Multi-system collaborative control method and system for mine water recycling

Abstract

This invention relates to the field of mine water circulation control technology, and particularly to a multi-system collaborative control method and system for mine water recycling. The method includes obtaining the initial suspended solids concentration in the influent of the pretreatment system and the actual transmembrane pressure difference in the membrane concentration and crystallization system; predicting the residual suspended solids concentration in the downstream pretreatment effluent based on the initial suspended solids concentration and historical dosage; extracting the acceleration of the transmembrane pressure difference change and the variance of the fluoride concentration to construct a cost function; substituting the obtained parameters into the cost function for optimization to obtain the current initial target dosage; and when the rate of change of the dosage corresponding to the initial target dosage exceeds a warning threshold, smoothly truncating the initial target dosage to output a smooth dosage command. This invention effectively suppresses defluoridator passivation and irreversible fouling of the membrane module caused by reagent escape.

Description

Technical Field

[0001] This invention relates to the field of mine water circulation control technology, and in particular to a multi-system collaborative control method and system for mine water recycling. Background Technology

[0002] Currently, in the environmental protection and water resource management system of large coal bases, the efficient centralized treatment and recycling of mine water is the core physical link to achieve zero emissions in mining areas. Taking large-scale mine water treatment plants as an example, they generally adopt a cascade series treatment process. The standard architecture typically includes three core systems: the first stage is a pretreatment system, which mainly removes large particulate suspended solids and colloids by adding chemical agents to coagulation sedimentation tanks; the second stage is a specialized defluorination system, which uses highly efficient defluorinating agents such as hydroxyapatite for targeted adsorption or chemical precipitation of fluoride ions; and the third stage is a membrane concentration and crystallization system, which uses reverse osmosis or nanofiltration membrane modules to achieve desalination, purification, and final crystallization of the water. However, due to different historical construction periods and the involvement of different environmental protection equipment suppliers, the existing three-stage mine water treatment systems are often built and operated independently by different manufacturers. The distributed control systems or automatic control systems of each water plant are physically isolated from each other, forming data silos. This results in a severe disconnect between the upstream and downstream data chains of the three-stage treatment and the daily operation and management chains, lacking a fully integrated real-time data flow.

[0003] Chinese patent document CN102749894B discloses an electrical control device and a mine water treatment system for underground coal mines. The control device includes a signal acquisition unit, an auxiliary control unit, a main control unit, and a parameter setting unit. The signal acquisition unit generates acquisition signals based on the operating status of the mine water treatment system. The auxiliary control unit is connected to the signal acquisition unit; the main control unit is connected to the auxiliary control unit; and the parameter setting unit is connected to the main control unit. This electrical control device is a control device for an underground mine water treatment system, capable of automatically controlling the operating status of various mine water treatment equipment according to set operating conditions to ensure the normal operation of the mine water treatment system.

[0004] In actual continuous production, the pretreatment system uses an automatic dosing device to add polyaluminum chloride and polyacrylamide. The supernatant effluent after sedimentation is directly used as the feed water for the dedicated defluorination system and membrane concentration and crystallization system through a cascaded pipeline network. Due to the lack of a global feedforward coordination mechanism across systems, when the initial water quality of the well water fluctuates drastically due to changes in the mining face, the pretreatment system can only passively rely on the turbidity detection at the effluent end of its own work station for delayed single-loop feedback regulation. This isolated and passive control mode leads to serious secondary disasters in real fluid dynamics and chemical reactions: on the one hand, when the dosing system experiences control lag in response to fluctuations in the feed water quality and when excessive dosing occurs, residual polymeric flocculants that have not fully cross-linked in the system will migrate downstream with hydraulic transport. These large molecules tend to physically adhere to the surface of hydroxyapatite in specialized defluorination systems, thereby blocking its microporous structure over a large area and obstructing the mass transfer channels of target ions. This leads to accelerated physical passivation and irreversible deactivation of the defluorinating agent. On the other hand, during the pretreatment stage, the floc rupture caused by mismatched initial dosage or subsequent flow field shearing can cause colloids and micro-nano-scale suspended matter that have not undergone effective gravity settling to directly penetrate the security filter components and deposit with the fluid at the retrieval surface of the subsequent reverse osmosis membrane, thus rapidly cross-linking to form a dense filter cake fouling layer. Summary of the Invention

[0005] To address the problem of isolated control nodes in existing cascade water treatment systems, which leads to severe lag and over-dosing risks in pretreatment stage chemical control and consequently causes uncrosslinked residual chemicals and suspended solids to escape downstream, this invention provides a multi-system collaborative control method and system for mine water recycling.

[0006] In a first aspect, the present invention provides a multi-system coordinated control method for mine water recycling, which adopts the following technical solution: A multi-system coordinated control method for mine water recycling includes the following steps: 1. A multi-system coordinated control method for mine water recycling, characterized by comprising the following steps: S1: Obtain the initial suspended solids concentration in the influent of the pretreatment system, and obtain the initial fluoride concentration in the influent of the dedicated defluorination system and the actual transmembrane pressure difference of the membrane concentration and crystallization system; S2: Predict the concentration of residual suspended solids in the pretreated effluent reaching the downstream based on the initial suspended solids concentration of the influent and the historical dosage. S3: Extract the acceleration of transmembrane pressure difference change and the variance of fluoride concentration based on the actual transmembrane pressure difference value and the initial fluoride concentration of the influent, respectively; construct a cost function that includes the pressure difference penalty weight coefficient, the concentration error penalty weight coefficient, the control increment penalty weight coefficient, and the water quality fluctuation penalty weight coefficient; substitute the acceleration of transmembrane pressure difference change, the residual suspended solids concentration of the pretreated effluent, the variance of fluoride concentration, and the historical dosing command value into the cost function for optimization, and calculate the current initial target dosing amount; S4: When the rate of change of the dosing rate corresponding to the initial target dosing amount exceeds the warning threshold, the initial target dosing amount is smoothly truncated to output a smooth dosing command, and the frequency command of the inverter of the downstream cascade booster pump is lowered according to the compensation delay parameter corresponding to the difference in the truncated dosing amount.

[0007] Compared to existing water treatment systems where control nodes are isolated and rely on delayed single-loop feedback, leading to problems such as excessive reagent dosage and insufficient reaction, this method uses cross-stage water quality feedforward prediction and constructs a cost function with multi-objective penalty weights including pressure difference, concentration, and water quality fluctuations for global optimization. When the source water quality fluctuates, it outputs smooth dosing commands in advance and links downstream frequency converters to reduce frequency compensation. This fundamentally prevents the escape of uncrosslinked residual reagents, which causes accelerated passivation of defluorination materials and irreversible fouling of reverse osmosis membranes, ensuring the stable operation of the entire process chain.

[0008] Preferably, the method for calculating the acceleration of the transmembrane pressure difference change is as follows: obtain the actual transmembrane pressure difference values ​​at the current time and at adjacent historical times; and calculate the acceleration of the transmembrane pressure difference change in the membrane concentration and crystallization system using a second-order backward differential discretization formula.

[0009] Compared to the traditional method of directly using the absolute value of transmembrane pressure difference, which is easily affected by steady-state drift caused by the aging of filter membrane components, this method accurately extracts the acceleration of transmembrane pressure difference change through a second-order backward differential discretization formula. This effectively removes background noise from complex industrial data, enabling the system to pinpoint the deterioration trend caused by filter cake layer pore compaction in water treatment membrane concentration and crystallization applications in real time. This provides an early and accurate warning of sudden membrane fouling and downstream flux decline.

[0010] Preferably, the method for predicting the concentration of residual suspended solids in the pretreated effluent reaching the downstream is as follows: construct an autoregressive moving average model with exogenous variables, and use the historical real-time dosing amount and the initial suspended solids concentration in the influent as the input sequence of exogenous variables; solve the parameter array using the least squares method with a small error-proofing coefficient; and calculate the predicted concentration of residual suspended solids in the pretreated effluent by adding the historical residual suspended solids concentration in the pretreated effluent, the historical dosing amount command value, and the initial suspended solids concentration in the influent and combining them with weighting coefficients.

[0011] Compared to conventional univariate time-series models, which cannot account for the limitations of long-term prediction failures caused by dynamic chemical dosing interventions in upstream workshops, this paper constructs an ARMAX model that uses historical dosing amounts and influent suspended solids concentrations as exogenous variables and is supplemented with a small error-proofing coefficient. This model accurately reflects the dynamic response of chemical agents after long-term flow in the physical pipeline network, providing a high-precision dynamic water quality prediction benchmark for downstream membrane separation systems to predict the risk of small suspended particles penetrating in advance and to formulate feedforward interception strategies.

[0012] Preferably, the method for calculating the current initial target dosage is as follows: normalize the transmembrane pressure difference change acceleration, the difference between the residual suspended solids concentration and the standard concentration in the pretreated effluent, the historical dosage command value, and the fluoride concentration variance; multiply each normalized parameter by the corresponding pressure difference penalty weight coefficient, concentration error penalty weight coefficient, control increment penalty weight coefficient, and water quality fluctuation penalty weight coefficient, and then sum them to calculate the cost function; minimize the cost function through rolling optimization using a bottom-level solver while satisfying physical boundary constraints, and output the calculated initial target dosage.

[0013] Compared to the problem of solution collapse caused by inconsistent parameter dimensions in conventional multi-objective control, this method performs normalization preprocessing on key parameters of different physical dimensions, such as differential pressure change acceleration and concentration difference, by performing offline calibration of extreme values. This allows each parameter to be smoothly mapped to the same dimension range before being multiplied by the corresponding weight. As a result, stable rolling optimization is achieved under the hard boundary constraint of the maximum flow rate of the physical equipment in the underlying solver, ensuring the best combined output of reagent dosing economy and sedimentation tank chemical balance.

[0014] Preferably, during the rolling optimization process, if the underlying solver fails to converge within a preset time, a timeout protection mechanism is triggered and the optimal solution of the previous cycle is output; if the calculated initial target dosage approaches or exceeds the set ratio of the rated flow limit for multiple consecutive control cycles, the output is truncated and the target dosage is locked to the steady-state average value to trigger an alarm.

[0015] Compared to traditional algorithms that rely solely on ideal state calculations and are susceptible to divergence and loss of control due to fluctuations in operating conditions in industrial settings, this system effectively avoids high-frequency extreme oscillations of equipment caused by occasional calculation anomalies or single parameter distortions by adding a timeout guarantee mechanism that outputs the optimal solution of the previous cycle when the solution fails to converge, and a dual error prevention strategy that triggers abnormal closed-loop circuit breakers and locks the steady-state mean when the dosing amount continuously approaches the flow limit. This improves the safety redundancy and control robustness of the water plant dosing system under all-weather operating conditions.

[0016] Preferably, the method for outputting a smooth dosing command and lowering the inverter frequency command of the downstream cascaded booster pump is as follows: the dosing rate change rate of the current cycle is calculated using a first-order backward differential; the initial target dosing amount is smoothed and truncated using a moving average filter to output a smooth dosing command with a limited slope; the truncated dosing amount difference is calculated, and the truncated dosing amount difference is multiplied by the water tank calibration conversion constant and the time step to calculate the compensation delay parameter, which is positively correlated with the truncated dosing amount difference; the pump displacement conversion ratio coefficient is multiplied by the compensation delay parameter to obtain the amount of frequency reduction in the inverter frequency command, so as to issue a frequency reduction and throttling command to the downstream cascaded booster pump.

[0017] Compared to the original MPC algorithm, which is prone to tearing the stable flow field in the sedimentation tank and causing floc rupture due to sudden changes in the output of dosing commands during step changes in operating conditions, the introduction of anti-impact distribution logic uses a moving average filter to truncate the command with a limited slope. The intercepted and reduced dosing gap is converted into a compensation delay parameter and sent to the downstream booster pump to perform frequency reduction and flow throttling. While mitigating the impact of sudden changes in dosing at the front end, it cleverly uses the hydraulic residence time of the pipeline network to make up for the dosing effect, and achieves dynamic hydrodynamic balance of the total treatment capacity of the entire plant.

[0018] Preferably, the multi-system collaborative control method further includes: calculating the comprehensive evaluation score of mine water recycling: retrieving the actual transmembrane pressure difference change acceleration, effluent compliance rate and chemical consumption statistics of the entire chain during the historical operating cycle; after removing outliers, using the analytic hierarchy process and entropy weight method to perform evaluation index weighting calculation on the operating status data matrix of the entire chain system, and obtaining the comprehensive evaluation score of mine water recycling.

[0019] Compared to the drawbacks of traditional control systems where strategies gradually deviate from optimal operating conditions after long-term equipment operation, this method automatically extracts real operating data from the entire chain in batches after a complete operating cycle and removes outliers. It then uses the analytic hierarchy process (AHP) and entropy weighting method to objectively reflect the weight relationships of different operating indicators and calculate a comprehensive evaluation score. This enables the quantitative diagnosis of implicit degradation processes such as mechanical wear of water pumps and natural aging of filter membrane pores within the system, laying a solid evaluation foundation for subsequent adaptive adjustment of strategies.

[0020] Preferably, the multi-system collaborative control method further includes: triggering a weight update mechanism when the comprehensive evaluation score of mine water recycling is continuously lower than the evaluation threshold; dynamically increasing the concentration error penalty weight coefficient in the cost function according to the linear mapping rule, proportionally reducing the control increment penalty weight coefficient, and adaptively adjusting the water quality fluctuation penalty weight coefficient; and setting a protection interval period after each update to avoid frequent updates.

[0021] Compared to the pain points of using fixed control parameters that cannot adapt to long-term water quality changes and equipment aging, a set of intelligent control closed loops with self-evolution capabilities has been established based on comprehensive evaluation results. When the evaluation score is lower than the threshold, the concentration error penalty is automatically increased and the control increment penalty is reduced according to the linear mapping rule. This allows the overall control model to dynamically evolve with the entire life cycle of the water plant hardware, effectively avoiding the continuous deviation of the overall energy efficiency and material consumption from the Pareto optimal point due to the solidification of parameters in the global system.

[0022] Preferably, during the process of obtaining the initial suspended solids concentration of the influent: a first-in-first-out queue buffer of a set depth is built in memory; when the incoming sensor value is detected to exceed the range limit or a communication interruption occurs, a state latching mechanism is triggered, and the nearest valid historical value in the queue buffer is used as the replacement.

[0023] Secondly, this invention provides a multi-system collaborative control system for mine water recycling, employing the following technical solution: The multi-system collaborative control system for mine water recycling includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the multi-system collaborative control method for mine water recycling described above is implemented.

[0024] The above-mentioned multi-system collaborative control method for mine water recycling is generated into a computer program and stored in a memory for loading and execution by a processor. This allows for the creation of a system based on the memory and processor, making it convenient to use.

[0025] The present invention has the following technical effects: This invention overcomes the shortcomings of traditional cascade systems, such as isolation and delayed closed-loop operation, by collecting multi-dimensional parameters of pretreatment, defluoridation, and membrane concentration systems and performing ARMAX water quality feedforward prediction. It then constructs an MPC cost function that comprehensively considers pressure difference, concentration, and water quality fluctuations to optimize the dosage, breaking through the data silos of each water plant. This effectively suppresses defluoridator passivation and irreversible fouling of membrane modules caused by chemical escape, and realizes full life cycle adaptive operation and global cost reduction and efficiency improvement of the mine water treatment system under complex operating conditions. Attached Figure Description

[0026] Figure 1 This is a flowchart of the multi-system collaborative control method for mine water recycling according to the present invention.

[0027] Figure 2 This is a schematic diagram of the drug dosing command interception and gap compensation of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] This invention discloses a multi-system coordinated control method for mine water recycling, referring to... Figure 1 This includes the following steps: Step S1: Cross-system multidimensional data acquisition and hidden derived parameter construction.

[0030] Based on the IoT Ethernet bus for water recycling deployed in the industrial site, the global IPC periodically sends trigger signals at a fixed sampling frequency of 1Hz to the PLC nodes at the bottom of the pretreatment system, the dedicated defluoridation system, and the membrane concentration and crystallization system. It reads the initial suspended solids concentration of the influent calculated from the online turbidimeter in the pretreatment influent pipeline, the initial fluoride concentration of the influent in the flow tank of the dedicated defluoridation system, and the current transmembrane pressure difference across the membrane concentration and crystallization system at both ends of the high-pressure pump. To prevent packet loss and outliers caused by electromagnetic interference or network congestion in the industrial site, a 60-depth FIFO queue buffer is forcibly built in the global IPC memory. When an incoming sensor value exceeds the range limit or a communication interruption occurs, a state latching mechanism is triggered, replacing the value with the nearest valid historical value in the queue to prevent subsequent divisions from having a zero denominator or algorithm divergence.

[0031] Since the absolute value of the transmembrane pressure drop in a conventional membrane concentration and crystallization system includes steady-state physical drift that occurs with the increasing age of the filter membrane module, in order to remove background noise and pinpoint the sudden worsening trend of membrane pore blockage, the acceleration of the transmembrane pressure drop change in the membrane concentration and crystallization system is extracted using a second-order backward differential discretization formula. This effectively filters out the slow drift phenomenon of the conventional pressure drop. The expression is as follows:

[0032] in, Indicates the current time The acceleration of the transmembrane pressure difference change in the membrane concentration and crystallization system; express The actual transmembrane pressure difference in the membrane concentration and crystallization system at any given time; express The actual transmembrane pressure difference in the membrane concentration and crystallization system at any given time; express The actual transmembrane pressure difference in the membrane concentration and crystallization system at any given time; Indicates the sampling time interval.

[0033] When the transmembrane pressure difference in the membrane concentration and crystallization system accelerates... When the concentration increases, it reflects the accelerated deterioration trend of irreversible compaction of the filter cake layer pores on the membrane surface, leading to a decline in downstream membrane flux, thus enabling accurate early warning of sudden membrane fouling.

[0034] For specialized defluoridation systems, a 10-minute timeframe is extracted along the historical timeline, using the current moment as the endpoint, to construct a sliding window for fluoride concentration data. The variance of the initial fluoride concentration sequence in the influent within the sliding window is then extracted. , when the variance An increase in this value reflects the degree of disturbance and turbulence in the current well water quality.

[0035] Step S2: Cross-stage water quality feedforward prediction based on ARMAX algorithm.

[0036] In physical pipe networks, there is a delay of several hours in the flow of water from the pretreatment system to the downstream. Since conventional univariate ARMA time-series models cannot incorporate dynamic chemical dosing interventions in the upstream workshop, long-term predictions are prone to failure. Therefore, this step constructs an autoregressive moving average model with exogenous variables (ARMAX) in the global IPC, using the historical real-time chemical dosing volume of the pretreatment system's dosing pumps and the initial suspended solids concentration of the downhole influent as independent exogenous variable input sequences. This enables feedforward prediction of the residual suspended solids concentration in the downstream pretreated effluent after chemical intervention.

[0037] Extract the initial suspended solids concentration and historical chemical dosage data of the influent cached in step S1, construct the exogenous variable input sequence, and solve the parameter array using the least squares method with Tikhonov regularization. To prevent division by zero or divergence anomalies caused by matrix ill-conditioning during the inversion operation, a small regularization error-proofing coefficient is forcibly set. , The method for obtaining the mean squared error of the model prediction is as follows: multiple sets of historical ill-conditioned matrix samples are collected in advance, and the statistical optimal solution is obtained by cross-validation.

[0038] The discretization prediction formula for ARMAX is as follows:

[0039] in, Indicates the predicted The concentration of residual suspended solids in the pretreated effluent reaches the downstream at any given time; express The concentration of residual suspended solids in the pretreated effluent at any given time; Represents the sliding window The historical dosage instruction value actually issued by the preprocessing system at any given time; Indicates the sliding window The initial suspended solids concentration of the downhole water at any given time; This represents the first element in the autoregressive coefficient array. Order weight coefficient; This represents the j-th order weight coefficient in the exogenous input array; This represents the weighting coefficients of the exogenous input of suspended matter obtained by fitting using the least squares method; This represents the Gaussian white noise sequence of the system; This parameter represents the physical delay time of the system pipeline. It is obtained by measuring the fluid residence time at different flow rates using isotope or dye tracing methods, and then using the least squares method to obtain the statistically optimal solution. Indicates the order of autoregression; Indicates the order of the moving average of the dosage; The moving average order of suspended solids concentration is represented by p. The methods for determining the autoregressive order p, the moving average order q of the dosage, and the moving average order r of the suspended solids concentration are as follows: Collect data on the suspended solids concentration of the pretreated effluent, the dosage, and the suspended solids concentration of the influent for 7 consecutive days, with the sampling period consistent with step S1. Perform differential stabilization on the data, iterate through the combinations of p, q, and r in the interval [1, 6], calculate the AIC (Akaike Information Criterion) value for each combination, and select the model order with the smallest AIC. For a typical mine water treatment system, the example orders are p=3, q=2, and r=2.

[0040] When the predicted concentration of residual suspended solids in the pretreated effluent An increase in concentration indicates poor flocculation and sedimentation in the pretreatment stage, leading to a large number of tiny suspended particles and unreacted reagents penetrating downstream. This allows for early prediction of potential membrane fouling risks and fluoride removal agent poisoning risks. Similarly, by replacing the output target vector, the model cascade outputs the predicted concentration of residual reagents reaching downstream at the next time step.

[0041] Step S3: Preliminary optimization of global dosing amount based on model prediction control.

[0042] Obtain the acceleration of the transmembrane pressure difference change in the membrane concentration and crystallization system extracted in step S1. With positive semivariance and the predicted concentration output from step S2. Subsequently, global IPC is substituted into the cross-workshop model predictive control (MPC) algorithm engine. This algorithm constructs a cost function with the joint optimization objective of minimizing the downstream membrane module pressure differential acceleration and maintaining dosing economics, while simultaneously introducing a front-end water quality turbulence penalty constraint mechanism. To prevent the direct hard fusion of unbounded variables and parameters with different dimensions from causing the underlying solver system to crash, the acceleration of the transmembrane pressure differential is evaluated before constructing the cost function. The difference between the concentration of residual suspended solids in the pretreated effluent and the standard concentration Historical dosage instructions and variance Perform Min-Max normalization preprocessing based on offline calibration extrema to map all variables to the [0,1] interval of the same dimension.

[0043] Subsequently, the cost function is constructed, with the expression:

[0044] in, Represents the cost function; This represents the differential pressure penalty weighting coefficient; This represents the concentration error penalty weighting coefficient; This indicates the weighting coefficient for controlling the incremental penalty; This represents the water quality fluctuation penalty weighting coefficient, with a value range of [0.1, 0.9]. The method for obtaining this coefficient is as follows: based on historical water quality step test data from the past 12 months of mine water, a genetic algorithm is used to find the Pareto optimal solution that minimizes the total global energy consumption and chemical consumption, and offline fitting is performed. For example, the value is [value missing]. ; Indicates the current time The normalized acceleration of the transmembrane pressure difference change in the membrane concentration and crystallization system; Indicates the predicted The normalized value of the difference between the concentration of residual suspended solids in the pretreated effluent downstream and the standard concentration, which is obtained from the surface water environmental quality standard. Indicates the control period The control increment of the dosage instruction after time-normalization.

[0045] When the control increment of the pretreatment system's dosage command within the control cycle When the volume increases, it reflects that the system is attempting to intervene in water quality by aggressively adjusting the dosing pump discharge rate, resulting in subsequent stages suffering from the impact and damage of a large amount of high molecular weight chemicals escaping, thereby achieving a dual safety limit on the dosing economy and the chemical stability of the physical system.

[0046] The MPC optimizer performs rolling optimization under the hard physical boundary constraint of satisfying the rated maximum flow rate of the dosing pump. If the underlying solver fails to converge within a preset 500ms, a timeout protection mechanism is triggered, immediately outputting the optimal solution of the previous cycle to ensure continuous control command flow. To compensate for error accumulation caused by the lack of a verification threshold, the system adds an adaptive multi-dimensional anomaly closed-loop circuit breaker mechanism: if the calculated initial target dosing amount is detected to approach or exceed 90% of the equipment's safe rated flow rate limit for three consecutive control cycles, the output is forcibly truncated and the target dosing amount is locked to the steady-state average of the past 24 hours. At the same time, a manual intervention alarm is triggered to the host computer, effectively suppressing algorithm divergence caused by a single feature anomaly. Finally, the initial target dosing amount of the pretreatment dosing equipment at the current time t is calculated and output. The unit is: .

[0047] Step S4: Adaptive throttling and dimensionality reduction compensation.

[0048] Under changing operating conditions, if the quality of the well water undergoes a sudden change, the MPC algorithm in step S3 will suppress the prediction error and output a sudden increase in the initial target dosage. Because such abrupt changes in dosing commands disrupt the established hydrodynamic flow field and chemical flocculation sedimentation equilibrium within the sedimentation tank, causing a large number of generated flocs to break up and escape under shear force, the original command cannot be used to drive the equipment directly. This step introduces anti-shock distribution logic, which smooths out abrupt changes in commands and compensates for the reduced efficacy gap by adjusting the hydraulic residence time in the downstream pipe network, thereby mitigating the abrupt dosing changes while maintaining a balance in the total treatment capacity.

[0049] Specifically, the initial target dosage of the pretreatment dosing equipment output from step S3 in the current cycle. Perform a first-order backward difference to calculate the rate of change of the dosing rate. The unit is: , .in, This represents the initial target dosage of the pretreatment dosing equipment at time t; This indicates that the pretreatment dosing equipment is in The initial target dosage at any given time; The control cycle time step is represented by the following method: multiple sets of mechanical response delay test samples of dosing pumps are collected in advance, and their mathematical expectation value is obtained by fitting with Gaussian distribution. The equipment safety dead zone margin is then added as the final control cycle.

[0050] When the rate of change of drug administration An increase in the warning threshold indicates that the front-end dosing equipment has performed a high-frequency, pulse-like destructive dosing action, causing physical shear disruption to the stable hydrodynamic flow field inside the flocculation sedimentation tank. This allows for the quantitative interception of unreasonable control actions. In this implementation, the warning threshold is set as follows: ,when When the absolute value exceeds the warning threshold, the following adaptive compensation mechanism is triggered.

[0051] Front-end smoothing: A one-dimensional moving average filter with a window size of 3×1 is constructed in memory, and the moving average filter is used to adjust the initial target dosage for the current cycle. Smoothing and denoising are performed to output a final smooth dosing command with a limited slope. The signal is transmitted as a 4-20mA current signal to the metering and dosing pump in the pretreatment workshop via the PLC's analog output module.

[0052] Back-end throttling cascade: Real-time calculation of the difference in drug dosage cutoff Based on the constant integral topological mapping of pool volume and flow rate, the compensation delay parameter required to compensate for the drug's efficacy is calculated. .in, Indicates the compensation delay parameter; Indicates the control cycle time step; This represents the calibration conversion constant for the water tank, which is obtained by physical modeling and calibration of the geometric volume and rated flow velocity of the water tank. For example, the value is 15 s / L. This represents the difference in dosage that was cut off. The constant 3600 in the formula is used to convert the units: Converted to .

[0053] The global IPC sends frequency reduction and throttling commands to the dedicated defluorination system and membrane concentration and crystallization system via the bus, lowering the frequency of the inverters of the downstream cascaded booster pumps. ,in, The pump displacement conversion ratio is set to 25Hz to prevent the motor cooling fan from stalling and burning out or the pipeline from losing pressure and disconnecting due to excessively low pump speed.

[0054] Step S5: System evaluation and multi-level closed-loop update based on AHP-entropy weight method.

[0055] After an 8-hour physical operation cycle per shift, the global IPC retrieves detailed statistics on the transmembrane pressure difference acceleration, effluent compliance rate, and chemical consumption of the entire membrane concentration and crystallization system from the relational database. Since the fixed control strategy gradually deviates from the actual physical optimum due to pump wear and natural aging of the filter membrane pores, this step introduces a closed-loop update mechanism. The Analytic Hierarchy Process (AHP) and entropy weighting method are used to assign evaluation index weights to the extracted full-chain system operation status data matrix, calculating the comprehensive evaluation score for the current cycle of mine water recycling. This achieves self-adaptation and dynamic energy reduction of the control strategy throughout the entire lifecycle of the hardware equipment. The data matrix used for entropy weighting calculations is derived from sensor records within the past 8-hour physical operating cycle. Outlier filtering is performed before evaluation: data segments with transmembrane pressure differential acceleration exceeding ±5 standard deviations or effluent concentration fluctuations exceeding 200% of the standard limit for more than 10 minutes are removed. The AHP judgment matrix is ​​independently completed by at least three operation experts using the 1-9 scale and then geometrically averaged. The overall evaluation score is calculated as follows. The calculation cycle is per production shift, for example, each production shift is 8 hours. Only when... MPC weight updates are only triggered when the weights are below the threshold of 0.6 for two consecutive periods. After a single update, there must be an interval of at least 3 periods before another update can be performed to avoid frequent weight oscillations caused by random fluctuations in the feedback signal.

[0056] When the comprehensive evaluation score When the value decreases, it reflects that the solidified model parameters currently used by the system can no longer adapt to the physical aging process of the on-site hardware and the long-term changes in water quality, resulting in the overall energy efficiency and material consumption of the global treatment process continuously deviating from the Pareto optimal point, thereby enabling full tracking and quantitative diagnosis of the system performance degradation cycle.

[0057] The overall evaluation score After normalization, the result is fed back to the MPC engine in step S3 as a global penalty factor. When When the evaluation threshold is lower than the preset threshold, the weight of the cost function in step S3 is dynamically increased according to the linear mapping rule. At the same time, the weights are reduced proportionally. And adaptively adjust the water quality fluctuation penalty weight. To prevent excessive control, the weight adjustment range is forcibly limited to ±10% at a time, thus forming a multi-system intelligent control closed loop with self-evolution capabilities.

[0058] For example, the comprehensive evaluation score If the value drops by 8% and falls below the preset evaluation threshold of 0.6, then the weight is adjusted. Increase by 8% to make the weight Reduce by 8%, then adaptively and dynamically adjust the weights. This ensures that the sum of all weight coefficients remains unchanged before and after adjustment.

[0059] The technical effects of this invention can also be illustrated in conjunction with the accompanying drawings. Figure 2 This diagram illustrates the interception and gap compensation of dosing commands. The horizontal axis represents the control cycle nodes of the underlying logic, and the vertical axis represents the given value for dosing control. The diagram shows a highly oscillating dotted line with large jumps and scattered markers, representing the initial target dosing command with extreme destructive pulses initially calculated by the system when facing a step change in operating conditions. Directly executing this command would inevitably destroy the fragile flocculation flow field already constructed in the pool through physical shear force. A smooth, continuous, thick solid line interspersed and aligned with the oscillating line represents the final safe dosing command actually issued by the system to the underlying equipment after moving average truncation and physical boundary limiting. Its trend is significantly smoother, successfully avoiding high-frequency pulses. When the oscillating dotted line breaks through the thick solid line, creating a large drop, the filling area enclosed between the two visually and accurately quantifies the difference in efficacy forcibly intercepted and reduced by the system.

[0060] This invention also discloses a multi-system collaborative control system for mine water recycling, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the multi-system collaborative control method for mine water recycling according to this invention is implemented.

[0061] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0062] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-system coordinated control method for mine water recycling, characterized in that, Including the following steps: S1: Obtain the initial suspended solids concentration in the influent of the pretreatment system, and obtain the initial fluoride concentration in the influent of the dedicated defluorination system and the actual transmembrane pressure difference of the membrane concentration and crystallization system; S2: Predict the concentration of residual suspended solids in the pretreated effluent reaching the downstream based on the initial suspended solids concentration of the influent and the historical dosage. S3: Extract the acceleration of transmembrane pressure difference change and the variance of fluoride concentration based on the actual transmembrane pressure difference value and the initial fluoride concentration of the influent, respectively; construct a cost function that includes the pressure difference penalty weight coefficient, the concentration error penalty weight coefficient, the control increment penalty weight coefficient, and the water quality fluctuation penalty weight coefficient; substitute the acceleration of transmembrane pressure difference change, the residual suspended solids concentration of the pretreated effluent, the variance of fluoride concentration, and the historical dosing command value into the cost function for optimization, and calculate the current initial target dosing amount; S4: When the rate of change of the dosing rate corresponding to the initial target dosing amount exceeds the warning threshold, the initial target dosing amount is smoothly truncated to output a smooth dosing command, and the frequency command of the inverter of the downstream cascade booster pump is lowered according to the compensation delay parameter corresponding to the difference in the truncated dosing amount.

2. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The method for calculating the acceleration of the transmembrane pressure difference change is as follows: obtain the actual transmembrane pressure difference values ​​at the current moment and at adjacent historical moments; The acceleration of the transmembrane pressure difference change in the membrane concentration and crystallization system was calculated using a second-order backward differential discretization formula.

3. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The method for predicting the concentration of residual suspended solids in the pretreated effluent reaching the downstream is as follows: an autoregressive moving average model with exogenous variables is constructed, and the historical real-time dosing rate and the initial suspended solids concentration of the influent are used as the input sequence of exogenous variables; the parameter array is solved by the least squares method with a small error-proofing coefficient; the historical residual suspended solids concentration in the pretreated effluent, the historical dosing rate command value, and the initial suspended solids concentration of the influent are added together and combined with the weighting coefficient to calculate and predict the concentration of residual suspended solids in the pretreated effluent.

4. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The method for calculating the current initial target dosage is as follows: normalize the transmembrane pressure difference acceleration, the difference between the concentration of residual suspended solids in the pretreated effluent and the standard concentration, the historical dosage command value, and the variance of fluoride concentration. Multiply each normalized parameter by the corresponding pressure difference penalty weight coefficient, concentration error penalty weight coefficient, control increment penalty weight coefficient, and water quality fluctuation penalty weight coefficient, and then sum them to calculate the cost function. Under the physical boundary constraints, the cost function is minimized by rolling optimization through the bottom-level solver, and the calculated initial target dosage is output.

5. The multi-system coordinated control method for mine water recycling according to claim 4, characterized in that, During the rolling optimization process, if the underlying solver fails to converge within a preset time, the timeout protection mechanism is triggered and the optimal solution of the previous cycle is output. If the calculated initial target dosage approaches or exceeds the set ratio of the rated flow limit for multiple consecutive control cycles, the output is truncated and the target dosage is locked to the steady-state average value to trigger an alarm.

6. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The method for outputting a smooth dosing command and lowering the inverter frequency command of the downstream cascaded booster pump is as follows: The dosing rate change rate of the current cycle is calculated using a first-order backward differential; the initial target dosing amount is smoothed and truncated using a moving average filter, resulting in a smooth dosing command with a limited slope; the truncated dosing amount difference is calculated, and the truncated dosing amount difference is multiplied by the water tank calibration conversion constant and the time step to obtain the compensation delay parameter, which is positively correlated with the truncated dosing amount difference; the pump displacement conversion ratio coefficient is multiplied by the compensation delay parameter to obtain the amount of frequency reduction in the inverter command, which is then used to issue a frequency reduction and throttling command to the downstream cascaded booster pump.

7. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The multi-system collaborative control method also includes: calculating the comprehensive evaluation score of mine water recycling: retrieving the actual transmembrane pressure difference change acceleration, water discharge compliance rate and chemical consumption statistics of the entire chain during the historical operation cycle; after removing outliers, using the analytic hierarchy process and entropy weight method to perform evaluation index weighting calculation on the entire chain system operation status data matrix, and obtaining the comprehensive evaluation score of mine water recycling.

8. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, The multi-system collaborative control method also includes: triggering a weight update mechanism when the comprehensive evaluation score of mine water recycling is continuously lower than the evaluation threshold; dynamically increasing the concentration error penalty weight coefficient in the cost function according to the linear mapping rule, proportionally reducing the control increment penalty weight coefficient, and adaptively adjusting the water quality fluctuation penalty weight coefficient; and setting a protection interval period after each update to avoid frequent updates.

9. The multi-system coordinated control method for mine water recycling according to claim 1, characterized in that, During the process of obtaining the initial suspended solids concentration of the influent: a first-in-first-out queue buffer with a set depth is built in memory; when the incoming sensor value is detected to exceed the range limit or a communication interruption occurs, a state latching mechanism is triggered, and the nearest valid historical value in the queue buffer is used as the replacement.

10. A multi-system collaborative control system for mine water recycling, characterized in that, include: The processor and memory, wherein the memory stores computer program instructions, which, when executed by the processor, implement the multi-system collaborative control method for mine water recycling according to any one of claims 1-9.

Images