle of the first target pulse and the first target pulse frequency And calculate the target peak current density. ; (3) Based on Calculate the duty cycle of the second target pulse Second target pulse frequency ; (4) The duty cycle of the final executed pulse Determined as and The smaller value in the value will determine the final pulse frequency. Determined as and The smaller value in; (5) Target peak current density With a dynamically calculated upper limit of safe current density The smaller value after comparison is used as the peak current density for controlling the pulse power supply. (6) Based on , and (5) the parameters of the peak current density adjustment pulse electric field; Step 3: Cool the supernatant after step 2 to crystallize and separate solid potassium chloride.

[0056] Step two also includes: real-time monitoring of the cell voltage of the electrodes in the electrochemical purification unit; when the cell voltage rises abnormally, automatically executing an adaptive cleaning program that includes polarity reversal and / or a high-frequency, low-duty-cycle descaling pulse mode.

[0057] Scaling leads to decreased conductivity, causing the tank voltage to rise slowly under constant current. A threshold value ΔU is set as the signal that cleaning is required. The pulse power supply 43 monitors the voltage across the plates in real time and transmits it to the controller. When the voltage reaches the ΔU threshold, feeding and discharging are stopped, and the system performs a polarity reversal operation. The controller reverses the output polarity of the pulse power supply 43 and applies a set of gradually increasing pulses to the new plates. The plates are then self-cleaned by generating hydrogen ions through electrolysis.

[0058] If the voltage drop rate is lower than the expected threshold, it switches to a high-frequency, low-duty-cycle strong pulse. The large bubbles generated instantly violently impact the surface of the electrode, causing the already formed loose scale layer to crack and loosen.

[0059] Current density at the reverse pole The value of β ranges from 0.5 to 1.2. In descaling mode, the frequency is 500Hz to 1000Hz and the duty cycle is 1% to 10%.

[0060] The reactor shell 41 is equipped with a liquid level detection signal connected to the intelligent controller 6. When the system is first started and no material is fed, if the liquid level in the reactor shell 41 is lower than the threshold, the controller will stop any power output. During operation, if the material is discharged from the bottom of the collection hopper 46 and the material is not replenished in time at the inlet, the liquid level will drop. When the liquid level is lower than the threshold, the intelligent controller 6 will stop any power output to protect the electrode plate.

[0061] A level gauge is installed at the bottom of the reactor shell 41 to detect the height of the sodium chloride crystal slurry and connects the level signal to the intelligent controller 6. When the solid accumulated in the collection tank below the electrode plate (sedimentation and separation of fine sodium chloride crystals) reaches a certain height and exceeds the set threshold of the level signal, the controller starts the solenoid valve to discharge the material into the pipeline below. No manual control is required, which improves efficiency and reduces labor intensity.

[0062] If the current evaporation and crystallization process stops due to various reasons, it sends a stop operation signal to the intelligent controller 6. The intelligent controller 6 then stops all power output to protect the electrode plates. When the evaporation and crystallization process returns to normal, it sends a start-up signal to the intelligent controller 6. Through the communication signals between processes, the equipment can be well protected, ensuring the stable and safe operation of the entire system.

[0063] Experimental Example 1 With all other conditions (such as influent water quality, flow rate, and electrical parameters) kept exactly the same, the electrode plate tilt angles were set to 6°, 55°, 70°, and 80° respectively, and the system was operated.

[0064] Monitoring indicators: 1) Effluent turbidity; 2) Solid settling rate at the bottom of hopper 46; 3) Crystal adhesion / scouring observed on the electrode plate surface.

[0065] When the angle is 6°, the settling path is short, the separation is incomplete, and the effluent turbidity is high. When the angle is 55°-70°, the settling efficiency and turbidity removal rate reach the best balance. When the angle is 80°, although the settling path is long, the effective settling area is reduced, and the scouring effect of the crystals on the plate surface is weakened, which may lead to local siltation. Therefore, 50° to 70° is the optimal range that balances separation efficiency and self-cleaning function.

[0066] Experimental Example 2 With the full dual-layer intelligent optimization architecture enabled, it ran for 30 days; using only single-loop PID control based on fixed rules, and adjusting the current only according to COD feedback, it ran for 30 days.

[0067] Comparison indicators: 1) Electricity consumption per ton of water; 2) Average purity of potassium chloride product; 3) Number of abnormal shutdowns due to scaling or parameter imbalance during the period; 4) Average COD removal rate.

[0068] The complete dual-layer intelligent optimization architecture is significantly superior to single-loop PID control in terms of power consumption per ton of water, product purity stability, and system fault-free operation time. Moreover, the energy consumption of the complete dual-layer intelligent optimization architecture will show a continuous optimization trend as the operating time increases, while the energy consumption of single-loop PID control is basically fixed.

[0069] Experiment Example 3 The key is to ensure efficiency and safety, rather than setting a fixed upper limit. To verify this setting, two scenarios are set up: Scenario 1: Simulating a sudden increase in influent COD concentration while maintaining a constant flow velocity, dynamically calculating the limiting diffusion current density. The current will increase within a safe range to cope with the impact and quickly restore water quality. Scenario 2: Simulating a decrease in influent flow velocity, at which point the limiting diffusion current density... The current density decreases, even if the COD concentration remains unchanged, according to the dynamically calculated safe upper limit. It will also reduce the system's automatic current limiting, avoiding oxygen evolution side reactions and energy waste caused by using high current under low mass transfer conditions.

[0070] Compared to a system that only sets a fixed high current limit, in scenario 2, the latter will generate a large number of invalid bubbles, resulting in a sharp increase in energy consumption and accelerated electrode wear.

[0071] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A fly ash washing water evaporation and salt separation system, characterized in that, include: Evaporation and crystallization unit (2) is used for evaporation and crystallization treatment of fly ash washing liquid; Sodium chloride separation unit (3) is used to separate sodium chloride and the first mother liquor; The intelligent electrochemical purification unit (4) has its inlet connected to the sodium chloride separation unit (3) to receive the first mother liquor. The intelligent electrochemical purification unit (4) includes a reactor shell (41), an inclined electrode plate group (42), a pulse power supply (43), a first sensor (44) for monitoring chemical oxygen demand, and a second sensor (45) for monitoring effluent turbidity. The first mother liquor achieves the removal of COD and hardness ions in the intelligent electrochemical purification unit (4). The external electric field applied by the pulse power supply (43) promotes the crystallization of sodium chloride to precipitate larger particles. A collection hopper (46) is provided at the lower part of the inclined electrode plate group (42). The solid outlet of the collection hopper (46) is connected to the inlet of the evaporation crystallization unit (2) through a pipeline to reflux the sodium chloride crystals separated in the electrochemical purification unit (4) as seed crystals. Potassium chloride crystallization unit (5) is connected to the supernatant outlet of the intelligent electrochemical purification unit; The intelligent controller (6) is connected to the pulse power supply (43), the first sensor (44), and the second sensor (45) and performs dual-loop coordinated control, including: (a) Real-time COD value obtained based on the first sensor (44) and its rate of change And electrode parameters, to calculate the duty cycle of the first target pulse. and the first target pulse frequency And calculate the target peak current density. ; (b) Real-time turbidity value obtained based on the second sensor (45) and its rate of change Calculate the duty cycle of the second target pulse. Second target pulse frequency ; (c) Based on the preset collaborative decision-making rules, determine the final pulse parameters to be executed, wherein the final pulse duty cycle is... Pick and The smaller value in the middle, the final pulse frequency Pick and The smaller value in; (d) Target peak current density With a dynamically calculated upper limit of safe current density The smaller value after comparison is used as the peak current density for controlling the pulse power supply (43) to operate; (e) according to , The peak current density determined in step (d) controls the operation of the pulse power supply.

2. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The target peak current density Determined using the following feedback-feedforward composite algorithm: ， in, To preset the reference current density, This is the real-time COD value. For the target COD value, The initial COD concentration. This represents the real-time COD change rate. The target COD removal rate, and This is an adaptively adjustable gain coefficient.

3. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, In step (a), the duty cycle of the first target pulse The algorithm, which includes a nonlinear concentration feedback term, a current efficiency compensation term, and a cleaning compensation term, is used to determine the specific terms: ， in, The preset base duty cycle is η, which is an adaptively adjustable weighting coefficient with a value range of 1.0 to 2.

0. and The gain coefficient is adaptively adjustable. For real-time current efficiency, This serves as a reference value for the expected or historically best current efficiency. This is a cleaning compensation amount triggered based on the electrode state. When it is determined that the electrode needs to be cleaned, a negative compensation is added.

4. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, In step (a), the first target pulse frequency The algorithm, which includes a nonlinear concentration feedback term, a removal rate feedback term, and a cleaning compensation term, is used to determine the specific details: ， in, Based on the base frequency, For real-time current efficiency, For reference current efficiency, This represents the real-time COD change rate. The target COD removal rate, and The gain coefficient is adaptively adjustable. For cleaning compensation frequency, positive compensation is superimposed when it is determined that the electrode needs to be cleaned.

5. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, In step (b), the second target pulse frequency Determined using the following algorithm, which is nonlinearly related to turbidity deviation: ， in, The base frequency, ΔNTU represents the real-time turbidity value and the set turbidity value. deviation, The gain coefficient is adaptively adjustable. For cleaning compensation frequency, positive compensation is superimposed when it is determined that the electrode needs to be cleaned.

6. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The second target pulse duty cycle Determined using the following algorithm: ， in, Based on duty cycle, The gain coefficient is adaptively adjustable. The coupling coefficient is... To compensate for the duty cycle during cleaning, negative compensation is added when it is determined that the electrode needs cleaning.

7. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The upper limit of safe current density Determined using the following algorithm: Real-time current density limit The minimum values ​​of the limiting diffusion current density of the electrode and the upper limit current density of the potential window are taken. , Among them, the upper limit current density of the potential window Determined by the electrode material itself, The limiting diffusion current density of the electrode is , ， The average number of electrons transferred during the complete oxidation of organic molecules. for Faraday constant, The mass transfer coefficient of organic matter is... This represents the concentration of organic matter in the bulk.

8. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The intelligent controller (6) adopts a two-layer intelligent optimization architecture, specifically including: The first layer is a real-time dynamic optimization layer. During a single processing step, it fine-tunes parameters based on real-time feedback. It employs a model predictive control algorithm to monitor parameters in real time. , and As input, pulse parameters are rapidly optimized and adjusted at the minute or second level; The second layer is the batch learning layer. After a complete batch is processed, deep learning and policy updates are performed. Its reward function, Reward, is as follows: ， Among them, α, β, γ, , These are dynamically configurable weighting coefficients, dynamically adjusted by a higher-level meta-controller based on the phased macro-level objectives of the system's operation. This represents the amount of COD degraded per unit time. The amount of COD removed per unit of energy consumption. This is a penalty term, representing the ratio of real-time turbidity to a set value. This is a penalty term for oxygen evolution potential. This is the initial oxygen evolution potential. This represents the change in oxygen evolution potential. An increase indicates that the anodic catalytic oxygen evolution reaction has become difficult, and the coating may be deactivated or the surface may be covered by inactive substances. This is a penalty term for the electrode active area. This represents the change in differential capacitance. As the initial differential capacitance decreases, the effective active area of ​​the surface electrode also decreases. After a complete batch of processing is completed, this layer uses the data from this operation to train the strategy network and the value evaluation network, and updates the control strategy. The reward function is a weighted sum of COD removal rate, processing energy consumption, turbidity and electrode loss. The pulse parameter control strategy is then iteratively optimized and learned offline. The batch learning layer also includes an expert knowledge base for storing the trained policy network parameters; the deep reinforcement learning algorithm is a deep deterministic policy gradient algorithm.

9. The fly ash washing water evaporation and salt separation system according to claim 8, characterized in that, It also includes an electrode condition monitoring unit for monitoring the slot voltage at both ends of the electrode plate assembly; The intelligent controller (6) executes an adaptive electrode cleaning procedure, including: When the rise in the slot voltage exceeds the first preset threshold, the control pulse power supply (43) performs an output polarity reversal operation. After the polarity reversal operation, if the rate of decrease of the cell voltage is lower than the second preset threshold, the pulse power supply (43) is controlled to switch to the descaling pulse mode. The frequency of the descaling pulse mode is 500 Hz to 5000 Hz and the duty cycle is 1% to 10%.

10. The fly ash washing water evaporation and salt separation system according to claim 9, characterized in that, During the polarity reversal operation, the intelligent controller (6) controls the current density applied by the pulse power supply (43). With normal operating current density satisfy: , where β ranges from 0.5 to 1.

2.

11. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The intelligent controller (6) performs the following safety interlock protection: When the liquid level in the intelligent electrochemical purification unit (4) is lower than the safe liquid level threshold, the output of the pulse power supply (43) is cut off; When the bottom solid material level is higher than the discharge material level threshold, the discharge valve will automatically open. Based on the start / stop signal from the evaporation and crystallization unit (2), the output of the pulse power supply (43) is enabled or cut off accordingly.

12. The fly ash washing water evaporation and salt separation system according to claim 1, characterized in that, The tilted electrode plate group (42) includes alternating anode plates and cathode plates, with the tilt angle of the anode plates and cathode plates being 50° to 70°.

13. A method for salt separation in a fly ash washing water evaporation and salt separation system as described in any one of claims 1 to 12, characterized in that, Including the following steps: S1. Evaporate and crystallize the fly ash washing liquid (2), and then separate solid sodium chloride and the first mother liquor through the sodium chloride separation system (3); S2. The first mother liquor is introduced into the intelligent electrochemical purification unit (4). While applying a pulsed electric field for electrochemical treatment, dual-loop coordinated control is executed, including: S21. Real-time monitoring of electrode parameters and chemical oxygen demand (COD) values ​​during the treatment process. and its rate of change and effluent turbidity value ; S22, based on and Calculate the duty cycle of the first target pulse and the first target pulse frequency And calculate the target peak current density. ; S23, based on Calculate the duty cycle of the second target pulse Second target pulse frequency ; S24. The duty cycle of the final executed pulse. Determined as and The smaller value in the value will determine the final pulse frequency. Determined as and The smaller value in; S25, Target peak current density With a dynamically calculated upper limit of safe current density The values ​​are compared, and the smaller value is taken as the peak current density for controlling the pulse power supply. S26, according to , And the parameters of the pulse electric field determined by peak current density in S25; S3. Cool the supernatant after S2 treatment to crystallize and separate solid potassium chloride.

14. The salt separation method of the fly ash washing water evaporation salt separation system according to claim 13, characterized in that, S2 also includes: Real-time monitoring of the cell voltage of the internal electrode of the electrochemical purification unit (4); when the cell voltage rises abnormally, an adaptive cleaning program is automatically executed, including polarity reversal operation and / or high-frequency, low-duty-cycle descaling pulse mode.

15. The salt separation method of the fly ash washing water evaporation and salt separation system according to claim 13, characterized in that, In S2, sodium chloride crystals separated from the bottom of the electrochemical purification unit (4) are returned to the evaporation and crystallization step of S1 as seed crystals.

Images