Wastewater and dirty salt treatment process optimization method and system based on digital twinning

Abstract

The invention discloses a digital twinning-based wastewater and dirty salt treatment process optimization method and system, and relates to the technical field of intelligent wastewater and dirty salt treatment. The method comprises the steps that a digital twinborn model synchronously mapped with a processing system is established, and the digital twinborn model integrates the technological parameter incidence relation of an evaporation section, a crystallization section and a filtering section; collecting component data and flow data of the polluted salt wastewater corresponding to each process section in real time; correcting model characteristic parameters of each process section in the digital twin model based on the component data and the flow data collected in real time; according to the corrected digital twin model and the optimization target, generating a process optimization strategy corresponding to each process section; and monitoring the component data of the treated discharge water, reuse water, industrial salt and solid waste in real time, and if the data does not reach the standard, triggering model parameter correction, optimization strategy iteration or cross-section collaborative logic recalculation. According to the invention, intelligent cooperation of wastewater and dirty salt treatment is realized.

Description

Technical Field

[0001] This application relates to the technical field of intelligent wastewater and sludge treatment. Specifically, it relates to a method and system for optimizing wastewater and sludge treatment processes based on digital twins. Background Art

[0002] Wastewater and sludge treatment is a process of separating salts (such as inorganic salts like sodium chloride and sodium sulfate) from wastewater containing high concentrations of salts and pollutants generated in industrial production, chemical engineering, energy and other fields, and performing harmless or resource-based treatment on the separated sludge.

[0003] The role of wastewater and sludge treatment is to separate salts from high-salt wastewater, which can not only avoid environmental problems such as soil salinization and water body ecological damage caused by direct discharge, but also reduce the corrosion damage of salts to industrial equipment. It can also purify and recycle the utilizable inorganic salts in the sludge and reuse them as industrial raw materials. At the same time, it eliminates the interference of salts for subsequent in-depth treatment of wastewater (such as reuse and biochemical treatment), achieving synergy benefits in multiple aspects such as environmental protection, production and resource utilization.

[0004] In existing wastewater and sludge treatment processes, it is usually dependent on a single sensor (such as a pH sensor) to regularly collect data of sludge wastewater before treatment, and then statically adjust the treatment process parameters based on manual experience. This method has the defects of single detection dimension and lagged adjustment, and it is difficult to respond to sudden changes in water quality, such as sudden high salt loads, resulting in a decline in treatment efficiency.

[0005] In response to the above problems, no effective solutions have been proposed yet. Summary of the Invention

[0006] An embodiment of this application provides a method and system for optimizing wastewater and sludge treatment processes based on digital twins to solve the above technical problems.

[0007] According to one aspect of the embodiments of this application, a method for optimizing wastewater and sludge treatment processes based on digital twins is provided, including: establishing a digital twin model that is synchronously mapped with the treatment system, where the digital twin model integrates the correlation relationships of process parameters in the evaporation section, crystallization section, and filtration section; through component analyzers and monitoring sensors deployed in each process section, real-time collecting the component data and flow data of the sludge wastewater corresponding to each process section; based on the real-time collected component data and flow data, correcting the model characteristic parameters of each process section in the digital twin model; according to the corrected digital twin model and the optimization target, generating process optimization strategies corresponding to each process section for adjusting the treatment system; real-time monitoring the component data of the treated discharged water, recycled water, industrial salt, and solid waste. If not up to standard, trigger the correction of model parameters, iteration of optimization strategies, or recalculation of cross-section collaborative logic.

[0008] According to another aspect of the embodiments of the present application, there is provided an optimization system for wastewater and sludge treatment process based on digital twin, including: a virtual model establishment module, configured to establish a digital twin model that is synchronously mapped to the treatment system, and the digital twin model integrates the correlation relationships of process parameters of the evaporation section, the crystallization section, and the filtration section; a real-time data acquisition module, configured to collect in real time the composition data and flow data of the sludge wastewater corresponding to each process section through the composition analyzers and monitoring sensors deployed in each process section; a model correction module, configured to correct the model characteristic parameters of each process section in the digital twin model based on the composition data and flow data collected in real time; an optimization strategy generation module, configured to generate process optimization strategies corresponding to each process section according to the corrected digital twin model and the optimization target, so as to adjust the treatment system; a closed-loop feedback module, configured to monitor in real time the composition data of the treated discharged water, recycled water, industrial salt, and solid waste, and if not up to standard, trigger the correction of model parameters, the iteration of optimization strategies, or the recalculation of cross-section collaborative logic.

[0009] According to yet another aspect of the embodiments of the present application, there is provided an electronic device, including a memory and a processor, where a computer program is stored in the memory, and the processor is configured to execute the above-mentioned optimization method for wastewater and sludge treatment process based on digital twin through the computer program.

[0010] Based on the embodiments provided in the present application, by constructing a system-level dynamic model through digital twin technology, combining multi-dimensional real-time data-driven and closed-loop optimization mechanisms, the transformation of wastewater and sludge treatment from manual experience-driven to intelligent collaboration has been realized, effectively solving the core problems of single detection and lagging adjustment in the prior art, and having remarkable progress in terms of the working condition adaptability, optimization accuracy, and long-term stability of the treatment system. Specifically: The present application breaks through the fragmented monitoring mode that relies on a single sensor in the prior art by establishing a digital twin model that is synchronously mapped to the treatment system and integrating the correlation relationships of process parameters of the evaporation section, the crystallization section, and the filtration section. In the scenario of wastewater and sludge treatment, the digital twin model can map the chain reactions between each process section in real time. The composition data and flow data of the sludge wastewater are collected in real time by using the composition analyzers and monitoring sensors. When a sudden high salt load or a sudden change in water quality composition occurs, the real-time data can directly drive the digital twin model to correct the characteristic parameters of each process section, enabling the system to timely complete the identification of abnormal working conditions and the calibration of model parameters, significantly improving the adaptability of the treatment system to water quality fluctuations, and avoiding the decline in treatment efficiency or abnormal equipment operation caused by detection lag in the prior art.

[0011] Based on the revised digital twin model and optimization target, a process optimization strategy is generated to form a complete closed loop from data collection to model correction to strategy execution to effect feedback. It changes the inefficient adjustment mode of the existing technology that relies on manual trial and error, while ensuring that the quality of discharged water and recycled water meets the standards, it improves the efficiency of waste salt resource utilization. The digital twin model integrates the correlation between the three-stage process parameters so that the optimization strategy can take into account the synergistic effect of each stage. This cross-stage collaborative optimization capability solves the problems of poor system stability and high energy consumption caused by the independent adjustment of each process stage in the existing technology, and achieves a balance between multiple objectives such as efficiency improvement, equipment protection, and cost control in wastewater and waste salt treatment. It is especially suitable for long-term stable operation under complex conditions in industrial high-salt wastewater treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 It is a flowchart of an optional method for optimizing a wastewater and salt treatment process based on digital twins according to an embodiment of the present application; Figure 2 is an optional interaction diagram of a digital twin model and a processing system according to an embodiment of the present application; Figure 3 It is a flowchart of another optional wastewater and salt treatment process optimization method based on digital twin according to an embodiment of the present application; Figure 4 This is a structural diagram of an optional digital twin-based wastewater and salt treatment process optimization system according to an embodiment of the present application.

[0013] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0014] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0015] It should be noted that the terms "first", "second", etc. in the description, claims and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0016] The applicant proposes an optimization method and system for wastewater and sludge treatment process based on digital twin. By constructing a digital twin model covering the coupling relationship of multiple process sections, real-time fusion of water quality and equipment status data is used to dynamically correct model parameters, generate an adaptive adjustment strategy and conduct closed-loop verification, so as to realize the intelligent optimization of the treatment process under complex working conditions.

[0017] According to one aspect of the embodiments of the present application, there is provided an optimization method for wastewater and sludge treatment process based on digital twin, as Figure 1 shown. The method includes: S101, establishing a digital twin model that is synchronously mapped with the treatment system, and the digital twin model integrates the process parameter correlation relationships of the evaporation section, the crystallization section, and the filtration section; In this embodiment, the treatment system is used to receive the input sludge wastewater, and after treating it, output four items: discharged water, recycled water, industrial salt, and solid waste; component analyzers and monitoring sensors are deployed at some positions of the physical devices corresponding to each process section in the treatment system; Optionally, the monitoring sensors include but are not limited to temperature sensors, pressure sensors, flow sensors, liquid level sensors, conductivity sensors, pH sensors, turbidity sensors, stirring speed sensors, membrane pressure difference sensors, particle size sensors, flow velocity sensors, vacuum sensors, and vibration sensors.

[0018] Optionally, the component analyzers include but are not limited to ion chromatographs, salinometers, total dissolved solid analyzers, potentiometric titrators, ultraviolet-visible spectrometers, infrared spectrometers, mass spectrometers, gas chromatographs, liquid chromatographs, total organic carbon analyzers, industrial salt purity analyzers, particle size analyzers, solid waste moisture content detectors, recycled water pollutant component analyzers, and on-line multi-parameter water quality analyzers, etc.; S102, through the component analyzers and monitoring sensors deployed in each process section, real-time collecting the component data and flow data of the sludge wastewater corresponding to each process section; S103. Based on the composition data and flow data collected in real time, correct the model characteristic parameters of each process section in the digital twin model; In this step, update the parameter mapping relationship of the digital twin model based on the composition data and flow data in the processing system to achieve the state consistency calibration between the digital twin model and the processing system; S104. According to the corrected digital twin model and the optimization goal, generate process optimization strategies corresponding to each process section for adjusting the processing system; S105. Real-time monitor the composition data of the treated discharged water, recycled water, industrial salt, and solid waste. If not up to standard, trigger the correction of model parameters, the iteration of optimization strategies, or the recalculation of cross-section collaborative logic.

[0019] Exemplarily, as Figure 2 shown, it is an interaction diagram between a digital twin model and a processing system. In this diagram, the processing system includes physical devices corresponding to each process section, namely evaporation section devices, crystallization section devices, and filtration section devices, which constitute the physical carrier for treating sewage salt wastewater. The digital twin model realizes the full-process mirroring of the correlation relationship of process parameters in the evaporation section, crystallization section, and filtration section by establishing a synchronous mapping relationship with the processing system.

[0020] Among them, the evaporation section devices include, but are not limited to, evaporators, preheaters, condensers, and vacuum pumps, etc., which are used to realize the evaporation and concentration of wastewater; the crystallization section devices include, but are not limited to, crystallizers, stirrers, coolers, and seed addition devices, etc., which are used to control the crystallization and precipitation of salts; the filtration section devices include, but are not limited to, filters, backwashing devices, and filtrate collection tanks, etc., which are used to complete the solid-liquid separation of crystalline salts and mother liquor. The three types of devices together constitute the physical carrier for treating sewage salt wastewater.

[0021] The processing system transmits the composition data and flow data of each process section to the digital twin model in real time, providing a measured basis for model correction. After the digital twin model corrects the model characteristic parameters of each process section (such as mass transfer efficiency coefficient, crystallization activation energy threshold) based on the real-time data, it generates process optimization strategies that meet the optimization goal and outputs control instructions (such as evaporation section temperature adjustment, crystallization section stirring power adjustment) to the processing system to realize the dynamic adjustment of the physical system.

[0022] That is to say, through the two-way interaction between the digital twin model and the processing system, a closed loop from data collection to model optimization and then to strategy execution is constructed, realizing the optimization of the wastewater and sewage salt treatment process and solving the problem of insufficient coordination in single-section process control optimization.

[0023] As a specific implementation manner, collect the composition data and flow data of the sewage salt wastewater corresponding to each process section, including: Based on multiple groups of salt concentration gradient sensor arrays longitudinally arranged in the evaporation section, a real-time salt concentration distribution heat map is generated, and the real-time salt concentration distribution heat map reflects the changing trend of the longitudinal salt concentration gradient in the evaporator; Based on an ultrasonic supersaturation detector installed in the mother liquor circulation pipeline in the crystallization section, combined with the data of the turbulence intensity sensor, the crystal nucleation critical point offset is calculated, and the crystal nucleation critical point offset is used to characterize the dynamic stability of the mother liquor in the crystallization section; Based on the online particle size analyzers set at the inlet and outlet of the filtration section, the median particle size and distribution uniformity coefficient of the intercepted salt particles are synchronously collected, and the distribution uniformity coefficient is used to evaluate the degree of filtration efficiency degradation; Through the steam flow velocity sensor in the evaporation section and the fluid vortex detector in the crystallization section, the steam flow turbulence intensity and the mother liquor circulation vortex frequency are obtained in real time; Deploy a micro-pressure difference array sensor on the membrane surface in the filtration section to generate a membrane surface flow velocity distribution cloud map. The membrane surface flow velocity distribution cloud map is used to correlate the sudden change in pressure difference with the salt particle interception mode.

[0024] Based on the embodiments provided in this application, a system-level dynamic model is constructed through digital twin technology. Combining multi-dimensional real-time data-driven and closed-loop optimization mechanisms, it realizes the leap from manual experience-driven to intelligent collaboration in wastewater and salt treatment, effectively solving the core problems of single detection and lagging adjustment in the prior art, and making significant progress in the working condition adaptability, optimization accuracy, and long-term stability of the treatment system. Specifically: This application breaks through the fragmented monitoring mode that relies on a single sensor in the prior art by establishing a digital twin model that synchronously maps with the treatment system and integrating the correlation relationships of process parameters in the evaporation, crystallization, and filtration sections. In the wastewater and salt treatment scenario, the digital twin model can real-time map the chain reactions between each process section. For example, when the high-salt wastewater causes the pressure difference in the filtration section to increase, the digital twin model can, based on the preset correlation relationships of process parameters, synchronously identify the possible uneven feed volume problem in the evaporation section, providing an accurate digital foundation for subsequent cross-section collaborative optimization, thereby solving the defect of insufficient analysis of the complex correlation relationships of the treatment system in the prior art.

[0025] The composition data and flow data of the salt-containing wastewater are collected in real time using a composition analyzer and monitoring sensors. When a sudden high-salt load or water quality composition mutation occurs, the real-time data can directly drive the digital twin model to correct the characteristic parameters of each process section, such as the heat transfer coefficient in the evaporation section and the supersaturation threshold in the crystallization section, enabling timely identification of abnormal working conditions and calibration of model parameters, significantly enhancing the adaptability of the treatment system to water quality fluctuations, and avoiding the reduction in treatment efficiency or abnormal equipment operation caused by detection lag in the prior art.

[0026] Based on the revised digital twin model and optimization target, a process optimization strategy is generated to form a complete closed loop from data collection to model correction to strategy execution to effect feedback. For example, when it is monitored that the purity of industrial salt does not meet the recovery standard, the supersaturation control parameters of the crystallization section or the impurity interception efficiency of the filtration section are traced through the digital twin model, and a cross-section adjustment strategy is automatically generated, such as adjusting the matching relationship between the evaporation section temperature and the crystallization section stirring rate, so that the treatment process can be self-optimized without manual intervention. The inefficient adjustment mode of the existing technology that relies on manual trial and error has been changed, and the efficiency of resource utilization of polluted salt has been improved while ensuring that the quality of discharged water and recycled water meets the standards.

[0027] The integration of the correlation between the three-stage process parameters by the digital twin model enables the optimization strategy to take into account the synergistic effects of each stage. For example, when the performance of the membrane assembly in the filtration stage degrades, the risk of decreased feed stability in the evaporation stage is foreseen through the model correlation, and a combination strategy of rotating the membrane group in the filtration stage and compensating the heating power in the evaporation stage is generated in advance, thereby maintaining the processing efficiency while reducing the risk of equipment overload. This cross-stage collaborative optimization capability solves the problems of poor system stability and high energy consumption caused by the independent adjustment of each process stage in the existing technology, and achieves a balance between multiple objectives such as efficiency improvement, equipment protection, and cost control in wastewater and salt treatment. It is especially suitable for long-term stable operation of complex working conditions in industrial high-salt wastewater treatment.

[0028] As an optional solution, when constructing a digital twin model, the equipment status data is incorporated into the model construction elements. The equipment status data includes mechanical performance parameters, physical structure parameters, and operating status parameters. Among them, the mechanical performance parameters include the vibration amplitude of the centrifugal pump and the transmembrane pressure difference of the membrane assembly; the physical structure parameters include the scaling thickness of the evaporator heat exchange tube and the porosity of the nanofiltration membrane; the operating status parameters include the equipment start and stop time and the load rate.

[0029] Optionally, the mechanical performance parameters also include but are not limited to agitator torque value, fan motor speed and pump operating temperature; the physical structure parameters also include but are not limited to pipeline inner wall corrosion thickness and membrane surface roughness; the operating status parameters also include but are not limited to cumulative operating time and fault alarm signals.

[0030] Based on the embodiments provided in this application, when constructing a digital twin model, device status data is incorporated. By fusing three types of data: mechanical properties, physical structure, and operating status, the model can accurately reflect the real-time health status of the device. For example, the degree of membrane fouling can be identified based on the growth trend of the transmembrane pressure difference of the membrane module, and the cut-off efficiency of the filtration section can be dynamically adjusted in combination with the change in the porosity of the nanofiltration membrane, avoiding model inaccuracy caused by device performance degradation. Compared with the defects of "relying on manual inspections for device status and estimating parameter adjustments based on experience" in traditional methods, the digital twin model can adapt to actual working conditions such as device aging, scaling, and wear in real time. For example, when the heat exchange tubes of the evaporator are fouled, the heat transfer coefficient is automatically corrected, and when the vibration of the centrifugal pump is abnormal, the load threshold is reduced in advance, providing a digital foundation that better fits the real operating status of the device for subsequent process optimization strategies, fundamentally enhancing the adaptability of the model to complex engineering scenarios.

[0031] As an alternative solution, based on the composition data and flow data collected in real time, the model characteristic parameters of each process section in the digital twin model are corrected, including: Analyze the deviation between the composition data and flow data collected in real time and the predicted values of the digital twin model; Based on the deviation analysis results, dynamically correct the mass transfer efficiency coefficient, crystallization activation energy threshold, and filter layer degradation rate function in the digital twin model.

[0032] As an alternative implementation, analyzing the deviation between the composition data and flow data collected in real time and the predicted values of the digital twin model includes: Perform time and space calibration on the composition and flow data of the fouled salt wastewater collected in real time, form a mapping with the predicted values of the digital twin model, identify the differences between the measured values of the key parameters of each process section and the model output, and provide a basis for model correction.

[0033] For key indicators such as the salt concentration gradient and heat transfer efficiency in the evaporation section, identify abnormal mass transfer processes by comparing the measured and predicted salt concentration change rates; For the crystallization section, analyze data such as the supersaturation of the mother liquor and the crystal size distribution, and combine parameters such as the stirring frequency and fluid shear force to locate the reasons for the fluctuations in the nucleation rate; For the filtration section, monitor the uniformity of the particle size of the retained salt particles and the flow velocity distribution on the membrane surface, and identify the characteristics of filter layer blockage or degradation through changes in the transmembrane pressure difference.

[0034] Based on the embodiments provided in this application, by analyzing the deviation between real-time data and model prediction values, key parameters are dynamically corrected, solving the problem in the prior art that model parameters are fixed and cannot respond to operating condition fluctuations. For example, when the measured value of the salt concentration in the evaporation section deviates from the model prediction value, the mass transfer efficiency anomaly can be identified and adjusted in a targeted manner to ensure that the salt separation efficiency during the evaporation process is consistent with the model calculation; when the interception effect of the filtration section decreases, the performance attenuation caused by membrane pollution or filter material aging is compensated in real time through the correction of the filter layer degradation rate function. This data-driven deviation feedback to parameter correction mechanism enables the digital twin model to calibrate the core physical characteristics of the process section in real time, avoid the decrease in processing efficiency due to sudden changes in water quality (such as high salt load) or equipment performance drift, and achieve dynamic synchronization between model output and actual process behavior, laying the foundation for accurate optimization strategy generation.

[0035] As an optional solution, based on the deviation analysis results, the mass transfer efficiency coefficient, crystallization activation energy threshold and filter layer degradation rate function in the digital twin model are dynamically corrected, including: According to the deviation between the predicted value and the actual detected value of the salt concentration gradient in the evaporation section, the mass transfer efficiency coefficient is adjusted by an adaptive gradient descent algorithm, wherein the learning rate of the adaptive gradient descent algorithm is positively correlated with the real-time salt load; if the deviation direction is positive, the mass transfer efficiency coefficient is reduced; For example, by dynamically adjusting the learning rate through salt load, when the salt load is high, the learning rate increases to accelerate the correction of the mass transfer efficiency coefficient and suppress the runaway salt concentration gradient; when the salt load is low, the learning rate decreases to avoid oscillation caused by excessive adjustment.

[0036] In this embodiment, the positive deviation refers to the measured value being greater than the model predicted value. For example, the measured salt concentration at the outlet of the evaporation section is 10% higher than the predicted value, indicating that the mass transfer efficiency currently predicted by the model is too high and the actual mass transfer capacity is not as expected. At this time, the mass transfer efficiency coefficient is reduced to make the model output close to the actual working conditions, such as adjusting the coefficient from 0.9 to 0.85, to avoid the calculation deviation of the supersaturation in the subsequent crystallization section due to the overestimation of the evaporation capacity by the model, thereby preventing abnormal crystal nucleation or equipment scaling risks. The model parameters are corrected by negative feedback to ensure the prediction accuracy of the core process parameter "salt concentration gradient in the evaporation section", providing a reliable data basis for cross-section collaborative optimization.

[0037] Based on the supersaturation deviation of the mother liquor in the crystallization stage and the fluid shear force data, the crystallization activation energy threshold is adjusted through the dynamic sliding mode control algorithm to suppress abnormal fluctuations in the crystal nucleation rate; when hidden delay effects are detected, such as the delayed feedback of the pressure difference fluctuation in the filtration stage to the crystallization stage, the time attenuation factor is introduced to dynamically weaken the hysteresis weight; Among them, the fluid shear force data is the shear action parameter generated by the fluid flow in the crystallization section on crystal particles and the vessel wall, which is collected in real time by a pressure sensor or a flow velocity sensor installed in the crystallization kettle, such as the turbulence intensity near the stirring paddle and the friction force between the fluid and the crystal surface. The fluid shear force data is part of the process parameter correlation relationship, used to characterize the stability of the flow field in the crystallization section, and is a key input in the digital twin model that affects the crystal nucleation rate and particle size distribution.

[0038] According to the uniformity deviation of the salt particle size distribution intercepted in the filtration section, multi-scale convolution kernels are used to extract local blockage features and reconstruct the filter layer degradation rate function; if the flow velocity distribution cloud map on the membrane surface shows regional flow velocity anomalies, the porosity parameters of the corresponding area are preferentially corrected.

[0039] For example, if the edge flow velocity drops by 30%, a regional flow velocity anomaly is determined; Based on the embodiments provided in this application, for the characteristics of different process sections, the model characteristic parameters of each process section in the digital twin model are dynamically corrected, significantly improving the parameter correction accuracy under complex working conditions. For example, the control accuracy of the crystal nucleation rate is improved, and the early warning of local blockage of the filter layer is advanced, providing a more reliable model support for cross-section collaborative optimization.

[0040] Specifically, for the evaporation section, based on the salt concentration gradient deviation, an adaptive gradient descent algorithm is used to dynamically adjust the mass transfer efficiency coefficient, which can quickly respond to the change of high-salt wastewater load and avoid the correction lag of the traditional fixed-step algorithm under extreme working conditions. For the crystallization section, combining the mother liquor supersaturation deviation and the fluid shear force data, the dynamic sliding mode control algorithm is used to suppress the abnormal fluctuation of the crystal nucleation rate and ensure the uniformity of crystal particle size, such as preventing crystal fragmentation caused by sudden change of supersaturation. For the filtration section, multi-scale convolution kernels are used to extract local blockage features on the membrane surface and reconstruct the filter layer degradation rate function. Especially for regional flow velocity anomalies, such as local scaling caused by too low flow velocity at the corners of the membrane module, the porosity parameters of the corresponding area are preferentially corrected, solving the problem that traditional global correction cannot accurately locate local faults.

[0041] As an optional solution, the optimization objectives include the component compliance indicators of the treated discharged water, recycled water, industrial salt, and solid waste; As an example, the optimization objectives may include, but are not limited to: the compliance indicators for the discharged water components: the salt concentration ≤ 1000 mg / L (calculated as sodium chloride), and the concentrations of heavy metal ions (such as nickel and copper) ≤ 0.1 mg / L (meeting the standard of GB / T 31962-2015); the compliance indicators for the recycled water components: the conductivity ≤ 500 μS / cm, the turbidity ≤ 5 NTU, meeting the water quality requirements for industrial circulating cooling water (such as no risk of scaling in cooling equipment). The compliance indicators for the industrial salt components: the purity of sodium chloride ≥ 95%, and the sulfate impurity ≤ 1.5%, applicable to the raw material grade recycling standard for the chlor-alkali industry. The compliance indicators for the solid waste components: the moisture content ≤ 60%, and the soluble salt content ≤ 5% (calculated as sodium sulfate), meeting the landfill entry standard (such as preventing leachate from polluting the soil).

[0042] As Figure 3 shown, according to the corrected digital twin model and the optimization objectives, process optimization strategies corresponding to each process section are generated, including: S301, according to the corrected digital twin model, analyze the explicit process correlations and implicit dynamic interaction effects of the evaporation section, crystallization section, and filtration section to construct a coupling weight matrix; S302, deduce the predicted component data of the four under the current process parameters through the corrected digital twin model, and analyze the deviation sources from the optimization objectives; S303, based on the analysis results of the deviation sources, generate a strategy solution space that meets the constraint conditions; In some embodiments of the present application, the constraint conditions include, but are not limited to: the adjustment range of process parameters is restricted by the physical limits of the equipment; the cross-section coupling effects need to comply with the laws of mass and energy conservation; the core indicators of the discharged water and industrial salt are preferentially guaranteed; S304, screen the Pareto optimal strategy combinations from the strategy solution space, and generate the coordinated control instructions for the evaporation section, crystallization section, and filtration section in combination with the cross-section influence intensity of the coupling weight matrix.

[0043] Optionally, after generating the coordinated control instructions, the instructions can be sent to the processing system for execution.

[0044] Based on the embodiments provided in this application, when generating an optimization strategy, explicit or implicit cross-segment associations are analyzed through a coupling weight matrix, and constraint conditions including equipment physical limits, mass and energy conservation, and core indicator prioritization are constructed, solving the problems of target conflicts and system instability caused by single-segment isolated regulation in the prior art, and suppressing systematic oscillations caused by single-segment optimization. For example, when the crystal impurity rate in the crystallization section exceeds the standard, not only the supersaturation in the crystallization section is adjusted, but also the implicit influence of the retention efficiency of the filtration section on the impurity rate is identified through the coupling matrix, and the membrane pore size parameters of the filtration section are synchronously optimized; under high salt load, the strategy solution space preferentially ensures that the salt concentration in the discharged water meets the standard, and at the same time restricts excessive heating in the evaporation section through energy conservation constraints to avoid the risk of equipment scaling. This optimal strategy screening mechanism under multi-objective constraints enables the optimization strategy to not only meet core indicators such as discharged water and industrial salt, but also take into account equipment safety and energy consumption balance, achieving a leap from efficiency optimization to multi-objective coordination, especially suitable for strongly coupled scenarios of efficiency, quality, and safety in wastewater and waste salt treatment.

[0045] As an alternative solution, the dynamic construction of the coupling weight matrix includes: Defining the direct coupling relationship between process sections through the mass conservation equation and the energy balance equation; Training a time series neural network based on historical operation data to identify cross-segment delays or non-linear effects; Optionally, the historical operation data includes historical process parameters, historical collected component data, and flow data; Verifying the effectiveness of explicit weights and implicit weights according to the real-time collected data.

[0046] Based on the embodiments provided in this application, when dynamically constructing the coupling weight matrix, physical equations and data-driven are integrated, which not only retains the rigor of the process mechanism but also explores implicit cross-segment effects, such as the delayed effect of the change in the stirring frequency in the crystallization section on the membrane flux in the filtration section. The digital twin model can not only accurately reflect the direct physical associations between process sections but also reveal the implicit interaction laws accumulated during long-term operation, providing a more comprehensive analysis of influencing factors for cross-segment collaborative strategies, such as predicting in advance the delayed effect of the adjustment in the previous section on the subsequent section, and solving the problem of strategy failure caused by incomplete cross-segment association analysis in the prior art.

[0047] Specifically, explicit weights are defined through physical equations to ensure that cross-segment mass or energy transfer conforms to basic laws, such as the change in the flow rate in the filtration section is directly related to the feed load in the evaporation section; implicit weights are identified by training a time series neural network with historical data to capture non-linear relationships, such as the influence on the pH value in the crystallization section 30 minutes after the temperature fluctuation of the condensate water in the evaporator; the real-time verification mechanism ensures the effectiveness of the weights and avoids the "black box" defect of a pure data-driven model, such as verifying the cross-segment mass conservation relationship through real-time flow data.

[0048] As an alternative solution, the predicted component data of the four under the current process parameters is deduced through the corrected digital twin model, and the deviation sources from the optimization objectives are analyzed, including: using causal inference algorithms to quantify the causal effects of process section parameters on the components of the four, locating the process sections with high contribution degrees, and verifying the authenticity of the deviation root causes through counterfactual analysis to exclude spurious associations.

[0049] In an alternative implementation, through a causal inference algorithm such as Do-Calculus, calculate the causal effect value between the stirring frequency in the crystallization section and the purity of industrial salt. For example, for every 5 Hz increase in the stirring frequency, the probability of purity improvement increases by 25%, to identify the key parameters whose influence on the deviation exceeds a certain proportion. For the association that the abnormal load in the evaporation section is caused by the decrease in the membrane flux in the filtration section, simulate through counterfactual analysis "if the membrane flux does not decrease, whether the load in the evaporation section is normal", exclude misjudgments caused by sensor noise, and ensure that the deviation source is the real interaction between process sections.

[0050] Based on the embodiments provided in this application, causal effect quantification makes the causal relationship between the adjustment of process section parameters and the components of the four explicit. For example, it clarifies the direct influence path of the change in the stirring frequency in the crystallization section on the purity of industrial salt, ensuring the physical authenticity of the deviation source analysis. For example, it can exclude false deviation alarms caused by sensor noise and accurately identify the increase in the impurity rate of industrial salt caused by abnormal supersaturation in the crystallization section; the counterfactual verification mechanism effectively distinguishes the real interaction between process sections and accidental data fluctuations through simulation. For example, verify whether the temporary fluctuation of the membrane flux in the filtration section really causes the abnormal load in the evaporation section, avoid misjudgments caused by accidental data fluctuations, and provide a reliable root cause basis for subsequent strategy generation.

[0051] As an alternative solution, based on the deviation source analysis results, generate a strategy solution space that meets the constraint conditions, including: using a hierarchical reinforcement learning framework to divide the hierarchical decisions of cross-section collaboration and single-section execution, formalize the physical limits of equipment and the law of conservation of energy and mass in combination with a constraint automaton, and explore feasible strategy combinations that meet the core indicators within the constraints.

[0052] In an alternative embodiment, the high-level strategy determines the adjustment priority through causal reasoning. For example, when the purity of industrial salt does not meet the standard, the priority of adjusting the activation energy threshold in the crystallization section is higher than that of adjusting the membrane flux in the filtration section. The low-level strategy sets parameter boundaries based on equipment status data. For example, when the transmembrane pressure difference of the membrane module is > 0.3 MPa, the adjustment range of the flux in the filtration section is limited to ±10%. The constraint automaton is used to define the mass conservation constraint and the energy balance constraint, such as the outlet flow rate of the filtration section = the feed flow rate of the evaporation section ± 5% and the energy balance constraint; when the stirring power in the crystallization section increases by 15%, the heating power in the evaporation section needs to be reduced by 8% - 12% to ensure that the strategy solution complies with physical laws. The reinforcement learning agent explores strategy combinations within the constraints. When it is detected that the salt concentration in the discharged water is close to exceeding the standard, a collaborative strategy of "raising the temperature of the evaporation section by 2°C + increasing the addition of crystal seeds in the crystallization section" is preferentially generated to avoid cross-section conflicts caused by single-section adjustment.

[0053] Based on the embodiments provided in this application, the hierarchical decision-making architecture divides strategy generation into a cross-section collaborative layer and a single-section execution layer. The high level determines the adjustment priority through causal reasoning. For example, when the purity of industrial salt does not meet the standard, the activation energy threshold in the crystallization section is adjusted first. The low level sets parameter boundaries based on equipment status data. For example, when the transmembrane pressure difference of the membrane module is too high, the adjustment range of the flux in the filtration section is restricted to reduce cross-section conflicts caused by single-section isolated optimization in the prior art. The formal modeling of the constraint automaton transforms the physical limits of the equipment and the laws of mass and energy conservation into computable state transition constraints. For example, a strong correlation relationship between the outlet flow rate of the filtration section and the feed flow rate of the evaporation section is defined to ensure that the generated strategy solution naturally meets engineering feasibility and avoids the problem that the strategy is not executable caused by traditional optimization algorithms ignoring the equipment safety boundary. For example, the temperature adjustment in the evaporation section will not exceed the critical value of heat exchange tube fouling. The reinforcement learning agent explores and preferentially searches for strategies that meet the core indicators within the constraints, and dynamically adjusts the parameter boundaries in combination with the equipment status. For example, when the vibration of the centrifugal pump is abnormal, the adjustment range of the flow rate is tightened, making the strategy generation process have both goal orientation and equipment protection. For example, when the salt concentration in the discharged water is close to exceeding the standard, a collaborative strategy of raising the temperature of the evaporation section and increasing the addition of crystal seeds in the crystallization section is preferentially generated to ensure compliance while avoiding equipment overload.

[0054] As an alternative solution, Pareto optimal strategy combinations are screened from the strategy solution space, and combined with the cross-section influence intensity of the coupling weight matrix, collaborative control instructions for the evaporation section, crystallization section, and filtration section are generated, including: dynamically weighting the coupling weight matrix based on the graph attention mechanism, and combining the tabu search algorithm to avoid invalid adjustment combinations, and preferentially screening Pareto optimal solutions that meet the core indicators and have consistent cross-section collaboration in the strategy solution space.

[0055] In an alternative embodiment, the coupling weight matrix is weighted in real time through a graph attention mechanism. When the purity of industrial salt is the main optimization target, the combined weight of the activation energy threshold in the crystallization section and the membrane pore size in the filtration section is increased, and the collaborative adjustment strategy for these two sections of parameters is preferentially explored. The tabu search algorithm is used to record invalid adjustment combinations in the past 30 minutes. For example, increasing the temperature in the evaporation section by 5°C and decreasing the flux in the filtration section by 20% leads to equipment overload, and such combinations are automatically excluded during strategy screening to reduce repeated trial and error. A 50% priority bonus is given to strategies that meet the standard of the salt concentration in the discharged water. At the same time, the cross-section collaborative consistency is verified through the coupling weight matrix. For example, a strategy to decrease the impurity rate in the crystallization section needs to be accompanied by an increase in the retention efficiency of the filtration section by ≥10% to ensure the optimal balance of the strategy combination under multiple objectives.

[0056] Based on the embodiments provided in this application, graph attention is used to focus on cross-section combinations with high causal impacts. For example, when optimizing the purity of industrial salt, the collaborative adjustment weights of the activation energy threshold in the crystallization section and the membrane pore size in the filtration section are automatically increased, making the search for strategies related to core indicators more targeted. For example, the collaborative strategy between the crystallization section and the filtration section that has a significant impact on the purity of industrial salt is preferentially explored; the tabu search avoids invalid adjustments by recording historical invalid strategy combinations, such as the evaporation section temperature increase plus filtration section flux decrease plan that once caused equipment overload, avoiding repeated trial and error of traditional optimization algorithms, especially suitable for scenarios in wastewater treatment with large parameter adjustment inertia and high equipment response delay, such as reducing the life loss of membrane components caused by frequent invalid adjustments; the core indicator priority screening principle and cross-section collaborative consistency verification, such as requiring a strategy to decrease the impurity rate in the crystallization section to be accompanied by an increase in the retention efficiency of the filtration section, ensuring that the generated strategy combination does not deviate from the core requirements of environmental protection and resource utilization in the multi-objective balance.

[0057] As an alternative solution, the component data of the treated discharged water, recycled water, industrial salt, and solid waste are monitored in real time. If the standards are not met, model parameter correction, optimization strategy iteration, or cross-section collaborative logic recalculation is triggered, including: Identifying sudden fluctuations or continuous deviations in the component data of the four through a time-series anomaly detection algorithm, and determining the non-compliance type; Among them, the non-compliance types include but are not limited to end-component anomalies and process-section correlation anomalies; Specifically, end-component anomalies include but are not limited to various component indicators deviating from the preset compliance standards, such as abnormal salt concentration, insufficient purity, pollutant residue, etc. Process-section correlation anomalies: Based on the analysis of the digital twin model, locate the previous process anomalies that cause non-compliance at the end, such as oversaturation imbalance in the crystallization section, decreased retention efficiency in the filtration section, and degradation of mass transfer capacity in the evaporation section.

[0058] Based on the non-compliance type and the health status of the process section, dynamically allocate the trigger priorities for model correction, strategy iteration, or cross-section recalculation; Optionally, in this embodiment, the process section health status includes, but is not limited to, equipment performance degradation, abnormal structural parameters, and operation status fluctuations, etc.; based on the influence degree of non-compliance types (such as sudden over-standard, continuous deviation) and the process section health status, a three-level trigger priority is established, specifically: Emergency response: For sudden anomalies that may lead to system failures or emissions exceeding the standard, such as sudden changes in the composition of discharged water and critical equipment operation parameters exceeding the limit, trigger the cross-section collaborative logic recalculation with priority, and synchronously adjust the parameters of multiple process sections to quickly contain the risks; Sub-emergency adjustment: For continuous deviations that affect the treatment efficiency or product quality, such as fluctuations in the quality of recycled water and a slow decrease in the purity of industrial salt, trigger the iteration of optimization strategies with priority, and achieve progressive optimization through local parameter combination adjustment; Progressive correction: For model inaccuracies caused by equipment performance degradation, such as a decrease in the porosity of the membrane module and the accumulation of scale in the evaporator, trigger the correction of model parameters with priority, and calibrate the model characteristic parameters (such as mass transfer efficiency coefficient, filter layer degradation function) through real-time data to compensate for the effects of equipment aging or gradual changes in working conditions.

[0059] According to the priority arbitration result, select to execute the correction of model parameters, the iteration of optimization strategies, or the cross-section collaborative logic recalculation, and verify the response effect in real time.

[0060] In some embodiments of the present application, the correction of model parameters includes the following steps: Multi-source anomaly feature extraction: Synchronously analyze the real-time composition data (such as salt concentration, particle size distribution, turbidity) of discharged water, industrial salt, and solid waste and the equipment status data (such as vibration amplitude, scale thickness) through a temporal convolutional network to generate anomaly feature vectors; Kernel mapping and regression modeling: Use a radial basis kernel function to map the anomaly feature vectors to a high-dimensional space, construct a support vector regression model, and learn the non-linear relationship between anomaly types (such as sudden changes in salt load, out-of-control crystal particle size) and model parameter correction amounts (mass transfer efficiency coefficient, activation energy threshold, etc.); Dynamic learning rate adaptation: Adjust the kernel function bandwidth according to the real-time salt load. When the salt load is high, narrow the bandwidth to enhance local sensitivity (such as triggering correction when the salt concentration gradient deviation is ±0.2 g / L·m), and when the salt load is low, widen the bandwidth to suppress overfitting; Parameter correction and verification: Import the corrected parameters into the digital twin model, and deduce the composition data of the four within the next 5 minutes. If the prediction deviation does not converge (such as the purity fluctuation of industrial salt > ±0.5%), trigger secondary correction.

[0061] In a specific implementation, when the inlet salt load suddenly increases to 250g / L, the discharge water salt concentration detection value is 75mg / L (target ≤50mg / L), and the evaporation section heat transfer efficiency model predicted value deviates from the actual value by 12%, the time series convolutional network extracts abnormal features (salt concentration gradient +0.3g / L·m, vibration amplitude exceeds the limit); the support vector regression model outputs the mass transfer efficiency coefficient correction; the dynamic learning rate is adapted (the bandwidth is reduced by 30%), and the salt concentration gradient deviation of the corrected model converges; and the reduction of the discharge water salt concentration and the recovery of the industrial salt purity are verified.

[0062] Based on the embodiments provided in this application, the kernel bandwidth of the radial basis kernel function is adjusted in linkage with the salt load, breaking through the scenario limitations of the traditional fixed kernel function; the time series convolutional network synchronously analyzes the components and equipment data to accurately locate complex anomalies (such as high salt + equipment aging).

[0063] In some embodiments of the present application, the optimization strategy iteration includes the following steps: Strategy space encoding: Mapping process parameters (evaporation temperature, crystallization stirring power, filtration backwash frequency) into high-dimensional strategy vectors, embedding salt phase transition kinetic constraints (such as solubility-temperature curve, critical nucleation supersaturation) as hard boundaries; Covariance matrix adaptive evolutionary search: Design the fitness function to integrate salt recovery rate, equipment life and energy consumption indicators, and dynamically adjust the search direction through the covariance matrix; introduce preference-inspired generative adversarial networks to simulate expert experience to generate initial strategy populations and accelerate convergence; Multi-objective constraint verification: When screening Pareto solutions, exclude strategies that violate mass conservation (such as evaporation section feed volume > crystallization section processing capacity) or equipment limits (temperature > 120°C); Dynamic strategy update: The optimal strategy is sent to the physical device for execution, and the execution effect is recorded in the case library. The covariance matrix update rules of the covariance matrix adaptive evolutionary search are iteratively optimized through online reinforcement learning.

[0064] In a specific embodiment, when the crystal purity in the crystallization stage is continuously lower than 97.5% and there is no effective solution in the traditional strategy library, the strategy space is encoded (for example, temperature 110-120°C, stirring power 18-25kW); the preference-inspired generative adversarial network generates the initial strategy population, and the covariance matrix is ​​adaptively evolved to search for the optimal solution; it is verified whether the strategy meets the solubility constraint (for example, the supersaturation of the mother liquor is 1.2 at 118°C), and the purity improvement effect after execution is verified; online reinforcement learning is used to update the covariance matrix to optimize the subsequent search efficiency.

[0065] Based on the embodiments provided in this application, salt phase change constraint embedding and a hard boundary exclusion physical violation strategy are adopted; preference-inspired generative adversarial networks are used to accelerate the search and generate an initial population that conforms to expert experience, solving the cold start problem of covariance matrix adaptation evolution search.

[0066] In some embodiments of this application, cross-segment collaborative logic recalculation includes the following steps: Memory-enhanced relationship modeling: Construct a spatio-temporal memory network to store historical abnormal events (such as sudden increase in differential pressure in the filtration section, abnormal crystal particle size in the crystallization section) and corresponding cross-segment regulation effects, and extract key memory fragments through the attention mechanism; Dynamic coupling weight reconstruction: Based on real-time data and memory fragments, reconstruct the coupling weight matrix to strengthen implicit associations (such as the lag effect of stirring power in the crystallization section on the differential pressure in the filtration section); Adversarial collaborative deduction: Simulate extreme working conditions (such as instantaneous doubling of salt load, membrane module damage), and train the spatio-temporal memory network to generate collaborative strategies; Closed-loop verification and self-evolution: After implementing the collaborative strategy, monitor the component data of the four, and if the standard is still not met, trigger the memory update of the spatio-temporal memory network and the iteration of the weight matrix, gradually reducing the dependence on manual intervention.

[0067] In a specific implementation manner, when the differential pressure in the filtration section suddenly increases by 0.3 MPa and the root cause is that the crystal particle size in the crystallization section is too fine (D50 = 40 μm), the spatio-temporal memory network retrieves historical similar events (for example, when D50 = 45 μm, increasing the stirring power + backwashing is effective); reconstruct the coupling weight to strengthen the influence coefficient from crystallization to filtration; generate a virtual extreme scenario (for example, D50 = 30 μm), train the spatio-temporal memory network to output a strategy, such as stirring power + 25% + backwashing every 10 minutes; after execution, verify the increase in D50 and whether the differential pressure is stable, and store the strategy in the library for self-learning iteration.

[0068] Based on the embodiments provided in this application, based on the spatio-temporal memory network, historical events are associated with real-time data to achieve causal reasoning of cross-segment influences; by simulating extreme working conditions, the stability of the strategy in scenarios such as sudden salt load changes is improved.

[0069] Based on the embodiments provided in this application, the abnormal types of component data such as discharge water and industrial salt are identified through a time series anomaly detection algorithm, and the correction priority is dynamically assigned based on the non-compliance type and the health status of the process section, such as the priority triggering of strategy iteration for sudden high salt load, and the priority correction of model parameters for continuous deviations caused by equipment aging, thereby building an intelligent closed-loop feedback mechanism. For example, when a sudden drop in industrial salt purity is detected, the supersaturation control anomaly of the crystallization section is located, and the cross-section collaborative logic recalculation is triggered first to adjust the matching relationship between the evaporation section temperature and the crystallization section stirring rate; if the salt concentration of the discharge water continues to approach the excess limit and the membrane pressure difference of the filtration section increases synchronously, it is determined to be an equipment degradation problem, and the filter layer degradation rate function is corrected first and the membrane group cleaning plan is scheduled. This adjustment mode can match the optimal response strategy for different problem sources, significantly improving the stability of the processing results and the ability to adjust abnormal working conditions.

[0070] As an optional solution, the method further includes: Based on the crystal growth rate in the crystallization stage and the flow velocity distribution data on the membrane surface in the filtration stage, a dynamic correlation model between crystal size distribution and filter layer pressure difference was constructed to predict the nonlinear effect of crystals of different sizes on membrane flux. The digital twin model is used to deduce the quantitative relationship between crystal agglomeration or breakage and the risk of clogging in the filter section; When abnormal crystal size distribution is detected in the crystallization stage, the high-frequency pulse backwash mode of the filtration stage is triggered, and the stirring power of the crystallization stage is adjusted in conjunction to suppress excessive crystal refinement; For example, if the D50 particle size is less than 80 μm, it is determined that the crystal size distribution of the crystal stage is abnormal. The D50 particle size is the median diameter of the particle size distribution, which refers to the maximum particle size corresponding to the particles that account for 50% of the total volume of the particles on the cumulative particle size distribution curve, reflecting the average size of the particles.

[0071] If the digital twin model predicts that the pressure difference of the filter layer will exceed the safety threshold within the preset time, the feed flow rate of the crystallization section will be reduced in advance and the backup filter unit will be activated; In this embodiment, the preset time includes but is not limited to 30 minutes, etc.; the safety threshold of the filter layer pressure difference can be set to 0.9MPa (normal operating range 0.3-0.6MPa, critical alarm value 0.8MPa, exceeding 0.9MPa will cause irreversible blockage of the membrane pores).

[0072] Monitor the filter layer pressure difference fluctuation and crystal retention efficiency after coordinated adjustment. If they fail to meet the standards, reversely correct the crystallization stage nucleation rate or the filtration stage membrane flux threshold.

[0073] In this embodiment, continuous monitoring is performed for 30 minutes after adjustment. If one of the following situations occurs, it is determined that the standard is not met: the filter layer pressure difference still exceeds the threshold: the pressure difference reaches 0.85MPa in the 20th minute after the adjustment (close to the safety threshold and the rising rate is not less than 0.03MPa / 10min); the crystal retention efficiency decreases: the purity of industrial salt drops from 98% to 94% (indicating that fine crystals penetrate the filter membrane and the retention rate is <95%).

[0074] Reverse correction of the nucleation rate of the crystallization stage or the membrane flux threshold of the filtration stage, including: Correction of crystallization stage parameters: If blockage is caused by excessive crystal refinement (D50 < 20 μm), reduce the crystallization stage nucleation rate parameter from 1.5 times the supersaturation trigger to 1.2 times to reduce the generation of fine crystals. At the same time, increase the stirring power by 10% to enhance crystal collision and agglomeration and promote the formation of large particles. Modification of filtration section parameters: If uneven distribution of membrane flux leads to local blockage, dynamically reduce the membrane flux threshold from 100L / (m²·h) to 80L / (m²·h), and increase the backwash frequency from once per hour to once every 20 minutes (to enhance membrane surface cleaning).

[0075] Based on the embodiments provided in this application, a dynamic correlation model of crystal particle size distribution and filter layer pressure difference is constructed to achieve deep collaborative control of the crystallization section and the filtration section: through the crystal growth rate and membrane surface velocity distribution data, the influence of crystals of different particle sizes on the membrane flux is quantified, such as when the proportion of crystals with a particle size of <50μm is too high, the membrane pore blockage risk index increases; when the crystal particle size of the crystallization section is abnormal, the high-frequency pulse backwashing of the filtration section is triggered in linkage, and the stirring power of the crystallization section is adjusted to suppress crystal refinement, forming a pre-control mechanism from crystallization control to filtration protection; if the model predicts that the filter layer pressure difference will exceed the safety threshold, the feed flow of the crystallization section is reduced in advance and the spare filtration unit is activated; the collaborative adjustment effect is monitored and the nucleation rate of the crystallization section or the membrane flux threshold of the filtration section is reversely corrected to form a closed loop of optimization logic. The common problem of "abnormal crystal particle size leads to decreased filtration efficiency" in wastewater and salt treatment is solved. Through the cross-segment dynamic correlation model and pre-control strategy, the risk of clogging in the filtration section is transformed from post-processing to pre-quantified prediction and active suppression, which significantly improves the collaborative efficiency and operation reliability of the system in the crystal growth and solid-liquid separation links.

[0076] According to another aspect of the embodiment of the present application, a wastewater and salt treatment process optimization system based on digital twin is provided, such as Figure 4 As shown in , the system includes: A virtual model building module 401 is used to build a digital twin model that is synchronously mapped with the processing system, and the digital twin model integrates the process parameter association relationship of the evaporation section, the crystallization section, and the filtration section; The real-time data acquisition module 402 is used to collect the component data and flow data of the contaminated salt wastewater corresponding to each process section in real time through the component analyzers and monitoring sensors deployed in each process section; The model correction module 403 is used to correct the model characteristic parameters of each process section in the digital twin model based on the component data and flow data collected in real time; The optimization strategy generation module 404 is used to generate process optimization strategies corresponding to each process section according to the corrected digital twin model and the optimization objectives, so as to adjust the treatment system; The closed-loop feedback module 405 is used to monitor the component data of the treated discharged water, recycled water, industrial salt and solid waste in real time. If not up to standard, it triggers the correction of model parameters, the iteration of optimization strategies or the recalculation of cross-section collaborative logic.

[0077] According to another aspect of the embodiments of the present application, an electronic device for implementing the above-mentioned digital twin-based wastewater and contaminated salt treatment process optimization method is also provided. This electronic device can be a terminal device or a server. In this embodiment, the electronic device is taken as an example of a server for illustration. The electronic device includes a memory and a processor. A computer program is stored in the memory, and the processor is configured to execute the steps in any of the above method embodiments through the computer program.

[0078] Optionally, in this embodiment, the above-mentioned electronic device can be at least one network device among multiple network devices in a computer network.

[0079] Optionally, those of ordinary skill in the art can understand that the electronic device can also be a terminal device such as a smart phone, a tablet computer, a handheld computer, a mobile Internet device, a PAD, etc.

[0080] In other embodiments, the above-mentioned terminal device or server can be a node in a distributed system. Among them, the distributed system can be a blockchain system, and the blockchain system can be a distributed system formed by connecting the multiple nodes in the form of network communication. Among them, the nodes can form a peer-to-peer (P2P, PeerToPeer) network, and any form of computing device, such as a server, a terminal and other electronic devices, can become a node in the blockchain system by joining the peer-to-peer network.

[0081] Optionally, in this embodiment, for the embodiments to be implemented by the above-mentioned respective unit modules, reference can be made to the above-mentioned respective method embodiments, which will not be elaborated here.

[0082] In the above embodiments of the present application, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0083] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A wastewater and salt treatment process optimization method based on digital twin, characterized in that: include: Establishing a digital twin model that is synchronously mapped with the processing system, wherein the digital twin model integrates the correlation relationship between the process parameters of the evaporation section, the crystallization section, and the filtration section; By deploying component analyzers and monitoring sensors in each process section, the component data and flow data of the salt wastewater corresponding to each process section are collected in real time; Based on the component data and flow data collected in real time, correcting the model characteristic parameters of each process section in the digital twin model; Generating a process optimization strategy corresponding to each process stage according to the modified digital twin model and the optimization target for adjusting the processing system; Real-time monitoring of the composition data of treated discharge water, recycled water, industrial salt and solid waste. If the standards are not met, model parameter correction, optimization strategy iteration or cross-segment collaborative logic recalculation will be triggered.

2. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 1 is characterized in that: When constructing the digital twin model, the device status data is incorporated into the model construction elements, and the device status data includes mechanical performance parameters, physical structure parameters and operating status parameters; Among them, the mechanical performance parameters include the vibration amplitude of the centrifugal pump and the transmembrane pressure difference of the membrane assembly; the physical structure parameters include the scaling thickness of the evaporator heat exchange tube and the porosity of the nanofiltration membrane; the operating status parameters include the equipment start and stop time and the load rate.

3. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 1 is characterized in that: The method of correcting the model characteristic parameters of each process section in the digital twin model based on the component data and flow data collected in real time includes: Analyze the deviation between the composition data and flow data collected in real time and the predicted values ​​of the digital twin model; Based on the deviation analysis results, the mass transfer efficiency coefficient, crystallization activation energy threshold and filter layer degradation rate function in the digital twin model are dynamically corrected.

4. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 3 is characterized in that: The method of dynamically correcting the mass transfer efficiency coefficient, crystallization activation energy threshold and filter layer degradation rate function in the digital twin model based on the deviation analysis results includes: According to the deviation between the predicted value and the actual detected value of the salt concentration gradient in the evaporation section, the mass transfer efficiency coefficient is adjusted by an adaptive gradient descent algorithm, wherein the learning rate is positively correlated with the real-time salt load; Based on the supersaturation deviation of the mother liquor in the crystallization stage and the fluid shear force data, the crystallization activation energy threshold is adjusted by a dynamic sliding mode control algorithm to suppress abnormal fluctuations in the crystal nucleation rate; According to the uniformity deviation of the particle size distribution of retained salt particles in the filtration section, a multi-scale convolution kernel is used to extract local blockage features and reconstruct the filter layer degradation rate function; if the membrane surface velocity distribution cloud map shows regional velocity anomalies, the porosity parameters of the corresponding area are corrected first.

5. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 1 is characterized in that: The optimization objectives include the component compliance indicators of the treated discharge water, recycled water, industrial salt and solid waste; the process optimization strategies corresponding to each process stage are generated according to the modified digital twin model and the optimization objectives, including: According to the revised digital twin model, the explicit process correlation and implicit dynamic interaction of the evaporation section, crystallization section and filtration section are analyzed to construct a coupling weight matrix; The predicted composition data of the four under the current process parameters are deduced through the revised digital twin model, and the sources of deviation from the optimization target are analyzed; Based on the analysis results of the deviation sources, a strategic solution space that meets the constraints is generated; the constraints include that the adjustment range of process parameters is subject to the physical limits of the equipment; the cross-segment coupling effect must comply with the laws of conservation of mass and energy; and the core indicators of discharged water and industrial salt are prioritized. The Pareto optimal strategy combination is selected from the strategy solution space, and the coordinated control instructions of the evaporation section, the crystallization section and the filtration section are generated in combination with the cross-segment influence strength of the coupling weight matrix.

6. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 5 is characterized in that: The dynamic construction of the coupling weight matrix includes: Define the direct coupling relationship between process stages through mass conservation equation and energy balance equation; Train a time series neural network based on historical operation data to identify cross-segment delays or nonlinear effects; Verify the effectiveness of explicit and implicit weights based on data collected in real time.

7. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 1, characterized in that: The real-time monitoring of the composition data of the treated discharged water, recycled water, industrial salt and solid waste will trigger model parameter correction, optimization strategy iteration or cross-segment collaborative logic recalculation if the data do not meet the standards, including: The time series anomaly detection algorithm is used to identify sudden fluctuations or continuous deviations in the four component data and determine the type of non-compliance; Dynamically assign triggering priorities for model correction, strategy iteration, or cross-stage recalculation based on non-conformance type and process stage health status; Based on the priority arbitration results, choose to execute model parameter correction, optimization strategy iteration, or cross-segment collaborative logic recalculation, and verify the response effect in real time.

8. The method for optimizing wastewater and salt treatment process based on digital twin according to claim 1, characterized in that: The method further comprises: Based on the crystal growth rate in the crystallization stage and the flow velocity distribution data on the membrane surface in the filtration stage, a dynamic correlation model between crystal size distribution and filter layer pressure difference was constructed to predict the nonlinear effect of crystals of different sizes on membrane flux. The digital twin model is used to deduce the quantitative relationship between crystal agglomeration or breakage and the risk of clogging in the filtration section; When abnormal crystal size distribution is detected in the crystallization stage, the high-frequency pulse backwash mode of the filtration stage is triggered, and the stirring power of the crystallization stage is adjusted in conjunction to suppress excessive crystal refinement; If the digital twin model predicts that the pressure difference of the filter layer will exceed the safety threshold within a preset time, the feed flow rate of the crystallization section is reduced in advance and the standby filter unit is activated; Monitor the filter layer pressure difference fluctuation and crystal retention efficiency after coordinated adjustment. If they fail to meet the standards, reversely correct the crystallization stage nucleation rate or the filtration stage membrane flux threshold.

9. A wastewater and salt treatment process optimization system based on digital twin, characterized in that: include: A virtual model building module, used to build a digital twin model that is synchronously mapped with the processing system, wherein the digital twin model integrates the process parameter association relationship of the evaporation section, the crystallization section, and the filtration section; The real-time data acquisition module is used to collect the composition data and flow data of the salt wastewater corresponding to each process section in real time through the component analyzers and monitoring sensors deployed in each process section; A model correction module, used to correct the model characteristic parameters of each process section in the digital twin model based on the component data and flow data collected in real time; An optimization strategy generation module, for generating a process optimization strategy corresponding to each process stage according to the modified digital twin model and the optimization target, so as to adjust the processing system; The closed-loop feedback module is used to monitor the composition data of treated discharge water, recycled water, industrial salt and solid waste in real time. If the standards are not met, it will trigger model parameter correction, optimization strategy iteration or cross-segment collaborative logic recalculation.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 8 through the computer program.

Images