Method for preparing battery-grade manganese sulfate by coupling of mvr concentration and intelligent temperature control

Abstract

This invention discloses a method for preparing battery-grade manganese sulfate using a combined MVR (Mechanical Vapor Reduction) concentration and intelligent temperature control system, belonging to the field of hydrometallurgical and chemical process control technology. This method aims to solve the technical challenges in the MVR concentration process, such as wide particle size distribution of manganese sulfate crystals due to temperature fluctuations and high energy consumption caused by control system response delays. The core solution involves: online monitoring of the density and temperature of the concentrate at the evaporator outlet of the MVR system, with the supersaturation calculated in real time by an intelligent temperature control system; dynamically coordinating and adjusting the compressor frequency and heater power by comparing the supersaturation with a preset optimal range of 1.05-1.20, thereby precisely stabilizing the evaporation temperature and supersaturation; after the solution is concentrated to 300-350 g / L, it is sent to a crystallizer to complete crystallization and separation, obtaining battery-grade manganese sulfate crystals. This invention is mainly used for the efficient preparation of battery-grade manganese sulfate with uniform crystal particle size and achieves energy savings in the process.

Description

Technical Field

[0001] This invention belongs to the field of hydrometallurgical and chemical process control technology, specifically relating to a method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling. Background Technology

[0002] In hydrometallurgy, the preparation of battery-grade manganese sulfate typically requires evaporation and concentration to achieve supersaturation, followed by cooling and crystallization. Traditional evaporation and concentration processes, such as multi-effect evaporation, suffer from high energy consumption. Mechanical vapor recompression (MVR) technology, due to its high thermal efficiency, shows promise for concentrating such materials. However, in actual operation, the evaporation temperature of the MVR system is easily affected by factors such as steam pressure, feed conditions, and system load, resulting in fluctuations. These temperature fluctuations directly lead to difficulties in maintaining stable supersaturation of manganese sulfate in the concentrate. Fluctuations in supersaturation affect the nucleation and growth rate of the subsequent crystallization process, easily resulting in a wide crystal size distribution in the final product, affecting product consistency and quality. To stabilize the crystallization process, an attempt has been made to introduce a temperature feedback-based control system to regulate the evaporation conditions. However, in complex MVR systems, there is an unavoidable lag in temperature sensor signal transmission, controller calculation, and actuator (such as compressor and heater) action, and the system's thermal inertia is also significant. This response delay makes it difficult for the control system to compensate for rapid process disturbances in a timely and accurate manner. Correction is often only made after the temperature has already deviated from the target value, causing the evaporation temperature to oscillate repeatedly around the setpoint. This not only fails to effectively stabilize supersaturation but may also increase energy consumption due to frequent equipment adjustments. Therefore, how to achieve rapid, stable, and precise control of the evaporation temperature while utilizing MVR technology for efficient concentration, thereby stabilizing the supersaturation of the solution within a narrow range suitable for crystallization, has always been a challenging technical difficulty in this field. Summary of the Invention

[0003] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0004] Another objective of this invention is to provide a method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling. This method can maintain the supersaturation of the manganese sulfate solution within a suitable crystallization range of 1.05-1.20 through real-time monitoring and feedback control during the MVR concentration stage. This effectively suppresses the runaway supersaturation caused by temperature fluctuations, promotes uniform nucleation and growth of crystals during crystallization, and obtains battery-grade manganese sulfate crystals with a concentrated particle size distribution. At the same time, timely and coordinated adjustment reduces ineffective energy consumption caused by system response delays, thereby improving the overall energy efficiency of the process.

[0005] To achieve these objectives and other advantages of the present invention, a method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling is provided, comprising the following steps: S1: The manganese sulfate raw material solution is pretreated by filtration to remove suspended impurities and obtain a clear manganese sulfate solution; S2: The clarified manganese sulfate solution is fed into the MVR concentration system, which includes an evaporator, a compressor, a heater and a condenser. An online density meter and a temperature sensor are installed at the outlet of the evaporator to monitor the density and temperature of the concentrated manganese sulfate solution in real time. S3: The density and temperature data monitored by the online density meter and temperature sensor are transmitted to the intelligent temperature control system in real time. The intelligent temperature control system calculates the supersaturation of the manganese sulfate solution based on the received density and temperature data. The intelligent temperature control system compares the calculated supersaturation with the preset supersaturation range of 1.05-1.20. Based on the comparison result, a control command is generated to coordinately adjust the frequency of the compressor and the power of the heater to control the evaporation temperature so that the supersaturation of the manganese sulfate solution is maintained at 1.05-1.20. When the concentration of the manganese sulfate solution reaches 300-350 g / L, the concentrated manganese sulfate solution from the MVR concentration system is sent to the crystallizer; in the crystallizer, cooling and crystallization are carried out under the conditions of a cooling rate of 0.5-2 ℃ / min and a stirring speed of 50-150 rpm. S4: After crystallization, solid-liquid separation is performed to obtain battery-grade manganese sulfate crystals. This invention monitors the density and temperature of the concentrated solution online in real time, calculates the current supersaturation, compares this calculated value with a preset suitable crystallization range (1.05-1.20), and then dynamically and coordinately adjusts the compressor frequency and heater power in the MVR system. This achieves precise and stable control of the evaporation temperature, maintaining the supersaturation within the target range. This invention uses supersaturation, a key parameter directly affecting crystallization quality, as a direct control target. Through closed-loop feedback and actuator coordinated adjustment, it overcomes the shortcomings of traditional single temperature control, such as lag and overshoot. Stable supersaturation creates conditions for uniform crystal nucleation and growth, significantly improving the concentration and consistency of crystal particle size distribution in manganese sulfate products. Furthermore, this invention reduces ineffective energy consumption caused by control oscillations or slow response through timely and matched power adjustments, improving the overall energy efficiency of the entire concentration-crystallization process.

[0006] Preferably, the concentrated manganese sulfate solution, which has reached a concentration of 300-350 g / L after being concentrated by the MVR concentration system, is transported from the outlet of the evaporator to the inlet of the crystallizer through an insulated conveying pipeline; A tubular heat exchanger is installed on the insulated conveying pipeline; a branch of the high-temperature secondary steam from the compressor outlet is used as the heat medium and introduced into the shell side of the tubular heat exchanger; a concentrated manganese sulfate solution with a temperature of 70°C to 85°C, which is to be conveyed to the crystallizer, is introduced into the tube side of the tubular heat exchanger. By adjusting the flow rate of the heat medium entering the shell side, the heat loss of the manganese sulfate concentrate in the insulated transport pipeline is compensated, so that the temperature fluctuation of the manganese sulfate concentrate is within ±2℃ throughout the entire process of transporting it from the evaporator outlet to the crystallizer inlet.

[0007] This invention employs an insulated conveying pipeline connecting the evaporator outlet and the crystallizer inlet, with a tubular heat exchanger installed on the pipeline. The core of its approach lies in diverting a branch stream of the high-temperature secondary steam generated during the MVR system's operation, which serves as the heat medium in the shell side of the heat exchanger. Meanwhile, the 70-85℃ manganese sulfate concentrate to be conveyed is introduced into the tube side. The heat loss of the concentrate during conveying is compensated in real-time by adjusting the heat medium flow rate in the shell side. This invention constructs an online heat compensation closed loop, using system waste heat as the source and the conveyed material as the object, actively maintaining the temperature stability of the concentrate using the sensible heat of the high-temperature steam. This invention ensures that the temperature fluctuation of the continuously conveyed concentrate is strictly controlled within ±2℃, effectively avoiding premature rise in supersaturation and premature crystallization in the pipeline caused by temperature drops during conveying. This guarantees that the concentrate enters the crystallizer in a stable and uniform physical state, laying a solid foundation for obtaining uniformly sized crystalline products.

[0008] Preferably, a static mixer is installed in series near the inlet of the crystallizer in the insulated conveying pipeline. The static mixer has multiple spiral mixing units arranged sequentially along the flow direction. The mixing units are made of stainless steel sheets, and each unit has a 180° twist angle along the axial direction. The spiral directions of adjacent mixing units are opposite. After the manganese sulfate concentrate flows through the static mixer, it directly enters the crystallizer.

[0009] This invention involves installing a static mixer with multiple sets of internal spiral mixing units in series at the end of the insulated delivery pipeline and before the crystallizer inlet. The spiral directions of adjacent units are set to opposite directions. As the solution flows through these alternating spiral units, the fluid is continuously cut, divided, rotated, and recombine, generating a strong radial mixing and back-mixing effect without external power. This forcibly eliminates the temperature and concentration gradients naturally formed in the pipeline. This ensures that the manganese sulfate concentrate entering the crystallizer achieves a highly uniform state at the microscale, providing a completely homogeneous initial condition for the subsequent cooling and crystallization process. This promotes the synchronous formation of crystal nuclei and the uniform growth of crystals, ultimately yielding a battery-grade manganese sulfate crystal product with a more concentrated and consistent particle size distribution.

[0010] Preferably, the control logic of the intelligent temperature control system is configured as follows: When it is necessary to increase the evaporation temperature of the evaporator, the power of the heater is increased first. After the temperature sensor starts to rise and stabilizes, the frequency of the compressor is gradually increased within the preset safe frequency range of the compressor, according to the overall energy efficiency model of the system. When it is necessary to reduce the evaporation temperature, the frequency of the compressor is reduced first. After the temperature sensor starts to drop and stabilizes, the power of the heater is reduced accordingly. The single adjustment of the compressor's operating frequency shall not exceed 10% of its current frequency value, and the time interval between two adjacent frequency adjustments shall not be less than 30 seconds.

[0011] This invention configures a specific control logic for the intelligent temperature control system: when heating is required, the heater power is increased rapidly to achieve initial temperature tracking. After the temperature feedback stabilizes, the compressor frequency is gradually increased based on the energy efficiency model for fine-tuning and steady-state maintenance. When cooling is required, the compressor frequency is decreased first. After the temperature begins to drop and stabilizes, the heater power is reduced accordingly. Simultaneously, the compressor's single adjustment amplitude is strictly limited to no more than 10% of the current value, and the adjustment interval is no less than 30 seconds. This invention identifies and utilizes the differences in the dynamic characteristics of different actuators to formulate a primary-secondary ordered strategy of "fast-response devices performing coarse adjustments first, and slow-response devices performing fine adjustments later." It also avoids system oscillations by limiting the amplitude and frequency of the large-inertia component (compressor). This achieves organic coordination and decoupling of the compressor and heater's adjustment actions, ensuring that the evaporation temperature can smoothly and without overshoot approach and stabilize at the target value. This provides a solid foundation for the continuous and stable control of supersaturation and guarantees the continuous and stable operation of the crystallization process.

[0012] Preferably, the method for establishing the overall energy efficiency model of the system is as follows: During the stable operation of the MVR concentration system, multiple sets of historical operating data under different operating conditions are collected and recorded. Each set of data includes: compressor frequency, heater power, real-time density and temperature of manganese sulfate solution at the evaporator outlet, steam pressure and temperature at the compressor inlet and outlet, and total electrical power of the system. Based on the first law of thermodynamics and the principle of heat transfer, with the goal of minimizing the total electrical power of the system, and with the compressor efficiency model, heater heat transfer model, and evaporator material and energy balance equation as constraints, regression fitting is performed on the historical operating data to obtain a steady-state energy efficiency regression model that describes the quantitative relationship between the total electrical power of the system and the compressor frequency, heater power, and solution evaporation temperature.

[0013] This invention proposes establishing a system-wide energy efficiency model. The method involves collecting historical data under multiple operating conditions during stable operation, including compressor frequency, heater power, solution properties, steam state, and total electrical power. Based on the first law of thermodynamics and heat transfer principles, and with the goal of minimizing total electrical power, constrained by equipment performance models and material energy balance, a steady-state energy efficiency regression model describing the quantitative relationship between total electrical power, operating parameters, and evaporation temperature is fitted. This invention condenses the physical characteristics and inter-coupling relationships of key equipment such as compressors, heaters, and evaporators in the system into a computable steady-state energy efficiency prediction model through a combination of data-driven and mechanistic modeling. This allows for the quantitative assessment of energy consumption levels under different combinations of operating parameters. This invention provides a model-based energy efficiency assessment and prediction tool for intelligent temperature control systems, enabling the system to optimize towards a lower energy consumption steady-state operating point while meeting supersaturation control requirements. This lays the foundation for achieving a balance between process stability and energy consumption optimization.

[0014] Preferably, the application of the overall system energy efficiency model is as follows: During operation, the intelligent temperature control system acquires the density and temperature of the manganese sulfate solution at the evaporator outlet in real time, and deduces the target evaporation temperature range to be maintained based on the preset supersaturation range of 1.05-1.20; then, the current target evaporation temperature and the real-time collected compressor inlet secondary steam pressure and temperature are used as inputs and substituted into the steady-state energy efficiency regression model to calculate and output a set of optimal operating parameters that minimizes the total power of the system under the current operating conditions. The optimal operating parameter set includes recommended compressor frequency setting values ​​and heater power setting values. In the control execution phase: when the intelligent temperature control system needs to adjust the operating parameters to maintain supersaturation, it first compares the calculated optimal compressor frequency setting with the current frequency. If the absolute value of the deviation between the two is greater than twice the allowable adjustment threshold of the compressor frequency, the model output is temporarily ignored, and the operation is performed only according to the control logic. If the deviation is within the threshold range, the recommended value output by the model is used as the final target value for compressor frequency adjustment during the execution of the control logic.

[0015] To address the coordination issue between model application and actual control, this invention designs specific model application rules: The intelligent temperature control system first uses real-time monitored density and temperature to deduce the target evaporation temperature range to be maintained. Then, this target temperature, along with real-time steam parameters, is input into the steady-state energy efficiency regression model to calculate the theoretically optimal combination of compressor frequency and heater power setpoints that minimizes total electrical power. During control execution, the system first compares the model-recommended compressor frequency setpoint with the current actual frequency. If the absolute value of the deviation exceeds twice the allowable adjustment threshold, the model recommendation is temporarily disregarded, and operation is performed solely based on the established coordinated control logic. Only when the deviation is within the safety threshold is the model-recommended value used as the final target value for frequency adjustment during the adjustment process following the coordinated control logic, and the frequency is progressively approximated. This invention sets up a "safety valve" and "guide" based on the actual operating state for model-intervention in real-time control. It distinguishes between the "safe optimization range" and the "risk disturbance range" through deviation threshold judgment, thereby safely and smoothly embedding the model's global steady-state optimization capability into a dynamic control loop where stability is the primary objective. This invention, while ensuring the absolute stability of process control, intelligently guides the system operating parameters to gradually shift towards a steady-state operating point with lower energy consumption, achieving a dynamic balance between process stability and energy consumption optimization, as well as ensuring safety.

[0016] Preferably, the intelligent temperature control system continuously monitors and records the optimal compressor frequency setpoint output by the steady-state energy efficiency regression model and the compressor frequency value during actual stable operation of the system. If the system runs continuously for more than one preset model evaluation cycle, and within that cycle, any of the following conditions continuously occur and exceed 20% of the total running time, the current steady-state energy efficiency regression model is determined to be mismatched with the actual operating state: (i) The absolute deviation between the optimal compressor frequency setpoint and the final stable compressor frequency value continues to be greater than the allowable adjustment threshold of the compressor frequency; (ii) After adjustment based on the model's recommended values, the actual measured value of the total power of the system continues to be higher than the minimum power value predicted by the model by a preset percentage. After determining that the model is mismatched, the intelligent temperature control system initiates an online self-update program for the model, including the following steps: 1) Maintain the system in stable operation near the current dominant operating condition for at least one data acquisition period. During this period, actively and slightly adjust the compressor frequency and heater power to cover the operating window around the current operating point, and simultaneously collect multiple sets of new operating data and corresponding total system power data. 2) The parameters of the steady-state energy efficiency regression model are refitted and updated using the newly collected dataset to generate an updated steady-state energy efficiency regression model; 3) After the model update is complete, switch to using the updated model for online energy efficiency optimization and operation guidance.

[0017] This invention incorporates an online self-updating program in the intelligent temperature control system: the system continuously monitors and compares the optimal compressor frequency setpoint output by the model with the actual stable operating frequency, and calculates in real time the difference between the actual total power and the minimum power predicted by the model. When continuous operation exceeds one evaluation cycle and the aforementioned deviation consistently exceeds a set threshold, the system determines that the current model is mismatched and immediately initiates the self-updating process. This process first maintains the system's stable operation near the current dominant operating condition, actively and slightly adjusting the compressor frequency and heater power, and collecting multiple sets of new operating data and their corresponding total power within the operating window around the current operating point. Then, these fresh data are used to refit and update the parameters of the original steady-state energy efficiency regression model. After the update is completed, the system switches to using the new model for subsequent online energy efficiency optimization and operational guidance. This invention constructs a closed-loop model maintenance mechanism of "monitoring-diagnosis-relearning." It establishes online diagnostic criteria based on continuous comparison of actual operating values ​​and model predictions, enabling the system to autonomously perceive model performance degradation. Furthermore, it actively acquires new data representing the current true characteristics of the system through controlled experiments, driving the model to update its parameters and thus adaptively tracking the slow time-varying characteristics of the system. This invention endows the energy efficiency model with the ability to self-evolve and maintain long-term accuracy, ensuring the long-term effectiveness of optimization guidance. It fundamentally solves the problem of model failure caused by equipment aging or operating condition drift, enabling continuous and reliable energy efficiency optimization throughout the entire process.

[0018] Preferably, in step 1) of the online self-updating procedure for the model, the specific execution method for actively and slightly adjusting the compressor frequency and heater power is as follows: During the data acquisition period, the intelligent temperature control system temporarily switches to model parameter exploration mode; the model parameter exploration mode alternately executes the parameter adjustment phase and the steady-state maintenance phase according to a preset cycle. The execution process for each cycle is as follows: a) During the parameter adjustment phase, using the operating condition established at the end of the previous steady-state maintenance phase as the baseline, and based on a pre-designed orthogonal experimental table, the compressor frequency and heater power of the system are sequentially adjusted to the operating points specified in each combination in the table; the orthogonal experimental table contains multiple sets of operating parameter combinations with finite deviation values ​​based on the baseline; at each operating point, the system maintains stable operation for a first preset duration, and a set of operating data containing the total electrical power of the system is collected; during this phase, the absolute value of the compressor frequency adjustment does not exceed 1.5 times the allowable adjustment threshold of the compressor frequency, and the absolute value of the heater power adjustment does not exceed 5% of its current power value; b) In the subsequent steady-state maintenance phase, the system restores the operating parameters to the reference point and, according to the control logic of the intelligent temperature control system, controls the supersaturation of the manganese sulfate solution to be maintained within the preset range of 1.05-1.20, and continues to run for a second preset duration; the second preset duration is not less than twice the first preset duration. After exploring all parameter combinations according to the orthogonal experimental table, the system exits the model parameter exploration mode and uses the new operational data collected during the all-parameter adjustment phase to perform model parameter refitting and updating in step 2).

[0019] To address the dual interference of data acquisition on production stability and data quality, this invention structures the proactive adjustment process into alternating periodic parameter adjustment and steady-state maintenance phases. In the parameter adjustment phase, based on the previous steady-state point, the system is sequentially driven to designated operation points according to a pre-designed orthogonal experimental table containing combinations of finite deviations. At each point, the system runs stably for a preset duration to collect data, with the adjustment range strictly limited within a safe range. In the subsequent steady-state maintenance phase, the system restores the operating parameters to the baseline point and runs for an extended period strictly according to oversaturation control logic to ensure the process returns to stability. This invention decouples the continuous model update data acquisition task into discrete, controlled "exploration" and "recovery" cycles. It utilizes orthogonal design to efficiently cover the current operating condition neighborhood, while the significantly extended steady-state maintenance period ensures the stability of the main production process and the qualification of product quality. This invention successfully acquires statistically representative steady-state operating point data for model updates while maximizing the continuous stability of the production process, achieving a safe balance and efficient coordination between model parameter exploration and smooth operation of the main process.

[0020] Preferably, in each parameter adjustment phase of the model parameter exploration mode, after the system is driven to an operating point specified by the orthogonal experimental table, the intelligent temperature control system executes a steady-state data acquisition trigger mechanism, specifically as follows: The system continuously monitors the real-time temperature and density of the manganese sulfate solution at the evaporator outlet, and calculates the moving average and moving standard deviation of its most recent N sampled values ​​at fixed time intervals. When both of the following conditions are met simultaneously, the system determines that the production operation has reached a quasi-steady state suitable for model updates, and automatically opens a data acquisition window with a duration of the first preset time: A) The moving average of the temperature and the moving average of the density fall within a preset allowable fluctuation range centered on the expected steady-state value of the operating point; B) The moving standard deviation of the temperature and the moving standard deviation of the density are both less than or equal to their respective steady-state determination thresholds; During the data acquisition window, the system acquires and records a complete set of operational data at a constant sampling frequency. If the quasi-steady-state determination condition is not met after the maximum waiting time limit is reached, the system abandons data acquisition at that operation point, records an acquisition failure, and jumps to the next combination of operational parameters to continue execution according to the orthogonal experimental table.

[0021] To address the issue of inaccurate data acquisition timing, this invention establishes a steady-state data acquisition trigger mechanism: the system continuously monitors the real-time temperature and density of the solution at the evaporator outlet and calculates the moving average and moving standard deviation of its most recent N sampled values. Only when the moving averages of these two parameters simultaneously fall within the allowable fluctuation range centered on their expected steady-state value, and their moving standard deviations are both less than their respective steady-state judgment thresholds, does the system determine that a quasi-steady-state suitable for model updating has been reached, and automatically opens a data acquisition window of fixed duration. If the conditions are not met within the maximum waiting time limit, data acquisition at that point is abandoned. This invention constructs an objective, online quasi-steady-state criterion based on the statistical stability of the data itself (mean represents the state, standard deviation represents fluctuation), replacing subjective or fixed-time waiting. This invention fundamentally ensures that every data point input into the model update program represents the true and stable performance state achieved by the system at a specific operating point, thereby greatly improving the quality and representativeness of the original data used for model fitting and guaranteeing the accuracy and reliability of model parameter updates.

[0022] Preferably, the pre-designed orthogonal experimental table is dynamically generated by the intelligent temperature control system based on the current operating conditions and real-time operational constraints each time the online self-update program of the model is executed. The generation method includes the following steps: Step 1: When starting the model parameter exploration mode, take the current steady-state operating point of the system as the reference operating condition and determine the reference value F0 of the compressor frequency and the reference value P0 of the heater power at this time. Step 2: Read the real-time operating constraints pre-stored in the system. The constraints include: the maximum absolute value ΔF that the compressor frequency is allowed to be increased. max The maximum absolute value of the allowable downward adjustment ΔF min And the maximum percentage increase α in heater power. max The maximum percentage that can be reduced, α min The constraints are dynamically set based on the equipment's safe operating limits and the real-time requirements for maintaining process stability. Step 3: Based on the reference value (F0, P0) and the operational constraints, define a safe and feasible operating space Ω on the compressor frequency-heater power plane, permissible in this parameter exploration; the operating space Ω is a rectangular region with its frequency boundary being [F0-ΔF]. min, F0+ΔF max The power boundary is [P0×(1-α)]. min ), P0×(1+α max )]; Step 4: Within the operating space Ω, an orthogonal experimental design method is used to select m statistically representative operating points; the selected operating points must satisfy the following conditions: (a) they are distributed as evenly as possible within the Ω space; (b) they include the reference operating point (F0, P0). Step 5: Encode and combine the m operation points selected in Step 4 and their preset adjustment order to generate an orthogonal experimental table specifically for this model parameter exploration; the setting of the adjustment order follows the path principle of starting from the reference point, exploring from near to far, and finally returning to the reference point, so as to reduce process disturbances during continuous switching of operation points.

[0023] To address the mismatch between static experimental tables and dynamic operating conditions, this invention designs orthogonal experimental tables for dynamic generation: Each time model parameter exploration is initiated, the system uses the current steady-state operating point as a benchmark, reads real-time updated equipment safety limits and process stability constraints, and dynamically defines a rectangular safe and feasible operating space permissible for this exploration on the compressor frequency-heater power plane. Subsequently, using orthogonal experimental design methods, a series of evenly distributed, statistically representative operating points, including the benchmark point, are selected within this space and encoded according to the principle of "from near to far, and finally back," generating a dedicated exploration sequence table for this purpose. This invention transforms experimental design from "static pre-setting" to "dynamic coupling," ensuring that the parameter exploration guidance closely aligns with the real-time operating conditions and immediate safety boundaries at each initiation. This invention ensures that each active data acquisition process is conducted within the current absolutely safe operating range, fundamentally avoiding the risk of exceeding limits. Simultaneously, by evenly distributing points within the current feasible space, high-quality data reflecting the latest energy efficiency characteristics of the system can be obtained more efficiently and representatively, thereby significantly improving the overall adaptability, safety, and reliability of the model self-updating program.

[0024] Preferably, the step of the intelligent temperature control system calculating the supersaturation of the manganese sulfate solution based on the received density and temperature data specifically includes the following data processing and control logic: S101. The system continuously receives real-time density data ρ(t) and temperature data T(t) from an online density meter and temperature sensor at a fixed sampling period; the latest consecutive N density data points and N temperature data points received are stored in a first-in-first-out (FIFO) density data buffer and a temperature data buffer, respectively. S102. In each control cycle, the system calculates the moving average and moving standard deviation of the data in the two buffers respectively; let the density moving average be ρ and the moving standard deviation be σ. ρThe temperature moving average is T, and the moving standard deviation is σ. T ; S103. Substitute ρ and T obtained in step S102 into the preset manganese sulfate solution supersaturation calculation model to calculate the candidate supersaturation value S for the current period. calc The supersaturation calculation model is constructed based on the functional relationship between manganese sulfate solubility and temperature and the definition of supersaturation. S104. The system has a preset density stability threshold Δρ th and temperature stability threshold ΔT th Determine whether the data in the current buffer satisfies the stability condition: σ ρ ≤Δρ th And σ T ≤ΔT th ; S105. If step S104 determines that the stability condition is met, then S... calc As an effective current supersaturation value S eff The output is used for subsequent closed-loop control; simultaneously, this S... eff The value is recorded in a queue of historical valid values ​​for oversaturation; S106. If step S104 determines that the stability condition is not met, then the system abandons the use of S. calc Instead, the most recently recorded valid value is retrieved from the queue of historical valid values ​​of oversaturation as the S value for this control cycle. eff Output; if the history queue is empty, maintain the output value of the previous control cycle.

[0025] To address the issue of feedback signal quality at its source, this invention incorporates specific data processing logic into the intelligent temperature control system: the system first stores continuously collected density and temperature data in a first-in-first-out buffer and calculates its moving average and moving standard deviation in real time; only the moving average is used for oversaturation calculation to filter out noise, while an online data stability criterion is constructed by comparing the moving standard deviation with a preset threshold. When the data is stable, the newly calculated value is adopted; when the data fluctuation is too large, it automatically switches to using the most recent historical valid value for maintenance. This invention embeds an adaptive filtering and decision-making module based on statistical process control principles at the very front of the control loop. It utilizes the statistical characteristics of the data (mean reflects state, standard deviation reflects reliability) to dynamically identify and eliminate unreliable instantaneous disturbance information. This invention provides a smooth, reliable, and interference-resistant oversaturation feedback signal for the intelligent temperature control system from the source, fundamentally avoiding control oscillations caused by measurement noise. This ensures that all subsequent advanced control and optimization logic can be executed smoothly on a stable feedback basis, thereby consolidating the foundation for the overall method's control stability, product consistency, and operational efficiency.

[0026] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Detailed Implementation

[0027] The present invention will now be described in further detail so that those skilled in the art can implement it based on the description.

[0028] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0029] It is important to note that the solubility of manganese sulfate increases with increasing temperature. Therefore, the adjustment logic of the intelligent temperature control system is based on the following principle: when it is necessary to increase the supersaturation of the solution, the solubility must be reduced by decreasing the evaporation temperature; when it is necessary to reduce the supersaturation, the solubility must be increased by increasing the evaporation temperature. Subsequent descriptions of the control actions will be based on this principle.

[0030] This invention aims to solve the technical challenge of unstable supersaturation of manganese sulfate solution due to temperature control lag and fluctuations in the MVR concentration process, which in turn affects the uniformity of the final crystal particle size. The closest existing technology typically uses a mechanical vapor recompression system to evaporate and concentrate the manganese sulfate solution, and monitors the temperature inside the evaporator, using a conventional PID controller to adjust the compressor speed or heater power to stabilize the evaporation temperature. However, this method only controls the intermediate process parameter—temperature—rather than the core parameter that directly determines crystal quality—supersaturation. Furthermore, the system has high thermal inertia and response delay, making it difficult to compensate for rapid process disturbances. This often leads to oscillations in temperature and supersaturation around the set values, making it impossible to maintain a precise and long-term range suitable for crystallization within a narrow range.

[0031] To address the aforementioned shortcomings, the specific implementation of this invention is as follows: First, manganese sulfate meeting industrial raw material standards is dissolved in water to prepare a raw material solution. This solution is then passed through a filter equipped with a micron-sized filter element to effectively remove suspended impurities, resulting in a clear manganese sulfate solution. Subsequently, the clear solution is smoothly fed into an MVR concentration system via a feed pump. This system includes an evaporator, a steam compressor, an auxiliary electric heater, and a condenser. An online vibratory density meter and a high-precision platinum resistance temperature sensor are installed in parallel on the pipe at the outlet of the evaporator concentrate to continuously and in real-time monitor the density and temperature values ​​of the manganese sulfate solution leaving the evaporator. These two measurement signals are transmitted at high speed via cable to the central processing unit of the intelligent temperature control system. The system has pre-set solubility-temperature relationship data for manganese sulfate and a density-concentration calibration curve. Within each fixed control cycle, the intelligent temperature control system first calculates the current concentration of the solution based on the real-time density value using the calibration curve, and simultaneously queries the equilibrium solubility of manganese sulfate at that temperature based on the real-time temperature value. Based on the definition of supersaturation, i.e., the ratio of the current concentration to the equilibrium solubility, the system instantly calculates the real-time supersaturation value of the solution.

[0032] Next, the intelligent temperature control system compares the calculated real-time supersaturation with the preset optimal range of 1.05-1.20. Based on the comparison results, the system generates specific control commands to adjust the compressor's operating frequency and the heater's input power in a coordinated and dynamic manner, thereby precisely controlling the evaporator's evaporation temperature. The ultimate goal is to stably maintain the solution's supersaturation within the target range of 1.05-1.20. During the coordinated adjustment of the equipment, the system follows a predetermined priority control logic: When the calculated supersaturation is below the target range (<1.05), indicating a need to increase the supersaturation to the target range, the system determines that the evaporation temperature needs to be lowered based on the principle that the solubility of manganese sulfate increases with temperature. Control commands are not simultaneously issued to the compressor and heater. The system prioritizes issuing commands to reduce the operating frequency of the steam compressor (the compressor lowers the solution boiling point by reducing system pressure, i.e., lowering the evaporation temperature; its response inertia is large, requiring priority for trend control). The system continuously monitors the feedback from the temperature sensor. Once the temperature decrease trend is clear and reaches a temporarily stable plateau, it indicates that the compressor's trend control is essentially in place. At this point, the system, based on the overall system energy efficiency model, gradually reduces the input power of the electric heater within the preset safe power range (the heater has a fast response speed and low inertia, used for precise calibration and compensation of the evaporation temperature). This collaboratively maintains the lowered evaporation temperature, preventing a temperature rebound due to system thermal inertia or disturbances, thereby precisely stabilizing the evaporation temperature near the target value and allowing the supersaturation to rise smoothly to the range of 1.05-1.20.

[0033] Conversely, when the online calculated supersaturation is higher than the target range (>1.20), meaning the supersaturation needs to be reduced to the target range, the system determines that the evaporation temperature needs to be increased. It then prioritizes issuing a command to increase the input power of the electric heater. Since the heater directly affects the heating intensity of the liquid in the evaporator by changing the heat flux, its response speed is fast and its inertia is low, allowing the evaporation temperature to rise rapidly. Once the temperature rise trend is obvious and tends to a temporarily stable plateau, it indicates that the rapid coarse adjustment of the heater is basically in place. At this point, the system, based on the overall system energy efficiency model, gradually increases the operating frequency of the steam compressor within a preset safe frequency range, completing fine matching, and ultimately stabilizing the evaporation temperature near the higher target value, causing the supersaturation to smoothly decrease to the range of 1.05-1.20.

[0034] Under the protection of this intelligent temperature control system, the MVR concentration process continues. When the online density meter reading indicates that the solution concentration has reached the preset endpoint value, such as 320 g / L, the concentration stage is marked as complete. At this point, the concentrated manganese sulfate solution is sent to the next process—the crystallizer. In the crystallizer, the programmable control system cools the solution at a uniform rate of approximately 1.0 °C / min, while the agitator maintains stirring at approximately 100 rpm, creating a uniform environment for crystallization. As the temperature decreases, supersaturation is gradually released, inducing the orderly nucleation and growth of manganese sulfate crystals. After the crystallization process is completed, the reaction slurry is sent to a solid-liquid separation device (such as a centrifuge). The separated solids undergo subsequent washing and drying to finally obtain the target product—battery-grade manganese sulfate crystals. Throughout the adjustment process, to ensure that the large inertia element does not cause oscillation, the system strictly stipulates that the amplitude of a single frequency adjustment of the compressor shall not exceed 10% of its current operating value, and there must be a time interval of at least 30 seconds between any two adjustment commands to the compressor, so that the system can fully respond and reach a new dynamic equilibrium.

[0035] In an optional embodiment of the present invention, after completing the MVR concentration and obtaining a manganese sulfate concentrate with a concentration of approximately 320 g / L, the high-temperature concentrate needs to be transported to a separately set crystallization section. During the transport process, if the temperature of the concentrate drops significantly, its supersaturation will increase uncontrollably, which can easily lead to premature crystallization on the inner wall of the pipeline, forming fine crystals and scale. This may not only block the pipeline and affect continuous production, but also cause inconsistencies in the initial state of the material entering the crystallizer, thereby impairing the crystal uniformity of the final product.

[0036] To address the inherent problems in this transport process, this embodiment employs an integrated active heat compensation insulation transport scheme. Specifically, a transport pipeline externally wrapped with high-efficiency insulation material connects the evaporator concentrate outlet to the crystallizer inlet. A tubular heat exchanger is installed in series on this pipeline, forming the core of the insulation system. The heat source is cleverly designed to utilize the excess heat energy generated by the MVR system itself during operation: a small stream of high-temperature, high-pressure secondary steam drawn from the compressor outlet is used as a heat medium branch flow and introduced into the shell side of the tubular heat exchanger. Simultaneously, the manganese sulfate concentrate to be transported, with a temperature between 70-85°C, is pumped through the tube side of the heat exchanger. By precisely controlling the flow rate of this heat medium steam entering the shell side through a regulating valve, the heat loss to the environment that is inevitable as the concentrate flows through the subsequent insulation pipeline can be dynamically compensated. This design aims to control the temperature fluctuation of the concentrate during transport from the evaporator outlet to the crystallizer inlet within a target range of ±2°C, thereby maximizing the stability of its supersaturated state.

[0037] Current technologies for handling concentrate transportation typically employ only passive insulation to wrap the pipeline, or at most, a separate, external heat tracing system. Passive insulation can only slow down the cooling rate and cannot prevent temperature drop over long distances or in low ambient temperatures; while external heat tracing requires additional energy and increases system complexity. None of these methods can achieve accurate, real-time heat loss compensation coupled with the main process. Therefore, the temperature and supersaturation state of the concentrate before entering the crystallizer cannot be reliably guaranteed, and the risks of scaling in the transportation pipeline and uneven product particle size remain.

[0038] This embodiment, through the above-mentioned active heat preservation and conveying design integrated into the process flow, ingeniously utilizes the system's own process waste heat without requiring additional significant energy consumption. This successfully solves the problem of maintaining the temperature of the concentrate during the transfer process, providing raw materials with stable and uniform physical properties for subsequent crystallization processes. It also ensures that the established precise supersaturation control effect can continue until the beginning of the crystallization stage.

[0039] In an optional embodiment of the invention, after the manganese sulfate concentrate with a temperature fluctuation strictly controlled within ±2℃ and a concentration of approximately 320 g / L is smoothly transported to near the crystallizer inlet via a heat-insulated conveying system, the invention further installs a static mixer of a specific structure in series in the pipeline at the end of the heat-insulated conveying pipeline, immediately before the feed inlet of the crystallizer. The mixer's shell is a straight pipe section, inside which multiple mixing units are fixedly installed, arranged sequentially and closely along the fluid flow direction. Each mixing unit is made of corrosion-resistant stainless steel sheet and precisely machined into a spiral shape, with each unit having a 180° axial twist angle. Crucially, the spiral twist directions of adjacent mixing units are set to be opposite to each other. When the macroscopically homogeneous manganese sulfate concentrate flows through this section of the pipeline after heat insulation, the liquid flow sequentially passes through these spiral units that alternately change rotation directions. The fluid is given a directional rotation and split when flowing through the first right-handed spiral unit, and then its flow direction is forcibly twisted to the opposite direction and split and recombined again in the next left-handed spiral unit. This process is repeated, and without the need for any external moving parts, the fluid undergoes sufficient cutting, shearing, rotation, and re-fusion on the cross-section of the pipe. This process effectively breaks down the uneven fluid velocity distribution caused by pipe wall friction (faster flow velocity at the center of the pipe and slower flow velocity near the wall) and the secondary flow that may be formed due to pipe deflection. This effectively weakens the radial temperature and concentration gradient that may be generated by the pipe flow, and significantly improves the uniformity of the concentrate entering the crystallizer at the microscale.

[0040] Existing technologies typically employ limited measures to address homogenization issues at the end of such conveying processes: they may rely on large agitators within the crystallizer itself for macroscopic mixing, but this fails to eliminate the initial state differences of the material upon entering the crystallizer; or they may add simple baffles to the pipeline, but these have limited mixing effects and may increase resistance. None of these methods offer a dedicated, efficient, and power-free intervention for the inherent radial property gradients in pipeline flows, which exist in both laminar and turbulent flows.

[0041] This technical solution, by implementing the aforementioned end-of-phase homogenization scheme based on a static mixer with a specific structure, further addresses the issue of microscopic uniformity in the final stage before the material enters the crystallization equipment, building upon the already completed macroscopic heat preservation. This ensures that each stream of concentrate entering different regions of the crystallizer almost simultaneously has a highly consistent initial supersaturation value, providing a crucial physical prerequisite for the synchronous and uniform generation of crystal nuclei during subsequent cooling and crystallization. This, in turn, guarantees a higher degree of concentration in the final product's crystal particle size distribution.

[0042] In an optional embodiment of the present invention, after the homogeneous concentrate is stably fed into the crystallizer, the core of maintaining precise control of the evaporation temperature within the MVR system lies in the execution strategy of the intelligent temperature control system. Specifically, this system is configured with the following collaborative control logic based on dynamic characteristic recognition: When the calculated supersaturation is below the target range (<1.05), to raise it to the target range, the system, based on the principle that the solubility of manganese sulfate increases with temperature, determines that the evaporation temperature needs to be lowered (lowering the temperature reduces solubility, thus increasing supersaturation). Control commands are not simultaneously sent to the compressor and heater. The system prioritizes issuing commands to reduce the operating frequency of the steam compressor (the compressor lowers the solution's boiling point by reducing system pressure, i.e., lowering the evaporation temperature; its response inertia is large, so trend control needs to be initiated first). After the temperature sensor clearly reports a decrease and stabilizes, the system then correspondingly reduces the input power of the electric heater (the heater has a fast response speed and low inertia, used for precise calibration and compensation of the evaporation temperature) to maintain the lowered evaporation temperature, preventing temperature rebound and thus precisely stabilizing the evaporation temperature near the target value, allowing the supersaturation to rise smoothly to the range of 1.05-1.20. The compressor's method of changing the solution's boiling point by lowering system pressure is a process with high inertia and a slow response.

[0043] Conversely, when the calculated supersaturation is higher than the target range (>1.20), the system determines that the evaporation temperature needs to be increased to reduce it to the target range (increasing the temperature increases solubility, thereby reducing supersaturation). The system then prioritizes issuing a command to increase the input power of the electric heater. Since the heater directly affects the heating intensity of the liquid in the evaporator by changing the heat flux, its response speed is fast and its inertia is small, allowing the evaporation temperature to rise rapidly. Once the temperature rise trend is obvious and tends to a temporarily stable plateau, it indicates that the rapid coarse adjustment of the heater is basically in place. At this point, the system, based on the overall system energy efficiency model, gradually increases the operating frequency of the steam compressor within a preset safe frequency range at a gentle pace to complete fine matching, ultimately stabilizing the evaporation temperature near the higher target value, and causing the supersaturation to steadily decrease to the range of 1.05-1.20. Throughout the adjustment process, to ensure that the large inertia element does not cause oscillation, the system strictly stipulates that the amplitude of a single frequency adjustment of the compressor shall not exceed 10% of its current operating value, and there must be a time interval of at least 30 seconds between any two adjustment commands to the compressor, so that the system can fully respond and reach a new dynamic equilibrium.

[0044] Current technologies often employ simple parallel PID control or master-slave control when dealing with the need to jointly regulate multiple actuators, but they fail to deeply distinguish the essential differences in the dynamic characteristics of different actuators. This can easily lead to commands from fast-responding heaters and slow-responding compressors "chasing" or "cancelling" each other, resulting in continuous temperature fluctuations and overshoot, which not only wastes energy but also undermines process stability.

[0045] This technical solution effectively resolves the conflict problem in multi-actuator coordination by implementing the aforementioned hierarchical, step-by-step, and gradually changing collaborative control logic. This strategy allows faster-responding devices to handle rapid following and coarse-tuning tasks, while slower-responding devices are responsible for subsequent fine-tuning and steady-state maintenance, using strict amplitude and time constraints to "tame" large-inertia components. This ensures that in the dynamic process of pursuing supersaturation stability, the evaporation temperature can approach and remain at the target value smoothly and without overshoot, thus providing a reliable control foundation for the continuous, stable, and efficient operation of the entire crystallization process.

[0046] In an optional embodiment of the present invention, after achieving stable regulation of the evaporation temperature based on the established collaborative control logic, in order to further guide the system's operating state towards a better energy efficiency level, the present invention is systematically implemented under the process condition that the MVR concentration system can stably maintain the target supersaturation. The system utilizes its comprehensive data acquisition capabilities to continuously collect and record multiple sets of historical operating data at several different but stable operating points (e.g., by actively making small adjustments to the feed flow rate, initial concentration, or ambient cooling water temperature to create differentiated operating conditions). Each set of data includes multiple dimensions of information reflecting the complete state of the system: including direct control variables, such as the compressor's operating frequency and the real-time power of the electric heater; key process state variables, such as the density and temperature of the manganese sulfate solution at the evaporator outlet measured by an online density meter and temperature sensor; core parameters reflecting the system's thermodynamic state, such as the pressure and temperature of the secondary steam at the compressor inlet and outlet; and finally, the total electrical power consumption of the system under that operating condition is summarized.

[0047] After obtaining a series of historical datasets covering different operating conditions, the core work of modeling lies in transforming these data into a mathematical model that can predict and optimize system energy efficiency. Its construction is not a simple data fitting process, but rather establishes the system's energy budget framework based on the first law of thermodynamics (energy conservation) and the fundamental principles of heat transfer. For example, when constructing the evaporator energy balance equation, key energy terms such as the heat of solution concentration, latent heat of vapor, and changes in sensible heat must be considered. Specifically, with minimizing the total electrical power of the system as the explicit optimization objective, the modeling process simultaneously introduces multiple constraints reflecting the inherent performance of the equipment: for example, the compressor efficiency model describes the relationship between its power consumption and the increase in steam pressure and temperature; the heater heat transfer model relates its input power to the effective heat transferred to the solution; and the evaporator's material and energy balance equations characterize the mass and heat balance during the solution concentration process. Under the joint constraints of these physical laws and equipment characteristic equations, regression analysis and other methods are used to fit the historical data, ultimately obtaining a steady-state energy efficiency regression model that quantitatively describes the relationship between "total system electrical power" and several key variables such as "compressor frequency," "heater power," and "solution evaporation temperature." The model is essentially a data-calibrated steady-state mapping of a "digital twin" that incorporates the physical properties of the system.

[0048] In seeking energy savings in MVR systems, current technologies often rely on rough estimations based on the rated efficiency curves of individual devices or simple statistical correlations to find the relationship between individual parameters and energy consumption. These methods fail to model the compressor, heater, evaporator, and process materials as a unified, coupled system, neglecting the complex interactions between devices and the nonlinear impact of operating condition changes on overall efficiency. Therefore, the energy-saving operations they guide are often one-sided and have limited accuracy.

[0049] By implementing the aforementioned systematic modeling method based on the coupling of multi-dimensional historical data and physical laws, this invention equips the intelligent temperature control system with a steady-state model that profoundly reflects the system's inherent energy efficiency characteristics. This goes beyond simply maintaining process stability; it provides a reliable and quantitative basis for intelligent operation that actively seeks and approaches the globally optimal energy consumption point while meeting crystallization process requirements, thus fulfilling the invention's objective of high efficiency and energy saving at a deeper level.

[0050] In an optional embodiment of the present invention, after successfully establishing the overall system energy efficiency model, the key step in achieving energy efficiency optimization is to apply this model safely, effectively, and without interfering with production stability to real-time control. The present invention intelligently integrates the offline optimization capability of the model with online dynamic control logic. In actual operation, the intelligent temperature control system first performs a basic calculation: it reads the density and temperature measurements of the manganese sulfate solution at the evaporator outlet in real time, and immediately, based on the preset supersaturation target range of 1.05-1.20 (the most stringent process requirement), it calculates the target evaporation temperature range required to maintain this supersaturation at the current solution concentration. This range directly reflects the process requirements. Subsequently, the system inputs this target temperature (usually the midpoint of the range as the setpoint), along with the pressure and temperature of the secondary steam at the compressor inlet—two key parameters reflecting the system's immediate thermodynamic state—into the established steady-state energy efficiency regression model.

[0051] The model is then "awakened," and its complex internal relationships are calculated instantly. Based on the input target evaporation temperature and real-time steam state, it optimizes and outputs a set of optimal operating parameters that theoretically minimizes the total power consumption of the entire MVR system under the current specific operating conditions, drawing from a vast amount of historical experience and physical laws. This set of parameters explicitly provides a recommended compressor frequency setpoint and a recommended heater power setpoint. However, the intelligent temperature control system does not immediately and rigidly apply these two recommended values ​​to the currently operating parameters. It incorporates a deliberate decision-making process: the system first compares the optimal compressor frequency setpoint recommended by the model with the compressor's actual operating frequency at the current moment, calculating the absolute deviation between the two. This deviation is then compared with a preset "compressor frequency adjustment threshold" based on process safety and equipment protection considerations.

[0052] Based on the comparison results, the system executes a branching strategy: if the calculated absolute value of the deviation exceeds twice the allowable adjustment threshold, it indicates a significant difference between the model-recommended "optimal point" and the system's current operating point, and a forced rapid switch risks causing drastic process fluctuations. In this case, the system temporarily ignores the model's output and fully relinquishes control to the collaborative control logic, which prioritizes stability, ensuring smooth process operation. Conversely, if the deviation falls within the allowable adjustment threshold, it means the model-recommended optimization path is gentle and safe. In this situation, the system continues to execute the collaborative control logic (e.g., prioritizing heater adjustment when heating is required), using the model's recommended compressor frequency setpoint as an ideal, gradually approached final target value. Subsequent adjustments to the compressor will be small, interval-based, gradual approaches towards this more energy-efficient target value, until the system operates at a steady-state point with lower energy consumption while maintaining process stability.

[0053] Existing technologies often exhibit two extremes when applying optimization models: one is "strong coupling," where the control system is forced to strictly track the model output, but this can easily lead to production accidents when the model is inaccurate or the operating conditions change drastically; the other is "weak coupling" or "offline optimization," where the model is only used to guide operators to make manual adjustments periodically, and cannot achieve real-time, dynamic energy efficiency tracking, resulting in delayed optimization and limited effectiveness.

[0054] By implementing the aforementioned model application mechanism that integrates real-time judgment and safety thresholds, this invention creatively resolves the potential contradiction between optimization objectives and stable control. It transforms the high-level energy efficiency model from an isolated computational tool into an "intelligent navigator" capable of assessing the situation and safely guiding the underlying control loop towards more economical operation, thereby achieving a closed-loop and practical implementation of energy efficiency optimization without sacrificing the robustness of the production process.

[0055] In an optional embodiment of the present invention, after the established mechanism safely applies the steady-state energy efficiency model to real-time optimization, the present invention further considers and solves the "aging" or "inaccuracy" problems that the model may face during long-term operation. The present invention endows the intelligent temperature control system with the ability to self-monitor and update the model performance online. During continuous system operation, a background process is activated. It does not interfere with normal control logic but silently performs a key task: continuously recording and comparing the "optimal compressor frequency setpoint" output by the steady-state energy efficiency regression model with the frequency value that the compressor ultimately maintains after the system has been running stably for a period of time. Simultaneously, it continuously compares the actual measured value of the system's total electrical power with the theoretical minimum electrical power value predicted by the model under the corresponding operating conditions.

[0056] These comparisons are not simply records, but form the basis for diagnosing the model's health status. The system pre-determines a model evaluation period (e.g., continuous operation for 100-500 hours) and sets clear quantitative mismatch criteria. When the cumulative duration of the system within such an evaluation period exceeds 20% of the total evaluation period, a diagnostic alarm is triggered: Scenario 1, the absolute deviation between the optimal frequency recommended by the model and the actual stable operating frequency of the system continuously exceeds the safe threshold for compressor frequency adjustment (e.g., 5% of the current frequency); Scenario 2, after adjustment based on the model's recommended values, the total actual power consumption of the system continuously exceeds the minimum power predicted by the model by a significant percentage (e.g., a pre-determined percentage of 5%-10%). Once either criterion is met, the system determines that the current steady-state energy efficiency regression model has mismatched with the actual operating state, and the accuracy of its optimization guidance has decreased.

[0057] Upon detecting a mismatch, the system doesn't simply issue an alarm; instead, it automatically initiates a structured online model self-updating program. This program first coordinates the control logic to ensure the MVR system operates stably near the dominant operating condition triggered by the mismatch. Based on this stability, the system enters a model parameter exploration mode. The total runtime of this mode can be set as needed (e.g., 8-16 hours), with the goal of accumulating a sufficient number (e.g., 20-30 sets) of steady-state valid data at different operating points. The actual effective time for parameter adjustment and data acquisition only accounts for a portion of this; the system spends most of its time in the steady-state maintenance phase to ensure production continuity. During this period, the system is allowed, within the framework of collaborative logic, to proactively and slightly adjust the compressor frequency and heater power, allowing the operating point to fluctuate within a safe window around the current operating point. The purpose of this adjustment is not control, but rather to purposefully "explore" the energy efficiency response of the actual system under different micro-operating conditions. The system synchronously and frequently collects new sets of operating data and their corresponding true total electrical power data generated at these exploration points; these data carry the most accurate characteristic information of the system at present.

[0058] Once a sufficient number of sets of fresh data covering the current operating conditions have been accumulated, the self-updating program enters its second phase. The system invokes its modeling algorithm, but this time it doesn't start from scratch. Instead, it uses the newly collected dataset, representing the current state of the system, to refit and refresh the parameters of the original steady-state energy efficiency regression model. This process is equivalent to using the latest "experience" to correct and calibrate the original "knowledge base." After the model parameters are updated, the program enters its final phase: the system smoothly switches the reference source for optimization guidance from the old, mismatched model to the new, updated model, and immediately begins using the new model for subsequent online energy efficiency optimization and operational guidance.

[0059] In existing technologies, most steady-state models used for optimization remain fixed after system debugging, or can only be manually updated periodically by engineers offline. This approach cannot detect the gradual deviation between the model and the actual performance of the equipment in real time, nor can it autonomously evolve the model without stopping the system or significantly interfering with production, leading to a gradual deviation of long-term operating energy efficiency from the optimal state.

[0060] By implementing the aforementioned closed-loop model self-updating procedure, which includes continuous monitoring, quantitative diagnosis, and controlled relearning, this invention endows the system with an "up-to-date" intelligence. It transforms energy efficiency optimization from a "one-off" action based on a static model into a dynamic, continuous, and self-improving intelligent process capable of adaptively tracking real-world factors such as equipment aging, scaling, and slow changes in raw material characteristics, fundamentally ensuring the sustainability and reliability of energy efficiency optimization throughout its entire lifecycle.

[0061] In an optional embodiment of the present invention, after the model's online self-updating program is initiated according to the aforementioned method and the system enters the stage of actively collecting new data, how to maximize the continuous stability of mainline production and product quality while exploring necessary parameters, and ensuring that the collected data can truly reflect steady-state characteristics, becomes an operational problem that requires meticulous design. To solve this problem, a specific embodiment of the present invention designs a structured parameter exploration mode for the system. When the system enters the data acquisition period, the intelligent control system will temporarily switch to this dedicated mode. The core operating mechanism of this mode is to alternately execute two distinctly different phases according to a preset time period (e.g., 30-60 minutes / cycle): a parameter adjustment phase and a steady-state maintenance phase.

[0062] Within each cycle, the parameter adjustment phase is executed first. The stable operating condition established at the end of the steady-state maintenance phase of the previous cycle is used as the baseline for this round of exploration. Based on a pre-designed orthogonal experimental table—which contains multiple combinations of operating parameters with finite, controllable deviations from the baseline (e.g., compressor frequency adjustment within ±(1.0~1.5) times the allowable frequency adjustment threshold, and heater power adjustment within ±(3%~5%) of the current power)—the system sequentially adjusts the compressor frequency and heater power to an operating point specified in the table. At each such operating point, the system does not pause briefly but maintains stable operation of that set of parameters for a preset duration (e.g., 5-15 minutes) to allow the process state to initially stabilize, and during this period, a complete set of operating data, including the total electrical power of the system, is collected. To ensure absolute safety during the exploration process, strict limits are imposed on the adjustment range: the absolute value of the compressor frequency adjustment does not exceed 1.5 times its allowable frequency adjustment threshold, and the absolute value of the heater power adjustment does not exceed five percent of its current power value.

[0063] Following the parameter adjustment phase, the system immediately enters the steady-state maintenance phase. At this point, the system restores the operating parameters to the baseline point at the start of this exploration round. Subsequently, based on its core supersaturation control logic (i.e., a coordinated strategy prioritizing the adjustment of the heater or compressor), the system strives to restore the supersaturation of the manganese sulfate solution to the optimal process range of 1.05-1.20, and continues to operate in this state for a longer preset duration (e.g., 10-30 minutes). The duration of this steady-state maintenance phase is set to be no less than twice the duration of the previous parameter adjustment phase to ensure that the production system has sufficient time to fully recover from the exploratory disturbance and produce qualified products. This "exploration-recovery" cycle continues until all planned operating points have been visited according to the orthogonal experimental setup. After completion, the system exits the parameter exploration mode and uses multiple sets of new operating data collected during all parameter adjustment phases to refit and update the parameters of the steady-state energy efficiency regression model.

[0064] Existing technologies often employ simple and direct methods when collecting data for similar model updates, such as continuously varying parameters over a long period or making isolated, large-scale jumps in test points. This approach can easily lead to production conditions deviating from the optimal process window for extended periods, affecting the consistency of product quality during that time. Furthermore, the collected data points may be in transitional states and cannot accurately represent steady-state performance, thus reducing the quality and reliability of model updates.

[0065] By implementing the aforementioned structured exploration mode, which features periodic alternation, strict amplitude limits, and sufficient recovery periods, this invention cleverly balances the contradiction between "acquiring fresh data" and "maintaining stable production." It compresses data collection activities that could potentially disrupt production into discrete, short timeframes, while reserving the majority of runtime for a controlled and stable production process. This efficiently completes the model self-updating task while reliably ensuring the continuity, stability, and uniformity of product quality throughout the entire production process.

[0066] In an optional embodiment of the present invention, after adjusting the operating point according to the aforementioned structured parameter exploration mode, accurately determining whether the system has reached a new steady state, and thus collecting high-quality data that truly represents the steady-state performance of that operating point, becomes a key detail determining the success or failure of the model update. To solve this specific problem, the present invention introduces a steady-state data acquisition triggering mechanism based on real-time data statistical characteristics. When the system is driven to a new operating point specified by the orthogonal experimental table during the parameter exploration phase, this mechanism is activated. The system does not start a countdown and wait, but continuously performs a precise monitoring task: it continuously acquires the real-time temperature and density signals of the manganese sulfate solution through a sensor at the evaporator outlet, and dynamically calculates the moving average and moving standard deviation of the most recent dozens of sampled values ​​(e.g., 30-50) of each of these two parameters at fixed time intervals (e.g., 1-5 seconds).

[0067] The moving average reflects the central trend of the parameter within the current short time window, while the moving standard deviation quantifies its degree of fluctuation. The system pre-sets clear quasi-steady-state judgment conditions, including two aspects: First, the calculated temperature moving average and density moving average must each fall within a small allowable fluctuation range centered on the expected steady-state value of the operating point (this range can be set based on process control accuracy, such as allowable temperature fluctuation ±1.0℃, allowable density fluctuation ±1.5g / L); Second, the calculated temperature moving standard deviation and density moving standard deviation must simultaneously be less than or equal to their respective preset steady-state judgment thresholds (temperature standard deviation ≤0.15℃, density standard deviation ≤0.5 g / L), which means that process fluctuations have been suppressed to an extremely low level. Only when the above two conditions are met simultaneously does the system determine that the production operation has reached a "quasi-steady-state" that can be used for model updates. Once the judgment is established, the system automatically opens a data acquisition window with a pre-set duration, during which a complete set of operating data is collected and recorded at a constant frequency (e.g., 1 second / time) within this window period. The allowable fluctuation range can be set to ±0.5% of the expected steady-state value. The steady-state judgment threshold can be set according to the sensor noise level and the general requirements for process stability, such as a temperature standard deviation of less than 0.1℃ and a density standard deviation of less than 0.5 kg / m³. 3 .

[0068] If, after reaching a new operating point, the system waits for a preset maximum time limit (5-10 minutes) and the above conditions are still not met simultaneously, it indicates that the process may be unable to quickly stabilize at this operating point for various reasons, or is experiencing abnormal fluctuations. In this case, the system will rationally abandon data acquisition at that point, record a failure, and, according to the experimental table, safely jump to the next combination of operating parameters to be explored and continue executing the program, thus avoiding wasting time and resources on invalid or unreliable data points.

[0069] Existing technologies for determining system steady state typically rely on operator experience or simple fixed-duration delays, such as "start recording after running parameters for X minutes." This method lacks objective, quantitative criteria and cannot distinguish between situations where "the time has elapsed but the state is not yet stable" and "the state has stabilized ahead of schedule." It is also prone to collecting data containing transient information or fluctuation noise, resulting in insufficient "purity" and "representativeness" of the data used for model updates, directly affecting the prediction accuracy and optimization effect of the updated model.

[0070] This technical solution, by implementing the aforementioned triggering mechanism based on online multi-parameter statistical stability criteria, endows the data acquisition stage during the model update process with intelligent "eagle eyes." It ensures that every data point sent to the model training program is a reliable "snapshot" of the system reaching a realistic and stable operating state under specific operating conditions, thereby maximizing the accuracy of model parameter updates and the reliability of the generated new model from the data source level.

[0071] In an optional embodiment of the present invention, based on ensuring data acquisition quality through the aforementioned steady-state triggering mechanism, the present invention further improves the core tool for guiding parameter exploration—the orthogonal experimental table. Traditionally, such experimental tables are pre-designed statically and fixed in the system. However, the actual operating conditions, equipment status, and safety boundaries of the system may shift due to production batches, equipment wear, or changes in raw materials. Static experimental tables may no longer be applicable at this time, as the parameter combinations they contain may have approached or even exceeded the currently permissible safe operating window. Forced execution may lead to process instability or equipment risks. Simultaneously, the points in a fixed table may not provide the optimal data point distribution within the changed operating condition neighborhood, affecting update efficiency.

[0072] To address this issue, the embodiments of this invention endow the intelligent control system with the ability to dynamically generate orthogonal experimental tables. Whenever the system initiates the model parameter exploration mode, the generation program is activated. The program first locks the system's steady-state operating point at the current moment as the baseline operating condition for this exploration, and accurately records the compressor's operating frequency baseline value F0 and the heater's power baseline value P0 at this moment. Next, the program reads a set of real-time updated operating constraints from the system (the constraints are derived from the equipment's factory safety parameters plus real-time process monitoring values, such as compressor maximum frequency ≤ 50Hz and heater maximum power ≤ 100kW). These conditions are not fixed; they combine the equipment's physical safety operating limits (such as compressor frequency upper and lower limits, heater maximum allowable power) with dynamic requirements calculated in real-time to maintain process stability (such as the frequency adjustment range temporarily constrained to ensure that supersaturation does not exceed limits).

[0073] Based on the baseline values ​​(F0, P0) and these dynamic constraints, the system clearly defines a rectangular safe and feasible operating space Ω specific to this exploration on the two-dimensional plane formed by the compressor frequency and heater power. The boundaries of this space are precisely defined by the constraints: the feasible range on the frequency axis is [F0 - maximum allowable downward adjustment, F0 + maximum allowable upward adjustment], and the feasible range on the power axis is [P0 × (1 - allowable downward adjustment percentage), P0 × (1 + allowable upward adjustment percentage)]. This space Ω represents all possibilities for the system to safely and stably explore parameters at the current moment.

[0074] Within the clearly defined safety space Ω, the system employs orthogonal experimental design principles to automatically select m statistically representative operation points (e.g., m = 9-12). Two selection principles apply: first, these points must be distributed as evenly as possible within the Ω space to efficiently cover the entire explorable area; second, they must include the baseline operating point (F0, P0) itself to ensure the continuity of exploration and provide anchor points for the recovery phase. Finally, the system sorts and encodes these selected operation points according to an optimized path principle (usually starting from the baseline, exploring neighboring points first, then exploring more distant points, and finally returning to the baseline), dynamically generating an orthogonal experimental table specifically for this model parameter exploration, fully adapted to the current operating condition.

[0075] Existing technologies typically rely on static experimental tables pre-designed by engineers, either general or specific to a particular operating condition. When the actual operating point deviates from the design condition, operators often need to manually adjust based on experience or bear potential risks, lacking an experimental design capability that is tightly coupled with the real-time system state and automatically adapts to changes.

[0076] This technical solution, by implementing the aforementioned method of dynamically generating orthogonal experimental tables, elevates parameter exploration from a fixed, "pre-programmed" action to a context-aware intelligent activity. It ensures that each data acquisition experiment conducted to update the model is firmly confined to the currently absolutely safe operating area, while simultaneously optimizing data point placement. This improves both the safety of model updates and the efficiency of data acquisition, while also enhancing the overall self-updating program's broad adaptability and long-term reliability under different operating conditions.

[0077] It should be noted that the quasi-steady-state determination thresholds set in the steady-state triggering mechanism (such as temperature fluctuation ±0.2-0.5℃, temperature standard deviation ≤0.1℃, and density standard deviation ≤0.001 g / cm³) 3The parameters (etc.) need to be comprehensively determined based on the actual sensor measurement accuracy, the inherent fluctuation level of the process system, and the requirements for data stability in model updates. In addition to ensuring that the operating points are uniformly distributed within the safety space Ω, the design of the orthogonal experimental table must also consider the setting of its factor levels (i.e., the adjustment range of compressor frequency and heater power) to effectively distinguish the impact of changes in operating parameters on system energy efficiency (total electrical power), thereby providing sufficient and representative experimental data for parameter fitting of the steady-state energy efficiency regression model.

[0078] In one optional embodiment of the present invention, the real-time signals acquired by the density and temperature sensors inevitably contain measurement noise, instantaneous fluid turbulence, and slight fluctuations in the process itself. If such raw signals or their simply filtered results are directly used for supersaturation calculation, the resulting values ​​will exhibit irregular and rapid fluctuations. Using this as the core feedback signal will cause the control system to generate unnecessary and frequent oscillation adjustments, thereby disrupting process stability.

[0079] To address this fundamental data quality issue, the technical solution of this invention constructs a sophisticated data purification and decision-making logic within the data processing core of the intelligent control system. The system continuously receives raw data streams ρ(t) and T(t) from an online density meter and temperature sensor at a fixed high frequency (e.g., 10 times per second). These instantaneous values ​​are not used directly but are instead fed into two first-in-first-out (FIFO) data buffers—one for density and one for temperature. Each buffer retains N historical data points from the most recent continuous time period (e.g., the most recent 30 seconds), forming a dynamically updated data window.

[0080] During each core control cycle (e.g., once per second), the system performs real-time statistical analysis on the data clusters within the two buffers. It calculates the moving average ρ and moving standard deviation σ of the density data window. ρ Simultaneously calculate the moving average T and moving standard deviation σ of the temperature data window. T The moving averages ρ and T represent the macroscopic trend of the process state within that time period, effectively smoothing out random noise. Subsequently, the system substitutes these two trend values ​​ρ and T into a pre-defined supersaturation calculation model based on the physical properties of manganese sulfate to obtain a candidate supersaturation value S for the current period. calc .

[0081] However, the system does not unconditionally adopt S. calc It introduces an intelligent decision criterion based on the inherent stability of the data. The system presets a density stability threshold Δρ. th and temperature stability threshold ΔT th In each cycle, it determines whether the volatility of the current buffer data is acceptable, i.e., whether it simultaneously satisfies: σρ ≤Δρ th And σ T ≤ΔT th If this stability condition is met, it indicates that recent data is stable and reliable, and the system will then... calc The effective supersaturation value S for this period has been officially adopted. eff The output is used by the high-level control logic, and this S... eff The value is stored in a historical valid value queue for later use.

[0082] If the stability condition is not met, it indicates that the recent data fluctuations are too large (possibly due to strong interference), and the calculated S... calc Unreliable. The system will decisively abandon S. calc Instead, it retrieves the most recently recorded, reliable S value from the historical valid value queue. eff The value is used as the output of the current control cycle. If the queue is empty (such as during system startup), the output value of the previous control cycle is conservatively maintained. This mechanism ensures that when the signal is subjected to a brief but strong disturbance, the value fed back to the control loop will not jump erratically, but will maintain the previously known and reliable state.

[0083] Existing technologies typically employ low-pass filters or moving average filters with fixed time constants when processing sensor signals. While these methods can smooth out noise, they cannot distinguish between "normal process slow variations" and "abnormal noise disturbances." When interference occurs, the filtered signal will still be distorted or delayed, and they lack the intelligent decision-making capability to automatically switch to a backup reliable value when the data is unreliable.

[0084] By implementing the aforementioned data processing logic based on dynamic statistical feature analysis and intelligent conditional decision-making, this invention constructs an "adaptive filter" and a "decision firewall" at the signal input port of the control system. It not only filters out high-frequency random noise but also dynamically adjusts decisions based on the quality of the data itself, fundamentally eliminating the possibility of control oscillations caused by unreliable measurement signals. This lays the most solid data foundation for the stable and accurate operation of the entire multi-level intelligent control architecture.

[0085] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details.

Claims

1. A method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling, characterized in that, Includes the following steps: S1: The manganese sulfate raw material solution is pretreated by filtration to remove suspended impurities and obtain a clear manganese sulfate solution; S2: The clarified manganese sulfate solution is fed into the MVR concentration system, which includes an evaporator, a compressor, a heater and a condenser. An online density meter and a temperature sensor are installed at the outlet of the evaporator to monitor the density and temperature of the concentrated manganese sulfate solution in real time. S3: The density and temperature data monitored by the online density meter and temperature sensor are transmitted to the intelligent temperature control system in real time. The intelligent temperature control system calculates the supersaturation of the manganese sulfate solution based on the received density and temperature data. The intelligent temperature control system compares the calculated supersaturation with the preset supersaturation range of 1.05-1.

20. Based on the comparison result, a control command is generated to coordinately adjust the frequency of the compressor and the power of the heater to control the evaporation temperature so that the supersaturation of the manganese sulfate solution is maintained at 1.05-1.

20. When the concentration of the manganese sulfate solution reaches 300-350 g / L, the manganese sulfate solution concentrated by the MVR concentration system is sent to the crystallizer; in the crystallizer, cooling and crystallization are carried out under the conditions of cooling rate of 0.5-2 ℃ / min and stirring speed of 50-150 rpm. S4: After crystallization, solid-liquid separation is performed to obtain battery-grade manganese sulfate crystals.

2. The method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling according to claim 1, characterized in that, The concentrated manganese sulfate solution, which has reached a concentration of 300-350 g / L after being concentrated by the MVR concentration system, is transported from the outlet of the evaporator to the inlet of the crystallizer through an insulated conveying pipeline. A tubular heat exchanger is installed on the insulated conveying pipeline; a branch stream of high-temperature secondary steam from the compressor outlet is used as the heat medium and introduced into the shell side of the tubular heat exchanger; a concentrated manganese sulfate solution with a temperature of 70-85 ℃, which is to be conveyed to the crystallizer, is introduced into the tube side of the tubular heat exchanger. By adjusting the flow rate of the heat medium entering the shell side, the heat loss of the manganese sulfate concentrate in the insulated transport pipeline is compensated, so that the temperature fluctuation of the manganese sulfate concentrate is within ±2℃ throughout the entire process of transporting it from the evaporator outlet to the crystallizer inlet.

3. The method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling according to claim 2, characterized in that, A static mixer is installed in series near the inlet of the crystallizer in the insulated conveying pipeline. The static mixer has multiple spiral mixing units arranged sequentially along the flow direction. The mixing units are made of stainless steel sheets with a spiral angle of 180° and the spiral directions of adjacent mixing units are opposite. After the manganese sulfate concentrate flows through the static mixer, it directly enters the crystallizer.

4. The method for preparing battery-grade manganese sulfate using MVR concentration-intelligent temperature control coupling according to claim 1, characterized in that, The intelligent temperature control system adjusts its operating parameters according to the following collaborative control logic: Based on the comparison results between the supersaturation and the preset range, when it is determined that the evaporation temperature of the evaporator needs to be increased, the power of the heater is increased first. After the temperature fed back by the temperature sensor starts to rise and tends to stabilize, the frequency of the compressor is gradually increased within the preset safe frequency range of the compressor according to the overall energy efficiency model of the system. When it is necessary to reduce the evaporation temperature, the frequency of the compressor is reduced first. After the temperature sensor starts to drop and stabilizes, the power of the heater is then reduced. The single adjustment of the compressor's operating frequency shall not exceed 10% of its current frequency value, and the time interval between two adjacent frequency adjustments shall not be less than 30 seconds.

5. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 4, characterized in that, The method for establishing the overall energy efficiency model of the system is as follows: During the stable operation of the MVR concentration system, multiple sets of historical operating data under different operating conditions are collected and recorded. Each set of data includes: compressor frequency, heater power, real-time density and temperature of manganese sulfate solution at the evaporator outlet, steam pressure and temperature at the compressor inlet and outlet, and total electrical power of the system. Based on the first law of thermodynamics and the principle of heat transfer, with the goal of minimizing the total electrical power of the system, and with the compressor efficiency model, heater heat transfer model, and evaporator material and energy balance equation as constraints, regression fitting is performed on the historical operating data to obtain a steady-state energy efficiency regression model that describes the quantitative relationship between the total electrical power of the system and the compressor frequency, heater power, and solution evaporation temperature.

6. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 5, characterized in that, The application of the overall energy efficiency model of the system is as follows: During operation, the intelligent temperature control system obtains the density and temperature of the manganese sulfate solution at the outlet of the evaporator in real time, and deduces the target evaporation temperature range that needs to be maintained based on the preset supersaturation range of 1.05-1.

20. Subsequently, the current target evaporation temperature and the real-time collected compressor inlet secondary steam pressure and temperature are used as inputs and substituted into the steady-state energy efficiency regression model to calculate and output a set of optimal operating parameters that minimizes the total power of the system under the current operating conditions. The optimal operating parameter set includes recommended compressor frequency setting and heater power setting. In the control execution phase: when the intelligent temperature control system needs to adjust the operating parameters to maintain supersaturation, it first compares the calculated optimal compressor frequency setting with the current frequency. If the absolute value of the deviation between the two is greater than twice the allowable adjustment threshold of the compressor frequency, the model output is temporarily ignored, and the operation is performed only based on the collaborative control logic. If the deviation is within the threshold range, then during the execution of the control logic, the recommended value output by the model will be used as the final target value for compressor frequency adjustment.

7. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 6, characterized in that, The intelligent temperature control system continuously monitors and records the optimal compressor frequency setpoint output by the steady-state energy efficiency regression model and the compressor frequency value during actual stable operation of the system. If the system runs continuously for more than one preset model evaluation cycle, and any of the following situations occur continuously within that cycle and exceed 20% of the total running time, then the current steady-state energy efficiency regression model is determined to be mismatched with the actual operating state: (i) The absolute deviation between the optimal compressor frequency setpoint and the final stable compressor frequency value continuously exceeds the allowable adjustment threshold of the compressor frequency; (ii) After adjustment based on the model's recommended values, the actual measured value of the system's total electrical power continues to be higher than the minimum electrical power value predicted by the model by a preset percentage. After determining that the model is mismatched, the intelligent temperature control system initiates an online self-update program for the model, including the following steps: 1) Maintain the system in stable operation near the current dominant operating condition for at least one data acquisition period. During this period, actively and slightly adjust the compressor frequency and heater power to cover the operating window around the current operating point, and simultaneously collect multiple sets of new operating data and corresponding total system power data. 2) The parameters of the steady-state energy efficiency regression model are refitted and updated using the newly collected dataset to generate an updated steady-state energy efficiency regression model; 3) After the model update is complete, switch to using the updated model for online energy efficiency optimization and operation guidance.

8. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 7, characterized in that, In step 1) of the online self-updating procedure for the model, the specific execution method for actively and slightly adjusting the compressor frequency and heater power is as follows: During the data acquisition period, the intelligent temperature control system temporarily switches to model parameter exploration mode; the model parameter exploration mode alternately executes the parameter adjustment phase and the steady-state maintenance phase according to a preset cycle. The execution process for each cycle is as follows: a) During the parameter adjustment phase, using the operating condition established at the end of the previous steady-state maintenance phase as the baseline, and based on a pre-designed orthogonal experimental table, the compressor frequency and heater power of the system are sequentially adjusted to the operating points specified in each combination in the table; the orthogonal experimental table contains multiple sets of operating parameter combinations with finite deviation values ​​based on the baseline; at each operating point, the system maintains stable operation for a first preset duration, and a set of operating data containing the total electrical power of the system is collected; during this phase, the absolute value of the compressor frequency adjustment does not exceed 1.5 times the allowable adjustment threshold of the compressor frequency, and the absolute value of the heater power adjustment does not exceed 5% of its current power value; b) In the subsequent steady-state maintenance phase, the system restores the operating parameters to the reference point and, according to the control logic of the intelligent temperature control system, controls the supersaturation of the manganese sulfate solution to be maintained within the preset range of 1.05-1.20, and continues to run for a second preset duration; the second preset duration is not less than twice the first preset duration. After exploring all parameter combinations according to the orthogonal experimental table, the system exits the model parameter exploration mode and uses the new operational data collected during the all-parameter adjustment phase to perform model parameter refitting and updating in step 2).

9. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 8, characterized in that, In each parameter adjustment phase of the model parameter exploration mode, after the system is driven to an operating point specified by the orthogonal experimental table, the intelligent temperature control system executes a steady-state data acquisition trigger mechanism, specifically as follows: The system continuously monitors the real-time temperature and density of the manganese sulfate solution at the evaporator outlet, and calculates the moving average and moving standard deviation of its most recent N sampled values ​​at fixed time intervals. When both of the following conditions are met simultaneously, the system determines that the production operation has reached a quasi-steady state suitable for model updates, and automatically opens a data acquisition window with a duration of the first preset time: A) The moving average of the temperature and the moving average of the density fall within a preset allowable fluctuation range centered on the expected steady-state value of the operating point; B) The moving standard deviation of the temperature and the moving standard deviation of the density are both less than or equal to their respective steady-state determination thresholds; During the data acquisition window, the system acquires and records a complete set of operational data at a constant sampling frequency; If the above quasi-steady-state determination condition is not met after the maximum waiting time limit is reached, the system will abandon data acquisition at this operation point, record a data acquisition failure, and jump to the next combination of operation parameters to continue execution according to the orthogonal experimental table.

10. The method for preparing battery-grade manganese sulfate by MVR concentration-intelligent temperature control coupling according to claim 8 or 9, characterized in that, The pre-designed orthogonal experimental table is dynamically generated by the intelligent temperature control system based on the current operating conditions and real-time operational constraints each time the online self-updating program of the model is executed. The generation method includes the following steps: Step 1: When starting the model parameter exploration mode, take the current steady-state operating point of the system as the reference operating condition and determine the reference value F0 of the compressor frequency and the reference value P0 of the heater power at this time. Step two, read the real-time operation constraints pre-stored in the system, including: the maximum absolute value of the compressor frequency allowed to be raised ΔF max , the maximum absolute value of the allowed to be lowered ΔF min , and the maximum percentage of the heater power allowed to be raised α max , the maximum percentage of the allowed to be lowered α min ; wherein the constraints are dynamically set according to the safe operation limits of the equipment and the real-time requirements for maintaining process stability; Step 3: Based on the reference value (F0, P0) and the operational constraints, define a safe and feasible operating space Ω on the compressor frequency-heater power plane, permissible in this parameter exploration; the operating space Ω is a rectangular region with its frequency boundary being [F0-ΔF]. min , F0+ΔF max The power boundary is [P0×(1-α)]. min ), P0×(1+α max )]; Step 4: Within the operating space Ω, use orthogonal experimental design to select m statistically representative operating points; the selected operating points must satisfy the following conditions: they should be distributed as evenly as possible within the Ω space; and they should include the baseline operating point (F0, P0). Step 5: Encode and combine the m operation points selected in Step 4 and their preset adjustment order to generate an orthogonal experimental table specifically for this model parameter exploration; the setting of the adjustment order follows the path principle of starting from the reference point, exploring from near to far, and finally returning to the reference point, so as to reduce process disturbances during continuous switching of operation points.