Intelligent control system and method for mine water treatment based on adsorption-reduction model

Abstract

The application relates to the technical field of mine water treatment, and discloses a mine water treatment intelligent control system and method based on an adsorption-reduction model, which first acquires and standardizes sensor data to obtain a standardized water quality vector; then, based on a treatment target and an adsorption-reduction model instance, the vector is intelligently solved to determine a target reaction pH value. Further based on material adding amount and the model instance, the proton consumption rate in the reaction process is predicted in advance, and combined with the acquired real-time pH feedback signal, the target reaction pH value is subjected to combined closed-loop control of feedforward and feedback, and finally, an actuator instruction for driving acid-base adding is generated. In this way, by introducing a predictive feedforward path, the systematic pH disturbance caused by reaction consumption of protons can be actively compensated, the hysteresis of traditional pure feedback control is overcome, the accuracy and stability of pH control are significantly improved, and thus the treatment effect of mine water is efficiently ensured.

Description

Technical Field

[0001] This application relates to the field of mine water treatment technology, and more specifically, to an intelligent control system and method for mine water treatment based on an adsorption-reduction model. Background Technology

[0002] Hexavalent chromium (Cr(VI)) in mine water poses a serious threat to the ecological environment and human health due to its high toxicity, strong migration, and carcinogenicity, making it a key and challenging aspect of mine water treatment. Adsorption treatment using waste coal gangue is a promising low-cost solution, but its own reduction capacity for Cr(VI) is limited. Research has found that adding soluble organic matter such as fulvic acid (FA) to the system can construct an adsorption-reduction synergistic system, utilizing its functional groups to reduce highly toxic Cr(VI) to less toxic trivalent chromium (Cr(III)), thereby significantly improving treatment efficiency.

[0003] Despite the immense potential of the coal gangue-fulvic acid system, its practical application is limited by a key factor: the system's reaction efficiency is highly sensitive to pH, with acidic conditions (such as pH 2-4) being the optimal reaction range. The core technical challenge lies in the fact that the reduction of Cr(VI) is a process that continuously consumes protons (H⁺) in the system. This inevitably leads to a spontaneous and continuous rise in pH during the reaction, deviating from the optimal reaction range and causing a dynamic decrease and instability in processing efficiency.

[0004] Currently, pH control largely relies on traditional feedback control methods. This "post-treatment compensation" approach inherently suffers from lag, making it difficult to effectively address rapid and sustained systemic disturbances caused by the main reaction. This leads to frequent pH fluctuations, affecting not only the final treatment outcome but also increasing the ineffective consumption of reagents. Existing technologies fail to utilize reaction kinetic models to proactively predict proton consumption rates and integrate them into the control loop for proactive pH regulation. Therefore, developing an intelligent control method capable of accurately predicting and proactively compensating for endogenous disturbances in the system is a pressing technical challenge in this field. Summary of the Invention

[0005] To address the technical problems of the "coal gangue-fulvic acid" synergistic system in treating mine water containing hexavalent chromium (Cr(VI)), where the main reaction continuously consumes protons (H⁺), causing the pH value to spontaneously rise and deviate from the optimal reaction range, and the lag in traditional feedback control methods that cannot effectively suppress this endogenous system disturbance, this invention aims to provide an intelligent control method for mine water treatment based on an adsorption-reduction model, so as to achieve precise, stable, and proactive regulation of the pH value during the reaction process.

[0006] According to one aspect of this application, a smart control method for mine water treatment based on an adsorption-reduction model is provided, comprising: acquiring raw sensor data; standardizing the raw sensor data to obtain a standardized water quality vector; intelligently calculating target parameters of the standardized water quality vector based on the treatment target and an adsorption-reduction model instance to obtain a target reaction pH value; acquiring a real-time pH feedback signal; making a forward-looking prediction of the proton consumption rate based on the gangue dosage, fulvic acid dosage, current reaction time, and the adsorption-reduction model instance to obtain a predicted proton consumption rate; and performing closed-loop control of the reaction process based on the predicted proton consumption rate and the real-time pH feedback signal to obtain actuator commands.

[0007] According to another aspect of this application, a mine water treatment intelligent control system based on an adsorption-reduction model is provided, comprising: a raw data acquisition module for acquiring raw sensor data; a data standardization module for standardizing the raw sensor data to obtain a standardized water quality vector; a target pH intelligent calculation module for intelligently calculating target parameters of the standardized water quality vector based on the treatment target and an adsorption-reduction model instance to obtain the target reaction pH value; a real-time pH acquisition module for acquiring real-time pH feedback signals; a proton consumption rate prediction module for making a forward-looking prediction of the proton consumption rate based on the gangue dosage, fulvic acid dosage, current reaction time, and an adsorption-reduction model instance to obtain a predicted proton consumption rate; and a closed-loop control execution module for performing closed-loop control of the reaction process based on the predicted proton consumption rate and the real-time pH feedback signal to obtain actuator commands.

[0008] Compared with existing technologies, the intelligent control system and method for mine water treatment based on the adsorption-reduction model provided in this application firstly, by introducing model-based predictive feedforward control, can proactively compensate for the pH increase trend caused by the main reaction, eliminating the lag of traditional feedback control at its source, resulting in faster system response and more timely control. Secondly, this application can significantly suppress pH fluctuations and overshoot during the reaction process, keeping it more firmly stable within the optimal reaction range, thereby significantly improving the overall reduction efficiency of Cr(VI) and the stability of the treatment effect. Furthermore, precise and smooth control avoids reagent waste caused by repeated over-adjustment, effectively reducing operating costs. Finally, this composite control architecture enhances the system's robustness, enabling it to better adapt to changes in different initial water quality conditions, ensuring stable achievement of mine water treatment standards, and possessing significant engineering application value. Attached Figure Description

[0009] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a schematic flowchart of an intelligent control method for mine water treatment based on an adsorption-reduction model according to an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of infrared spectrum peak fitting for gangue.

[0012] Figure 3 This is a schematic diagram showing the effect of initial fulvic acid concentration on the adsorption-reduction of Cr(VI) in gangue.

[0013] Figure 4 This is a schematic diagram showing the effect of initial pH on the adsorption-reduction of Cr(VI) in gangue mediated by fulvic acid.

[0014] Figure 5 This is a schematic diagram illustrating the effect of the initial C:Fe molar ratio on the adsorption-reduction of Cr(VI) in gangue mediated by fulvic acid.

[0015] Figure 6 This is a schematic diagram of the fitting curve of the kinetics of Cr(VI) adsorption on gangue mediated by fulvic acid.

[0016] Figure 7 This is a schematic diagram of linear fitting of the intraparticle diffusion model of Cr(VI) adsorption on gangue mediated by fulvic acid.

[0017] Figure 8 This is a schematic diagram of the X-ray diffraction pattern of gangue.

[0018] Figure 9 This is a schematic diagram of a kaolinite structure ball-and-stick / polyhedron model.

[0019] Figure 10 This is a schematic diagram simulating the adsorption of Cr(VI) molecules by kaolinite mediated by fulvic acid.

[0020] Figure 11 This is a schematic diagram of the radial distribution function of Cr(VI) adsorption in kaolinite mediated by fulvic acid.

[0021] Figure 12 This is a schematic block diagram of an intelligent control system for mine water treatment based on an adsorption-reduction model according to an embodiment of this application. Detailed Implementation

[0022] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0023] Figure 1 This is a schematic flowchart illustrating an intelligent control method for mine water treatment based on an adsorption-reduction model according to an embodiment of this application. Figure 1 As shown, this application provides an intelligent control method for mine water treatment based on an adsorption-reduction model, including: S1, acquiring raw sensor data; S2, standardizing the raw sensor data to obtain a standardized water quality vector; S3, intelligently calculating target parameters on the standardized water quality vector based on the treatment target and an adsorption-reduction model instance to obtain the target reaction pH value; S4, acquiring real-time pH feedback signals; S5, making a forward-looking prediction of the proton consumption rate based on the gangue dosage, fulvic acid dosage, current reaction time, and the adsorption-reduction model instance to obtain a predicted proton consumption rate; S6, based on the predicted proton consumption rate, performing closed-loop control of the reaction process on the target reaction pH value and the real-time pH feedback signal to obtain actuator commands.

[0024] For example, in step S1, raw sensor data is acquired. It should be understood that the adsorption-reduction efficiency of hexavalent chromium in the gangue-fulvic acid system is significantly affected by various environmental factors. The reduction rate of hexavalent chromium is closely related to the initial fulvic acid concentration, pH value, and background iron concentration, and the reaction process is sensitive to temperature changes. To achieve precise closed-loop control based on a kinetic model, it is first necessary to acquire raw sensor data.

[0025] Specifically, acquiring raw sensor data is performed using a probe array deployed at the front end of the mine water treatment system (e.g., installed in pipelines or buffer tanks before entering the reactor). This probe array integrates multiple sensors for different water quality parameters. The control system acquires continuous or discrete electrical or digital signals from these sensors synchronously or time-divisionally at a preset sampling frequency through a data acquisition device. The acquired raw sensor data specifically includes: a current-voltage response signal representing hexavalent chromium concentration, a potential difference signal representing pH value, an ultraviolet absorbance signal representing fulvic acid concentration (a typical component of dissolved organic matter (DOM), a specific wavelength absorbance signal representing background Fe ion concentration that may affect the reaction pathway, and a resistance signal from a temperature sensor reflecting the system's reaction kinetics. These signals are unprocessed, raw electrical signals that directly reflect the physical or chemical responses of each sensor, collectively constituting a dataset immediately input to the next processing unit.

[0026] In one specific embodiment, a multi-parameter sensor probe array deployed on the mine water intake pipeline is activated. A central data acquisition module polls each sensor channel in the probe array sequentially once per second. At a certain sampling moment, the module acquires a current signal of 3.5 μA from a voltammetric Cr(VI) sensor, a potential difference signal of -59.16 mV from a pH electrode, an absorbance value of 0.8 representing fulvic acid concentration from an ultraviolet absorption spectroscopy sensor at a specific wavelength, an absorbance value of 0.2 representing Fe ion concentration from another specific wavelength absorption sensor, and a resistance value of 109.73 Ω from a PT100 thermistor sensor. This set of instantaneous electrical signals with a uniform timestamp, consisting of {3.5 μA, -59.16 mV, 0.8, 0.2, 109.73 Ω}, constitutes the raw sensor data at that moment. This data is then packaged and transmitted to a subsequent data standardization module for further processing.

[0027] For example, in step S2, the raw sensor data is standardized to obtain a standardized water quality vector. It should be understood that in actual mine water treatment scenarios, the removal efficiency of hexavalent chromium is complexly affected by various environmental factors such as fulvic acid concentration, pH value, and background iron concentration. Accurately obtaining the true physical values ​​of these key parameters is crucial for precise control. However, the data collected by the raw sensors is essentially analog electrical signals such as volt-ampere response current, potential difference, absorbance, or resistance. These signals not only have inconsistent physical dimensions but are also highly susceptible to electromagnetic noise interference such as pump start-up and shutdown in harsh industrial environments. Furthermore, different types of sensors exhibit significant differences in response time. Directly using these noisy, asynchronous, and meaningless raw data to drive the adsorption-reduction kinetic model will lead to distorted model parameter inversion, resulting in control command deviations. Therefore, data standardization is necessary to eliminate random noise, compensate for environmental drift, and transform heterogeneous signals into a precise, unified, and time-aligned set of physical parameters—the standardized water quality vector—providing a reliable data foundation for subsequent intelligent calculations.

[0028] In one embodiment, data standardization of the raw sensor data to obtain a standardized water quality vector includes: S21, digital filtering of the raw sensor data to obtain a denoised digital sequence set; S22, parameter inversion and calibration of the denoised digital sequence set based on a feature mapping model to obtain a standardized water quality vector, wherein the standardized water quality vector includes initial hexavalent chromium concentration, initial pH, initial fulvic acid concentration, initial Fe ion concentration and water temperature.

[0029] First, the raw sensor data is digitally filtered to obtain a denoised digital sequence set. Due to the complex environment in mines, sensor signals are susceptible to high-frequency noise caused by electromagnetic interference and other factors. Therefore, digital filtering algorithms need to be applied to the acquired raw sensor data sequences. For example, a moving average filtering method can be used. For the raw sample value of any sensor channel at time k, its arithmetic mean with the historical N-1 sample points is calculated to obtain a smoother, more representative filtered signal value, thus effectively filtering out random fluctuations and forming a denoised digital sequence set.

[0030] After obtaining the denoised digital sequence set, parameter inversion and calibration based on a feature mapping model are performed to obtain water quality parameters with practical physical meaning. This process converts the dimensionless or electrical unit-based pure digital signal output from the previous sub-step into standard physical quantity units through a pre-established mathematical model. Each parameter inversion corresponds to a specific feature mapping model, which is a functional relationship between the sensor response signal and the actual physical quantity established through regression analysis and other methods based on a large amount of experimental data. For example, the calculation of hexavalent chromium concentration specifically involves parameter inversion and calibration of the volt-ampere response current signal representing the hexavalent chromium concentration based on a feature mapping model using the following formula: in, The signal is the current-voltage response representing the concentration of hexavalent chromium after noise reduction processing. For sensor sensitivity coefficient, This is the temperature compensation coefficient, used to correct for the influence of water temperature changes on the measurement results. For water temperature, The zero-point intercept is used to eliminate system background bias, and 25 is the standard reference temperature. To calibrate the hexavalent chromium concentration. For the potential difference signal representing pH, the system uses an existing electrochemical model based on the Nernst Equation for inversion. This model describes the logarithmic linear relationship between the electrode potential difference and the hydrogen ion activity in the solution, and incorporates real-time temperature parameters to correct the slope, thus accurately converting the millivolt-level potential signal into a dimensionless pH value. For the ultraviolet absorbance signal representing fulvic acid concentration and the specific wavelength absorbance signal representing background Fe concentration, the system uses an existing spectral analysis model based on the Beer-Lambert Law for inversion. This model utilizes the principle that the degree of absorption (absorbance) of a substance to a specific wavelength of light is proportional to its concentration, combined with pre-determined molar absorptivity and optical path length, to directly convert the optical absorbance value into the corresponding fulvic acid concentration (mg·L⁻¹). -1) and Fe ion concentration (mol·L -1 For the resistance signal from the temperature sensor, the system calls upon existing metal resistance temperature characteristic models (such as the Callendar-Van Dusen equation or lookup table method) to calculate the current water temperature based on the strict monotonic correspondence between resistance value and temperature.

[0031] Finally, the individual parameters are fused and vectorized to form a standardized water quality vector. Since the response times of different sensors may vary, this process first aligns the multiple physical parameters output from the previous step, each with its own timestamp, to ensure that all parameters accurately reflect the water state at the same moment. Then, these time-aligned water quality parameters with standard physical units are arranged and combined according to a pre-defined data structure, encapsulating them into a single, structured data record—the standardized water quality vector.

[0032] In one specific embodiment, following the aforementioned steps, the system receives the raw sensor data set {3.5μA, -59.16mV, 0.8, 0.2, 109.73Ω}. First, the data set is processed into a smoother, denoised digital sequence, such as {3.48μA, -59.20mV, 0.79, 0.21, 109.73Ω}, using a moving average filter with a window size of 5. Subsequently, the system calls the feature mapping model for each parameter for inversion and calibration. Regarding hexavalent chromium concentration, in this embodiment, the sensor sensitivity coefficient... 10 mg·L -1 / μA, temperature compensation coefficient 0.1 mg·L -1 / °C, zero intercept -4.8 mg·L -1 And from the resistance value of 109.73Ω, the current water temperature is deduced to be 25°C. Therefore, the hexavalent chromium concentration is calculated as follows: =10×3.48+0.1×(25-25)+(-4.8)=30mg·L -1 Similarly, a potential difference of -59.20 mV was inverted to a pH value of 2.0, and an absorbance of 0.79 was inverted to a fulvic acid concentration of 100 mg·L⁻¹. -1 The absorbance of 0.21 was inverted to an Fe concentration of 0.05 mol·L⁻¹. -1 Finally, these calculation results are vectorized and encapsulated to obtain the final standardized water quality vector: {Initial hexavalent chromium concentration: 30 mg·L⁻¹} -1 Initial pH: 2.0, Initial fulvic acid concentration: 100 mg·L⁻¹ -1 Initial Fe ion concentration: 0.05 mol·L -1The vector, representing the water temperature at 25°C, is then transmitted to the target parameter intelligent calculation module.

[0033] For example, in step S3, based on the treatment target and the adsorption-reduction model instance, the target parameters of the standardized water quality vector are intelligently calculated to obtain the target reaction pH value. It should be understood that in the fulvic acid-mediated gangue adsorption-reduction system, pH is a key environmental factor determining the reaction kinetic rate. An acidic environment is conducive to the protonation of fulvic acid functional groups, thereby enhancing their electron-donating capacity and promoting the reduction of hexavalent chromium. However, excessively low pH values ​​lead to a surge in the consumption of acid regulators, increasing operating costs. Conversely, if the pH value is too high, not only will electrostatic repulsion be enhanced, but precipitation may also form covering active sites, resulting in substandard treatment. Therefore, it is necessary to use intelligent calculation to accurately calculate the target reaction pH value that balances treatment efficiency and economic cost, while meeting the rigid treatment target of a specific hexavalent chromium removal rate, comprehensively considering current water quality conditions such as hexavalent chromium concentration, fulvic acid concentration, and background iron concentration, in order to achieve dynamic matching of process parameters.

[0034] In one embodiment, based on the treatment objective and the adsorption-reduction model, the target parameters of the standardized water quality vector are intelligently calculated to obtain the target reaction pH value. This includes: constructing an adsorption-reduction model instance based on the standardized water quality vector; using the standardized water quality vector as the initial state boundary and a preset hexavalent chromium concentration removal rate as the treatment objective, performing multi-objective iterative optimization based on constraints on the adsorption-reduction model instance to obtain the optimal process parameter set, which includes the target reaction pH value and the optimal equilibrium adsorption amount.

[0035] First, based on the standardized water quality vector, a specific adsorption-reduction model instance for the current water quality is constructed. The first step in this process is to decouple the state vector from the standardized water quality vector and extract feature variables, separating a set of state variables that significantly influence reaction kinetics. This set includes at least the initial fulvic acid concentration, initial pH, and initial Fe ion concentration. Next, the system calls a multivariate regression model pre-trained and stored using a large amount of experimental data (e.g., data on the influence of initial fulvic acid concentration, pH, and Fe background values ​​on Cr(VI) adsorption-reduction efficiency) to perform kinetic coefficient inversion based on multivariate regression on this set of state variables. Specifically, the initial fulvic acid concentration, pH, and initial Fe ion concentration are used as input features, and the experimentally determined and fitted pseudo-second-order rate constant and equilibrium adsorption capacity are used as output labels. The data is fitted and trained using the least squares method, and the determined regression model structure is shown below: Where Y represents the predicted or , and These represent the normalized initial fulvic acid concentration, pH value, and initial Fe ion concentration, respectively. Represents the constant intercept term. Represents the linear effect coefficient. This represents the quadratic effect coefficient, used to describe the characteristics of nonlinear surfaces. Represents the interaction effect coefficient. This represents the random error term. During real-time operation, the system substitutes the real-time state variables obtained from the standardized water quality vector into the above regression model to calculate the pseudo-secondary adsorption rate constant for the current specific operating condition. ) and equilibrium adsorption capacity ( Finally, the two specific coefficient values ​​calculated for the current operating conditions are embedded into a general pseudo-second-order kinetic equation, thereby constructing a specific adsorption-reduction model instance capable of predicting the change of pollutant removal over time under different reaction conditions. This model instance can be represented by the following formula: in, The pseudo-second-order adsorption rate constant. To balance the adsorption capacity, This is the current reaction time. This represents the predicted instantaneous adsorption capacity at time t.

[0036] After constructing a specific adsorption-reduction model instance, a multi-objective iterative optimization based on constraints was performed on the model instance, using a standardized water quality vector as the initial state boundary and a preset hexavalent chromium concentration removal rate as the treatment objective, to obtain the optimal set of process parameters, including the target reaction pH value. This optimization process first requires constructing an optimization problem model containing a vector of control variables to be optimized, which is at least determined by the fulvic acid dosage concentration (…). ) and set the reaction pH value ( The composition is as follows. Simultaneously, an objective function is defined to minimize the total cost of the pharmaceutical preparation. This function is typically a weighted sum of the costs of fulvic acid and pH adjusters, and can be expressed by the following formula: in, and This is the cost weighting coefficient. The concentration of fulvic acid to be added. To set the pH value for the reaction, The initial pH in the standardized water quality vector, This represents the total cost.

[0037] Subsequently, the system employs a nonlinear programming algorithm (such as particle swarm optimization or genetic algorithm) to perform a global iterative search within the preset feasible region of process parameters (e.g., pH value set between 2 and 4, where acidic conditions are more favorable). In each iteration, the algorithm generates a set of temporary control variables (i.e., a set of...). and The algorithm then uses the adsorption-reduction model instance built in the previous sub-step to predict the Cr(VI) removal rate under these conditions when the reaction reaches a predetermined endpoint time (e.g., 24 hours). The algorithm determines whether the predicted removal rate meets the preset treatment target (e.g., a removal rate of no less than 90%). For all solutions that satisfy the constraints, the algorithm calculates the corresponding cost function J and iteratively searches for the combination of control variables that minimizes J. When the algorithm converges, the globally optimal solution it finds is the optimal set of process parameters. This step extracts the optimal set reaction pH value from this set as the final output, i.e., the target reaction pH value.

[0038] In one specific embodiment, the system receives a standardized water quality vector {initial hexavalent chromium concentration: 30 mg·L⁻¹}. -1 Initial pH: 2.0, Initial fulvic acid concentration: 100 mg·L⁻¹ -1 Initial Fe ion concentration: 0.05 mol·L -1 The water temperature is 25°C, and the preset treatment target is "Cr(VI) removal rate not less than 90%". First, a model instance is constructed, and the rate constant is calculated based on the regression model under the current water quality. and balanced adsorption capacity With variables and The functional relationship between them. Then, the particle swarm optimization algorithm is initiated, and the settings are... The search range is 0 mg·L. -1 Up to 300 mg·L -1 , The search range is [2.0, 4.0]. The algorithm begins iterating; for example, test point A: { =150, =2.5}. The model predicts that under these conditions, the removal rate can reach 96% after 24 hours of reaction, meeting the 90% target, and the corresponding cost is also calculated. Next, test point B: { =120, =2.2}. The model predicts a removal rate of 92%, which also meets the target. The cost was calculated. .like < If point B is the better solution, then point B is the current better solution. After hundreds of such iterations, the algorithm eventually converges to an optimal solution, for example, { =115, =2.3}. At this point, the system has completed the intelligent calculation and outputs the pH value of 2.3 from the optimal solution as the target reaction pH value to guide the subsequent closed-loop control process.

[0039] For example, in step S4, a real-time pH feedback signal is acquired. It should be understood that in the fulvic acid-mediated gangue adsorption-reduction system, pH is a core environmental factor determining the removal efficiency of hexavalent chromium. An acidic environment is conducive to maintaining the protonated state of functional groups such as carboxyl and phenolic hydroxyl groups in fulvic acid, thereby enhancing their ability as electron donors and promoting the reduction of hexavalent chromium. The reduction reaction of hexavalent chromium itself is accompanied by a severe consumption of protons. This stoichiometric proton consumption leads to an endogenous dynamic drift in the pH of the reaction system. If the pH signal cannot be acquired in real time and adjusted accordingly, the reaction environment will rapidly deviate from the optimal acidic range, resulting in the deprotonation of fulvic acid functional groups and precipitation covering active sites, thus severely affecting the treatment efficiency.

[0040] Specifically, real-time pH feedback signals are obtained through an online pH sensor installed inside the mine water treatment reactor. The probe of this pH sensor (typically an industrial-grade composite pH electrode) is continuously immersed in a mixture of mine water, gangue, fulvic acid, and other components undergoing an adsorption-reduction reaction. Throughout the reaction cycle, the sensor continuously measures the hydrogen ion activity in the mixture and converts it into a proportionally varying, continuously changing electrical signal (typically a millivolt-level potential difference signal). This raw electrical signal is transmitted via a signal cable to a data acquisition module or programmable logic controller (PLC). The data acquisition module first performs analog-to-digital conversion on the analog signal, discretizing it into a digital signal sequence. Subsequently, to eliminate electrical noise that may be introduced by equipment such as stirrers during the measurement process, a low-pass digital filter is applied to smooth the digital sequence. Finally, and most crucially, the system calls a pre-installed sensor calibration model (typically based on the Nernst equation with temperature compensation) to accurately convert the filtered electrical signal value into an internationally recognized, dimensionless pH value. The continuous data stream obtained after the complete processing, representing the instantaneous true pH value inside the reactor, constitutes the real-time pH feedback signal output in this step. This process is carried out continuously at a high frequency (e.g., several or tens of times per second) to ensure that the control system can keep abreast of the latest dynamics of the pH value of the reaction system.

[0041] In one specific embodiment, the preliminary steps determined the target reaction pH to be 2.3. After the reaction started, the pH electrode immersed in the reactor continued to operate. At the 15-minute mark of the reaction, the pH of the system tended to rise due to the large amount of H⁺ consumed by the reduction of Cr(VI). At this moment, the pH sensor measured a potential difference signal of -68.9 mV. This signal was captured by the data acquisition card, filtered, and input into the calibration program. The calibration program, based on the Nernst equation and the current water temperature of 25.5°C synchronously obtained from the temperature sensor, calculated and converted the -68.9 mV electrical signal into an actual pH value, for example, 2.41. This value of 2.41 is the real-time pH feedback signal at that moment. It is immediately transmitted to the subsequent closed-loop control module. The control module compared it with the target value of 2.3, obtaining a deviation of +0.11. This deviation signal drives the PID feedback controller to generate a correction command, which, after being integrated with other control signals, precisely adjusts the acid dosing equipment to bring the pH value back to the target level of 2.3.

[0042] For example, in step S5, a forward prediction of the proton consumption rate is made based on the amount of gangue added, the amount of fulvic acid added, the current reaction time, and an adsorption-reduction model instance to obtain the predicted proton consumption rate. It should be understood that in a system containing siderite and fulvic acid, the reduction of hexavalent chromium is accompanied by a complex acid consumption process. This consumption is continuous and intense. If adjustment is based solely on pH feedback signals, the system pH often deviates from the optimal reaction range by the time a deviation is detected, severely affecting the protonation state of the fulvic acid functional groups and the final reduction efficiency. Since the hexavalent chromium reduction reaction follows specific kinetic laws, the resulting pH increase trend is a predictable, endogenous systemic disturbance. Therefore, by introducing a forward prediction mechanism, the source of chemical disturbance that is difficult to measure directly is transformed into a quantitative signal that can guide the control system to act in advance. This effectively compensates for the intense acidity fluctuations caused by high concentrations and rapid reactions in the early stages of the reaction, ensuring that the reaction always proceeds in a kinetically optimal acidic environment.

[0043] In one embodiment, a prospective prediction of the proton consumption rate is performed based on the gangue dosage, fulvic acid dosage, current reaction time, and an adsorption-reduction model instance to obtain the predicted proton consumption rate. This includes: first, obtaining all the input information required to perform this step, including the already determined gangue and fulvic acid dosages (the sum of which constitutes the total adsorbent mass), the current reaction time t (starting from the start of the reaction), and the pseudo-second-order adsorption rate constant constructed in the previous steps for the current water quality, which includes specific parameters. and balanced adsorption capacity A pseudo-second-order adsorption-reduction model was established. Subsequently, the first-order derivative of the established pseudo-second-order adsorption-reduction model was taken over time to obtain the instantaneous reaction rate per unit adsorbent. This rate was then combined with the total adsorbent mass *m* and the stoichiometric coefficient. The calculation is performed, and the final output is a value that varies with the current reaction time. Predicted rate of change in proton consumption The core of this calculation lies in extending the adsorption-reduction kinetic model from describing the cumulative adsorption amount to describing the instantaneous reaction rate, and ultimately relating it to the proton consumption rate through stoichiometry. Specifically, the proton consumption rate is prospectively predicted using the following formula: in, Represents the predicted proton consumption rate, in mol / min; The dimensionless stoichiometric conversion factor represents the amount of stoichiometry consumed to reduce one mole of Cr(VI). The number of moles, the value of which depends on the dominant reduction reaction pathway (e.g., if Cr2O7 is used). 2- +14H + →... as the main focus, then 7) Pre-set and store in the system; The total mass of adsorbent added is obtained by adding the amount of gangue and the amount of fulvic acid added; Here is the molar mass of chromium; Let be the instantaneous adsorption capacity at time t; It is the pseudo-second-order adsorption rate constant; To balance the adsorption capacity; t represents the current reaction time. By substituting the current reaction time t into this formula within each control cycle, the system can continuously and dynamically generate a time-varying predicted proton consumption rate data stream.

[0044] In one embodiment, the adsorption-reduction model instance of the system has determined the parameters under the current operating conditions as follows: =0.0558 g·(mg·min) -1 , = 2.5351 mg / g. Total mass of adsorbent added. The value is 1000g. The stoichiometric conversion coefficients stored in the system... The molar mass of chromium is 7. The concentration is 52000 mg / mol. At t = 10 minutes, the system needs to predict the proton consumption rate. First, the instantaneous adsorption capacity at t = 10 minutes is calculated using a model example. (10) (10)≈1.49 mg / g. Then, substituting all known values ​​into the proton consumption rate prediction formula, the predicted proton consumption rate at the 10th minute of the reaction was obtained. Approximately 0.0082 mol / min. This value means that the system predicts that at the current moment, due to the reduction of Cr(VI), the reactor is consuming approximately 0.0082 mol / min of H₂. + The process is proceeding at a certain rate. This data will be immediately used in subsequent complex closed-loop control calculations to guide the controller to add acid in advance and in sufficient quantity, thereby precisely counteracting the impending pH rise.

[0045] For example, in step S6, based on the predicted proton consumption rate, closed-loop control of the target reaction pH and the real-time pH feedback signal is performed to obtain actuator commands. It should be understood that in the system containing siderite and fulvic acid, the reduction of hexavalent chromium is accompanied by a complex acid consumption process. This continuous proton consumption means that the upward trend of the pH is predictable and an endogenous systematic disturbance of the main reaction, rather than random noise. If only traditional PID control is used, its logic only begins correction after detecting a pH deviation from the set value. This lag in control leads to frequent deviations of the pH from the optimal reaction range in the early stages of the reaction, thus affecting the protonation state of the fulvic acid functional groups and the overall processing efficiency. Therefore, this step aims to construct a feedforward-feedback composite control mechanism to achieve precise pH control through a prediction plus correction mode, fundamentally suppressing systematic fluctuations caused by reaction consumption.

[0046] In one embodiment, closed-loop control of the reaction process based on the predicted proton consumption rate and the real-time pH feedback signal is performed to obtain actuator commands, including: fusing the feedforward-feedback composite control signal based on the predicted proton consumption rate and the real-time pH feedback signal to obtain the composite acid injection acceleration rate; and converting the composite acid injection acceleration rate into actuator commands.

[0047] Specifically, firstly, the predicted proton consumption rate obtained in the previous step is... Through a feedforward gain This is directly converted into a basic acid dosing rate to compensate for most known systemic disturbances; simultaneously, a parallel PID feedback controller continuously calculates the deviation between the target pH value and the real-time pH feedback signal. The feedforward channel generates a feedback correction signal. Finally, the base addition signal generated by the feedforward channel and the correction signal generated by the feedback channel are algebraically superimposed to obtain the final composite acid addition acceleration rate. At the algorithm level, it embodies a highly creative division of labor and collaboration approach. The feedforward section takes the lead, actively and smoothly adding acid to address major, predictable pH disturbances and suppressing large fluctuations at their source. The feedback section then acts as a supervisor, precisely correcting minor, random deviations such as model prediction errors and temperature fluctuations. By combining the advantages of these two control strategies, the pH control system exhibits faster response, higher robustness, and smaller overshoot.

[0048] The core calculation formula is as follows: in, This represents the acceleration rate of the composite acid solution, expressed in L / min. This is the feedforward gain coefficient; This is the standard concentration of the acid solution; , and These are the proportional, integral, and derivative gain coefficients of the PID controller, respectively. This represents the deviation between the target pH value and the real-time pH value.

[0049] Finally, the drive commands for the metering pump are calibrated and output. That is, the theoretical dosage calculated by the upper-level algorithm is accurately converted into an electronic control signal that can be recognized and executed by the underlying physical hardware.

[0050] In a specific example, the system calls a pre-stored function or lookup table of the metering pump flow-control signal characteristic curve. The composite acid solution obtained in the previous step is added to the acceleration rate. As input, the actual control signal value required by the metering pump is obtained through querying or calculation. Examples include PWM duty cycle or 4-20mA current signal. It serves as the last mile connecting data algorithms and the physical world, ensuring that the decisions of the upper-level composite control algorithm can be accurately executed by the underlying hardware devices without distortion, thereby guaranteeing the closed-loop integrity of the entire advanced control strategy and its feasibility in industrial settings.

[0051] The core calculation formula is as follows: in, This represents the real-time control signal output to the metering pump actuator. This is the calibration function for the metering pump.

[0052] In this way, by introducing predictive feedforward control based on a reactive adsorption-reduction model, proactive and precise regulation of pH value during mine water treatment can be achieved. Compared with traditional single feedback control, this method transforms the pH control system from a passive responder to an active predictor, which can significantly suppress systemic pH fluctuations caused by the consumption of H⁺ in the Cr(VI) reduction reaction at the source. The final technical effects are: significantly improving the stability of pH value throughout the entire reaction cycle, keeping it more closely within the optimal reaction range, thereby maximizing the reduction rate and total removal efficiency of Cr(VI); at the same time, by avoiding frequent overdosing and adjustments, it can also effectively reduce the consumption of acid-base regulators and reduce operating costs; the system response is faster, the robustness is stronger, and the adaptability to water quality changes is also enhanced, ultimately ensuring the stable achievement of mine water treatment standards.

[0053] In one embodiment, at the 10th minute of the reaction, the system receives three key inputs: a predicted proton consumption rate. The flow rate is 0.0082 mol / min, the target reaction pH is 2.3, and the real-time pH feedback signal is 2.41. The preset control parameters within the system are: feedforward gain... =1.0, acid concentration =1mol / L, PID parameters =0.05L / min =0.01L / (min·s) =0.001 L·s / min. First, the acid dosing rate of the feedforward section is calculated to be (1.0 / 1)*0.0082 = 0.0082 L / min. Simultaneously, the pH deviation of the feedback section is calculated. =2.41-2.3=0.11. To fully apply the PID algorithm in the formula, the system acquires historical deviation data from the start of the reaction to the present moment. In this embodiment, after system sampling and integration, the current deviation integral term... The cumulative value is 1.5 (units). (Adaptation); At the same time, the system detected the rate of change of deviation at the most recent moment. 0.5 (units) (Adaptation). The system then performs the following calculations based on the given PID parameters: the proportional term output is... =0.05×0.11=0.0055L / min; the integral term output is ×Integral value = 0.01 × 1.5 = 0.0150 L / min;Derivative output is × rate of change = 0.001 × 0.5 = 0.0005 L / min. Adding these three terms together, the actual feedback correction flow rate of the PID controller is 0.0055 + 0.0150 + 0.0005 = 0.0210 L / min. Finally, the system uses the formula to superimpose and fuse the feedforward basic injection flow rate and the feedback correction flow rate to calculate the final compound acid injection rate. =0.0082+0.0210=0.0292L / min.

[0054] Finally, the calibration and output of the metering pump drive command are performed. In this embodiment, a high-precision electromagnetic metering pump suitable for micro-dosing is selected, with a maximum design flow rate of 0.1 L / min. The system has a pre-set calibration function for this specific model of metering pump. This function is based on the linear response characteristics determined in the laboratory, and its mathematical expression is: in, The PWM duty cycle instruction (%) for output; The input is the desired flow rate (i.e., the acceleration rate of the composite acid solution calculated in the previous step). ; This is the maximum constant flow rate of the metering pump (0.1 L / min). This is the dead zone compensation bias (set to 2.0% in this example to overcome the mechanical resistance during pump startup). Based on the compound acid injection acceleration rate calculated in the previous step... =0.0292L / min, the system substitutes this value into the above calibration function for calculation: =29.2% + 2.0% = 31.2%, the system calculates the final PWM duty cycle value to be approximately 31.2%. Subsequently, the system encapsulates this value into a standard actuator command packet and sends it to the frequency converter controller of the metering pump via the RS-485 industrial bus. After parsing the command, the controller immediately adjusts the on / off timing of the electromagnetic drive coil to perform the stroke action with a 31.2% duty cycle, thereby accurately outputting a flow rate of 0.0292 L / min, achieving closed-loop precision dosing of the reaction system.

[0055] To better illustrate the scientific principles and practical feasibility of the intelligent control method proposed in this application, the following section will elaborate on the influence mechanism of fulvic acid (FA), as a representative of dissolved organic matter (DOM), on the adsorption-reduction process of hexavalent chromium (Cr(VI)) in a system using coal gangue as the adsorbent, based on specific experimental data and molecular simulation results. These experimental results provide a solid theoretical basis and data support for the construction of the adsorption-reduction model, the selection of key parameters, and the formulation of control strategies in this method.

[0056] The specific experimental conditions are as follows: Fulvic acid molecular formula C 14 H 12 O8, molecular weight 308.24. Fresh gangue was crushed to 100 mesh (0.149 mm) and dried to constant weight. Several conical flasks were taken, and 1 g of gangue and 0, 100, 200, and 300 mg of FA were added respectively. A 30 mg·L⁻¹ solution was added at a solid-liquid ratio of 1:100. -1 Adsorption-reduction experiments were conducted on Cr(VI) contaminated liquid. The adsorption-reduction reaction was carried out using 0.1 mol·L⁻¹ water. -1 The initial pH values ​​were adjusted to 2, 4, 6, and 8 with hydrochloric acid or NaOH solution to simulate different acidity and alkalinity of mine water. Different masses of siderite (FeCO3) were added to achieve an initial C:Fe molar ratio of FA to Fe of 1:0.5, 1:1, 1:2, and 1:3, respectively, simulating the Fe enrichment background in mine water caused by a large amount of iron minerals such as siderite nodules in a coal-bearing stratum. Three parallel experiments were conducted for each reaction condition. The conical flasks were sealed with a sealing film, and the reaction was carried out in a constant temperature shaking chamber at 25℃ and 175 rpm in darkness. At 0.5, 1, 2, 4, 6, 8, 12, and 24 h, 10 mL of the contaminated liquid was taken using a syringe, filtered through a 0.22 μm polycarbonate filter membrane, and the Cr(VI) content was determined by diphenylcarbazide spectrophotometry.

[0057] The reduction rate (R) of Cr(VI) and the adsorption capacity (q) of gangue for Cr(VI) are calculated according to the following formulas: In the formula, C0 is the initial concentration of Cr(VI) (mg·L). -1 ), C e Instantaneous concentration of Cr(VI) (mg·L) -1 V is the volume of the solution (L); The total mass (g) of gangue and fulvic acid.

[0058] The pseudo-first-order and pseudo-second-order kinetic equations for the adsorption process are nonlinearly fitted using the following formulas: In the formula, The instantaneous adsorption capacity at time t (mg·g) -1 ), The adsorption capacity at equilibrium (mg·g) -1 ), The quasi-first-order rate constant (min) -1 ), The quasi-second-order rate constant (g·(mg·min)) -1 ), where t is the adsorption time (min).

[0059] The intraparticle diffusion model of the adsorption process was linearly fitted using the following formula: In the formula, k3 represents the intraparticle diffusion rate constant (mg·(g·min)). 1 / 2 ) -1 C is the boundary layer drag constant (mg·g). -1 The remaining parameters are the same as above.

[0060] according to Figure 2 As shown, the FTIR spectrum of gangue is between 1106.87 and 1016.53 cm⁻¹. -1 The region exhibits a strong absorption peak, accounting for 88.85%, which is characteristic of CO stretching vibration; the second strongest peak is at 3415.22 cm⁻¹. -1 The absorption peak belongs to hydroxyl-hydroxyl hydrogen bonds formed by self-associated hydroxyl groups, accounting for 9.19%; there is also a small amount of aliphatic structure (1432.21 cm⁻¹). -1 (1.35%) and carbonyl (1633.57cm) -1 (Accounting for 0.62%). Following the FA-mediated adsorption-reduction of Cr(VI) in gangue, ... Figure 2 It can be seen that the intensity of the infrared absorption peak of the gangue generally increases after the reaction, which is related to the large number of oxygen-containing functional groups contained in FA itself. With the interaction of FA and gangue through complexation, coordination or covalent bond formation, the CO absorption peak significantly blue-shifts to 1160.30 ~ 1093.52 cm⁻¹. -1 The proportion decreased to 59.92%; while the hydroxyl absorption peak broadened to 3545.82 cm⁻¹. -1 and 3284.83 cm -1 The newly formed absorption peaks at 3418.40 cm⁻¹ represent hydroxyl-π hydrogen bonds formed between hydroxyl groups and π-phase atoms on the aromatic ring, and hydrogen bonds in the cyclic structure, accounting for 9.53% and 8.23% respectively. -1 The proportion of the absorption peak increased to 17.23%. The results indicate that when there is a large amount of FA in the water, FA can be adsorbed onto the surface of gangue through physical and chemical processes. Therefore, during the adsorption-reduction of Cr(VI) in gangue, FA will compete with Cr(VI) for adsorption sites.

[0061] Table 1. Proportion of infrared absorption peaks in gangue The initial FA concentration significantly affects the reduction efficiency of Cr(VI), such as Figure 3 As shown, during the 24-hour reaction process: when the initial FA = 0 mg·L⁻¹-1 At that time, the Cr(VI) concentration was 30 mg·L⁻¹. -1 Decreased to 16.73 ± 0.29 mg·L -1 The corresponding reduction rate was 44.22 ± 0.96%; the initial FA concentration was increased to 100 mg·L⁻¹. -1 At that time, 68.52 ± 1.20% of Cr(VI) in the system was reduced, and the concentration decreased by 20.55 ± 0.36 mg·L⁻¹. -1 Increase FA concentration to 200 mg·L -1 At that time, the removal amount of Cr(VI) reached 25.32 ± 0.31 mg·L. -1 The reduction rate was 84.39 ± 1.04%; when the initial FA concentration reached 300 mg·L⁻¹. -1 At that time, the residual concentration of Cr(VI) after the reaction was only 2.75 ± 0.36 mg·L. -1 The reduction rate is as high as 90.82±1.21%.

[0062] Therefore, the adsorption-reduction rate of Cr(VI) by gangue mediated by FA is positively correlated with the initial FA concentration. This concentration dependence mainly stems from the unique pore structure of gangue and its adsorption characteristics for FA. Studies have shown that the surface of gangue particles has a well-developed pore structure, which increases the amount of FA adsorbed by gangue with increasing initial solution concentration. High concentrations of FA exacerbate the competition for adsorption sites on the gangue surface, leading to a decrease in the direct adsorption of Cr(VI). However, FA can compensate for the loss of Cr(VI) adsorption by gangue through reduction. FA can act as an electron shuttle, promoting electron transfer between gangue and Cr(VI), while functional groups such as carboxyl groups (-COOH) and phenolic hydroxyl groups (-OH) in FA can directly serve as electron donors. ] It promotes the valence transformation of Cr(VI) to Cr(III) through complexation, thereby significantly improving the reduction efficiency. In addition, high concentrations of FA contain more organic functional groups, which effectively increases the number of active sites participating in the reduction reaction.

[0063] Initially, FA = 100 mg·L -1 At that time, the reduction rate of Cr(VI) decreased as the pH value increased. For example Figure 4 As shown, when the initial pH of the system was 2, 4, 6, and 8, after 24 h of reaction, the reduction rates of Cr(VI) by gangue mediated by FA were 91.26±0.68%, 85.42±0.90%, 81.81±1.46%, and 51.47±1.11%, respectively. It can be seen that acidic conditions favor the reduction of Cr(VI). When the initial pH increased from 2 to 8, the reduction rate of Cr(VI) by gangue mediated by FA decreased by approximately 43.60%.

[0064] This pH-dependent characteristic is closely related to the protonation state of functional groups: in acidic environments, the -COOH and -OH functional groups of FA remain protonated, enhancing their ability as electron donors and promoting the reduction of Cr(VI); while under alkaline conditions, the FA functional groups deprotonate to form -COO. - and -O - The presence of functional groups (F) results in a negatively charged FA surface, enhancing electrostatic repulsion with Cr(VI). This also increases FA solubility and causes it to dissociate from the gangue pores, weakening the electron-donating capacity of the active sites on the gangue surface. Furthermore, low pH increases the porosity and pore size of the gangue surface, enhancing its adsorption capacity; while the Cr(OH)3 precipitate formed in alkaline media covers the active sites on the gangue surface, forming a physical barrier effect and further inhibiting the reduction reaction.

[0065] like Figure 5 As shown, the reduction rate of Cr(VI) increases with increasing background Fe concentration. When the initial C:Fe molar ratio of the system is increased from 1:0.5 to 1:3, after 24 h of reaction, the reduction rate of Cr(VI) mediated by FA under gangue increases from 73.44±0.38% to 90.98±0.61%. After the introduction of siderite into the system, it gradually dissolves and releases Fe(II) (Equation 5-6). The increase in background Fe concentration significantly increases the supply of Fe(II), and Fe(II), as a strong reducing agent, directly participates in the reduction reaction of Cr(VI) to generate Fe(III) and Cr(III) (Equations 5-7, 5-8). FA can form a complex with Fe(III), inhibiting Fe(III) precipitation, while promoting the Fe(II) / Fe(III) cycle, reducing Fe(III) to Fe(II), and maintaining a continuous supply of Fe(II) in the system. ] FA, as an electron shuttle, can accelerate the electron transfer rate from Fe(II) to Cr(VI). Furthermore, Fe(III) and Cr(III) in the system can be co-precipitated and adsorbed by gangue in the form of iron-chromium hydroxide (Equation 5-9), thus achieving Cr fixation.

[0066] FeCO3 + 2H + →Fe 2+ +CO2↑+H2O 6Fe 2+ +Cr₂O₂⁻ + 14H⁺ + →6Fe 3+ +2Cr 3+ +7H2O 3Fe 2+ +HCrO-4+7H + →2Fe 3+ +2Cr 3+ +4H2O xCr3+ +(1-x)Fe 3+ +3H2O→(Cr x Fe 1-x (OH) 3(s) +3H + The pseudo-first-order and pseudo-second-order kinetic fitting curves of Cr(VI) adsorption on gangue under different initial conditions are shown in the figure. Figure 6 As shown, the overall pattern exhibits characteristics of "rapid adsorption followed by slow equilibrium." Based on the fitting parameters (Table 2), it can be seen that only when the initial FA = 0 mg·L⁻¹... -1 The pseudo-first-order kinetic equation showed a higher goodness of fit than the pseudo-second-order equation, indicating that in the absence of FA in the water, Cr(VI) was mainly adsorbed by gangue through physical processes. However, with the presence of FA, even after changing the initial FA concentration, pH value, or C:Fe molar ratio, the pseudo-second-order kinetic equation consistently showed a higher goodness of fit than the pseudo-first-order equation, demonstrating that the introduction of FA significantly enhanced the chemisorption behavior, consistent with the reported adsorption results of Cr(VI) by the mineral-DOM complex. The pseudo-second-order kinetic curves showed that as the initial FA concentration increased, the pH value decreased, and the Fe background value increased, the time required for gangue to reach equilibrium for Cr(VI) adsorption decreased. The pseudo-second-order rate constant k2 gradually increased, increasing by 11.56 times, 5.86 times, and 10.30 times under the three conditions, respectively. The adsorption capacity q at equilibrium was... e These figures increased by 85.38%, 31.07%, and 24.22%, respectively.

[0067] Table 2 Fitting parameters of Cr(VI) adsorption kinetics in gangue mediated by fulvic acid Under different initial conditions, the FA-mediated adsorption process of Cr(VI) by gangue conforms to the three-stage characteristics of the intraparticle diffusion model, such as... Figure 7As shown, the intraparticle diffusion rate constant k3 gradually decreases with increasing reaction time in each stage. In the rapid adsorption phase of stage 1, the FA-Cr(VI) complex rapidly diffuses from the bulk liquid phase to the liquid film layer on the surface of the gangue particles, undergoing chemisorption at surface active sites, such as the coordination of Cr(VI) with oxygen-containing functional groups on the surface. At this stage, the concentration of Cr(VI) in the liquid phase is high, the mass transfer driving force is strong, and k3 is relatively high. In stage 2, Cr(VI) enters the internal pores of the gangue particles. The extended diffusion path and reduced active sites lead to a significant decrease in the reaction rate, with k3 decreasing by 83.19 ± 8.71% compared to stage 1. The rate in this stage is limited by the tortuosity of the pore structure and the size effect of the FA-Cr(VI) complex. In stage 3, the active sites on the gangue surface are close to saturation, the Cr(VI) concentration decreases significantly, and the entire system reaches dynamic equilibrium, with k3 approaching 0. Furthermore, none of the fitted straight lines passed through the origin (C ≠ 0), indicating that the adsorption process is controlled by multiple steps, including liquid film diffusion, intraparticle diffusion, and surface reaction.

[0068] During adsorption, Cr(VI) undergoes mass transfer between the non-reactive phase (Cr(VI) in the bulk liquid phase that does not interact with the gangue surface) and the phase interface (solid-liquid boundary layer). C reflects the boundary layer resistance; the higher the value, the stronger the resistance to adsorption. C less than 0 only occurs when the initial FA = 0 mg·L⁻¹. -1 In the first stage, the lack of FA may have reduced the thickness of the liquid film, shortening the diffusion path of Cr(VI). The gangue directly enhanced the liquid film mass transfer efficiency of Cr(VI) through electrostatic attraction. Under other conditions, C was greater than 0, and the boundary layer resistance was significant. This indicates that FA forms a macromolecular complex with Cr(VI), increasing the liquid film mass transfer resistance, or that FA molecules loaded at the pore inlets on the gangue surface hindered the diffusion of Cr(VI) into the particle interior.

[0069] The XRD pattern of gangue produced in a certain coal mining area is as follows: Figure 8 As shown, quartz and kaolinite are the most abundant components in the gangue, accounting for 42.96% and 31.80% respectively, followed by siderite (12.33%), calcite (8.44%), and serpentine (4.47%). Kaolinite, a layered clay mineral, can adsorb various organic matter and ions in mine water, purifying mine water resources. Its chemical formula is Al₄(Si₄O₂). 10(OH)8, belonging to the dioctahedral TO-type layered silicate minerals, lacks interlayer cations or water molecules, with structural layers connected by hydrogen bonds. Using kaolinite, which has a high content, to represent the gangue structure, and based on the spatial coordinates of atoms and interlayer cations in the kaolinite unit cell reported by Bish, the unit cell parameters were set in Materials Studio 2020 software as follows: a = 5.1535 Å, b = 8.9419 Å, c = 7.3906 Å, α = 91.926°, β = 105.046°, γ = 89.797°. Subsequently, a 4a×2b×2c supercell structure model was constructed, as shown below. Figure 9 As shown.

[0070] exist Figure 9 Under periodic boundary conditions, the task was set to Fixed pressure using the Sorption module, the calculation method was Metropolis, and the accuracy was Ultra-fine. The adsorption of Cr₂O₂⁻⁷ was simulated, reaching a saturation adsorption capacity of 12 molecular weights / uc after 14225 iterations. This saturation adsorption contained 28 Cr(VI) atoms distributed on the surface of the aluminum-oxygen octahedra in kaolinite. Figure 10 As shown. A composite of FA and kaolinite crystal planes was constructed using the Build Layers module. After 58,200 iterations, the saturated adsorption capacity of Cr2O2-7 was still 12 molecular / uc. However, the adsorption sites of Cr2O2-7 changed. Since FA occupies some sites on the kaolinite surface, 1 / 3 of the Cr2O2-7 was converted to bind with FA and indirectly adsorbed.

[0071] From the perspective of energy change, van der Waals forces dominate the adsorption of Cr2O2-7 by kaolinite, which verifies that the initial FA = 0 mg·L. -1 The adsorption of Cr(VI) in gangue is mainly a result of physical processes. Under FA-mediated action, the energy of the van der Waals forces is -1593.55 ± 35.23 kcal·mol⁻¹. -1 There is still some physical adsorption. At adsorption equilibrium, the intramolecular energy reaches 5009.02 kcal·mol⁻¹. -1This reflects a deformation of the internal structure of the system. Cr(VI) adsorption may cause local distortion of the aluminum-oxygen octahedral or silicon-oxygen tetrahedral layers of kaolinite, with bond lengths or bond angles deviating from equilibrium values, increasing bond stretching energy and bond angle bending energy. Meanwhile, FA may fold or stretch to accommodate Cr(VI) adsorption, leading to an increase in dihedral torsion energy. Normally, adsorption is an exothermic process, but the total energy during FA-mediated Cr(VI) adsorption in kaolinite is positive, indicating that the overall reaction requires energy absorption. This may be due to a competitive mechanism or reduced thermodynamic stability: because FA competes with Cr(VI) for active sites on the kaolinite surface, when some sites on the kaolinite surface are covered by FA, ​​Cr(VI) adsorption requires overcoming the binding energy of kaolinite-FA, increasing energy consumption during adsorption; and the complexation of Cr(VI) with FA functional groups may disrupt the original hydrogen bond network, also requiring additional energy to maintain the stability of the new structure.

[0072] When Cr(VI) is saturatedly adsorbed onto kaolinite under FA-mediated conditions, the radial distribution functions of Cr(VI) with oxygen-containing functional groups in FA and O atoms in the Al-OH layer and Si-O plane on the kaolinite surface are as follows: Figure 11 As shown, the interparticle distances corresponding to the first peak in the radial distribution function of hydroxyl (-OH), carbonyl (-C=O), and Al-OH layers on the kaolinite surface with Cr(VI) in FA are all less than 3.5 Å, specifically 3.36 Å, 3.21 Å, and 3.43 Å, respectively. This indicates that Cr(VI) closest to these three structures is adsorbed by forming chemical bonds or hydrogen bonds, with -C=O exhibiting the strongest affinity, followed by -OH and Al-OH. In contrast, the carboxyl (-COOH), ether bond (COC), and kaolinite Si-O surface in FA adsorb Cr(VI) through van der Waals forces. The radial distribution function of -C=O with Cr(VI) has the highest first peak g(r) value of 9.41, while the first peak g(r) values ​​of -OH and Al-OH with Cr(VI) are 5.74 and 1.29, respectively. It can be seen that the probability of combining with Cr(VI) through chemical bonding during adsorption is from largest to smallest as -C=O > -OH > Al-OH.

[0073] The above experiments and simulations revealed the regulatory mechanism of DOM, represented by FA, ​​on the adsorption-reduction of Cr(VI) in gangue. The results show that: (1) FTIR showed that FA could be adsorbed by gangue, leading to an increase in oxygen-containing functional groups on the surface. During the adsorption-reduction of Cr(VI) by gangue, FA would compete with Cr(VI) for adsorption sites. The reduction rate of Cr(VI) by gangue itself was 44.22±0.96%. The increase in FA concentration significantly improved the reduction rate of Cr(VI). Its functional groups promoted the reduction of Cr(VI) to Cr(III) through electron donation, compensating for the decrease in the amount of direct adsorption by gangue due to competition for adsorption sites. Acidic conditions were more conducive to the protonated functional groups of FA enhancing the electron donation capacity, while alkaline media inhibited the reduction reaction. The introduction of siderite enabled FA to promote the Fe(II) / Fe(III) cycle, leading to an increase in the reduction rate of Cr(VI), which was eventually co-precipitated with Fe and adsorbed by gangue, thus achieving Cr fixation.

[0074] (2) Quasi-first-order dynamic model (R 2 =0.9605) indicates that the gangue itself adsorbs Cr(VI) through physical processes, while the FA-gangue system is mainly characterized by chemical adsorption (pseudo-second-order kinetic model R). 2 >0.93); The intraparticle diffusion model indicates that the adsorption of Cr(VI) in gangue mediated by FA is controlled by multiple steps, including liquid film diffusion, intraparticle diffusion and surface reaction. The formation of macromolecular complexes between FA and Cr(VI) increases the resistance to liquid film mass transfer, and the FA molecules loaded on the pore inlets of the gangue surface hinder the diffusion of Cr(VI) into the particle interior.

[0075] (3) Kaolinite, which accounts for as much as 31.80% of the gangue, can adsorb a variety of organic matter and ions in mine water. Molecular simulations show that during the adsorption of Cr(VI) by kaolinite mediated by FA, ​​the internal structure of the system undergoes deformation and the internal energy of the molecules increases. FA competes for kaolinite surface sites, which means that the Cr(VI) adsorption process needs to overcome the binding energy of kaolinite-FA, and the entire system needs additional energy to maintain the stability of the new structure. The radial distribution function shows that the carbonyl group (-C=O) of FA has the strongest binding ability with Cr(VI) and the highest probability of forming a chemical bond, followed by the hydroxyl group (-OH) and the Al-OH layer on the surface of kaolinite.

[0076] In summary, the intelligent control method for mine water treatment based on the adsorption-reduction model provided in this application has been elucidated. Firstly, by introducing model-based predictive feedforward control, this application can proactively compensate for the pH increase trend caused by the main reaction, eliminating the lag inherent in traditional feedback control at its source, resulting in faster system response and more timely control. Secondly, this application can significantly suppress pH fluctuations and overshoot during the reaction process, keeping it more firmly stable within the optimal reaction range, thereby significantly improving the overall reduction efficiency of Cr(VI) and the stability of the treatment effect. Furthermore, precise and smooth control avoids reagent waste caused by repeated over-adjustment, effectively reducing operating costs. Finally, this composite control architecture enhances the system's robustness, enabling it to better adapt to changes in different initial water quality conditions, ensuring stable achievement of mine water treatment standards, and possessing significant engineering application value.

[0077] This application also provides an intelligent control system for mine water treatment based on an adsorption-reduction model, used to execute the aforementioned intelligent control method for mine water treatment based on an adsorption-reduction model, such as... Figure 12 As shown, the intelligent control system 100 for mine water treatment based on the adsorption-reduction model includes: a raw data acquisition module 110 for acquiring raw sensor data; a data standardization module 120 for standardizing the raw sensor data to obtain a standardized water quality vector; a target pH intelligent calculation module 130 for intelligently calculating target parameters of the standardized water quality vector based on the treatment target and the adsorption-reduction model instance to obtain the target reaction pH value; a real-time pH acquisition module 140 for acquiring real-time pH feedback signals; a proton consumption rate prediction module 150 for making a forward-looking prediction of the proton consumption rate based on the gangue dosage, fulvic acid dosage, current reaction time, and the adsorption-reduction model instance to obtain a predicted proton consumption rate; and a closed-loop control execution module 160 for performing closed-loop control of the reaction process based on the predicted proton consumption rate, the target reaction pH value, and the real-time pH feedback signal to obtain actuator commands.

[0078] This application also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-mentioned method steps to implement the intelligent control method for mine water treatment based on the adsorption-reduction model provided in the above embodiments.

[0079] This application also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to realize the intelligent control method for mine water treatment based on the adsorption-reduction model provided in the above embodiments.

[0080] In this application, the system, computer-readable storage medium, or computer program product provided in the embodiments are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0081] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments.

[0082] The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous. The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A smart control method for mine water treatment based on an adsorption-reduction model, characterized in that, include: Acquire raw sensor data; The raw sensor data is standardized to obtain a standardized water quality vector; Based on the treatment target and an adsorption-reduction model example, the target parameters of the standardized water quality vector are intelligently calculated to obtain the target reaction pH value, including: Based on a standardized water quality vector, an adsorption-reduction model instance is constructed, including: decoupling the state vector from the standardized water quality vector and extracting feature variables to obtain a set of state variables, including initial fulvic acid concentration, initial pH, and initial Fe ion concentration; performing kinetic coefficient inversion based on multivariate regression on the set of state variables to obtain the pseudo-second-order adsorption rate constant and equilibrium adsorption capacity; and constructing an adsorption-reduction model instance based on the predicted pseudo-second-order adsorption rate constant and predicted equilibrium adsorption capacity. The adsorption-reduction model instance is represented as follows: in, The pseudo-second-order adsorption rate constant. To balance the adsorption capacity, This is the current reaction time. This represents the predicted instantaneous adsorption capacity at time t. Using a standardized water quality vector as the initial state boundary and a preset hexavalent chromium concentration removal rate as the treatment target, the adsorption-reduction model instance is subjected to multi-objective iterative optimization based on constraints to obtain the optimal set of process parameters, which includes the target reaction pH value and the optimal equilibrium adsorption amount. Obtain real-time pH feedback signal; Based on the amount of gangue added, the amount of fulvic acid added, the current reaction time, and an adsorption-reduction model example, a prospective prediction of the proton consumption rate is performed to obtain the predicted proton consumption rate, including: The prospective prediction of the proton consumption rate is performed using the following formula: in, Represents the predicted proton consumption rate, in mol / min; The dimensionless stoichiometric conversion factor represents the amount of stoichiometry consumed to reduce one mole of Cr(VI). Number of moles; This refers to the total mass of the adsorbent added. Here is the molar mass of chromium. Let be the instantaneous adsorption capacity at time t; It is the pseudo-second-order adsorption rate constant; To balance the adsorption capacity; This is the current reaction time; Based on the predicted proton consumption rate, the target reaction pH value and the real-time pH feedback signal are used to perform closed-loop control of the reaction process to obtain actuator commands.

2. The intelligent control method for mine water treatment based on the adsorption-reduction model according to claim 1, characterized in that, The raw sensor data includes the volt-ampere response current signal representing the hexavalent chromium concentration, the potential difference signal representing the pH value, the ultraviolet absorbance signal representing the fulvic acid concentration, the specific wavelength absorbance signal representing the background Fe concentration, and the resistance signal from the temperature sensor.

3. The intelligent control method for mine water treatment based on the adsorption-reduction model according to claim 2, characterized in that, The raw sensor data is standardized to obtain a standardized water quality vector, including: The raw sensor data is digitally filtered to obtain a set of denoised digital sequences; The denoised digital sequence set is subjected to parameter inversion and calibration based on a feature mapping model to obtain a standardized water quality vector, which includes the initial hexavalent chromium concentration, initial pH, initial fulvic acid concentration, initial Fe ion concentration and water temperature.

4. The intelligent control method for mine water treatment based on the adsorption-reduction model according to claim 3, characterized in that, To obtain a standardized water quality vector, parameter inversion and calibration based on a feature mapping model are performed on the denoised digital sequence set. This includes: performing parameter inversion and calibration based on a feature mapping model on the volt-ampere response current signal representing the hexavalent chromium concentration using the following formula: in, The signal is the current-voltage response representing the concentration of hexavalent chromium after noise reduction processing. For sensor sensitivity coefficient, This is the temperature compensation coefficient. For water temperature, The zero intercept is... The concentration of hexavalent chromium after calibration.

5. The intelligent control method for mine water treatment based on the adsorption-reduction model according to claim 1, characterized in that, Using a standardized water quality vector as the initial state boundary and a preset hexavalent chromium concentration removal rate as the treatment objective, a multi-objective iterative optimization based on constraints is performed on an adsorption-reduction model instance to obtain the optimal set of process parameters, including: Based on standardized water quality vectors and treatment objectives, an optimization problem model and a vector of control variables to be optimized are constructed. The vector of control variables to be optimized is a vector composed of fulvic acid dosage concentration and set reaction pH value. The optimization problem model is expressed as follows: in, and This is the cost weighting coefficient. The concentration of fulvic acid to be added. To set the pH value for the reaction, The initial pH in the standardized water quality vector, Indicates the total cost; Based on the initial hexavalent chromium concentration in the standardized water quality vector, a global iterative optimization based on nonlinear programming is performed on the optimization problem model and the adsorption-reduction model instance to obtain the optimal set of process parameters.

6. The intelligent control method for mine water treatment based on the adsorption-reduction model according to claim 1, characterized in that, Based on the predicted proton consumption rate, closed-loop control of the reaction process is performed using the target reaction pH value and real-time pH feedback signal to obtain actuator commands, including: Based on the predicted proton consumption rate, the target reaction pH value and the real-time pH feedback signal are fused by a feedforward-feedback composite control signal to obtain the composite acid injection acceleration rate. The acceleration rate of the compound acid solution injection is converted into actuator instructions.

7. A smart control system for mine water treatment based on an adsorption-reduction model, characterized in that, include: The raw data acquisition module is used to acquire raw sensor data; The data standardization module is used to standardize the raw sensor data to obtain a standardized water quality vector. The target pH intelligent calculation module is used to intelligently calculate the target parameters of a standardized water quality vector based on the treatment target and an adsorption-reduction model instance to obtain the target reaction pH value, including: Based on a standardized water quality vector, an adsorption-reduction model instance is constructed, including: decoupling the state vector from the standardized water quality vector and extracting feature variables to obtain a set of state variables, including initial fulvic acid concentration, initial pH, and initial Fe ion concentration; performing kinetic coefficient inversion based on multivariate regression on the set of state variables to obtain the pseudo-second-order adsorption rate constant and equilibrium adsorption capacity; and constructing an adsorption-reduction model instance based on the predicted pseudo-second-order adsorption rate constant and predicted equilibrium adsorption capacity. The adsorption-reduction model instance is represented as follows: in, The pseudo-second-order adsorption rate constant. To balance the adsorption capacity, This is the current reaction time. This represents the predicted instantaneous adsorption capacity at time t. Using a standardized water quality vector as the initial state boundary and a preset hexavalent chromium concentration removal rate as the treatment target, the adsorption-reduction model instance is subjected to multi-objective iterative optimization based on constraints to obtain the optimal set of process parameters, which includes the target reaction pH value and the optimal equilibrium adsorption amount. The real-time pH acquisition module is used to acquire real-time pH feedback signals; The proton consumption rate prediction module is used to perform forward prediction of the proton consumption rate based on the amount of gangue added, the amount of fulvic acid added, the current reaction time, and the adsorption-reduction model instance to obtain the predicted proton consumption rate, including: performing forward prediction of the proton consumption rate using the following formula, wherein the formula is: in, Represents the predicted proton consumption rate, in mol / min; The dimensionless stoichiometric conversion factor represents the amount of stoichiometry consumed to reduce one mole of Cr(VI). Number of moles; This refers to the total mass of the adsorbent added. Here is the molar mass of chromium. Let be the instantaneous adsorption capacity at time t; It is the pseudo-second-order adsorption rate constant; To balance the adsorption capacity; This is the current reaction time; The closed-loop control execution module is used to perform closed-loop control of the reaction process based on the predicted proton consumption rate, the target reaction pH value, and the real-time pH feedback signal to obtain actuator commands.

Images