Mine water defluorination dosing intelligent control method adaptive to water quality fluctuation

Abstract

The present application belongs to the technical field of dosing control, and particularly relates to a mine water defluorination dosing intelligent control method suitable for water quality fluctuation, which comprises the following steps: obtaining water quality fluctuation intensity according to mine water quality historical data, and obtaining dynamic window length in combination with rated volume and real-time water inflow; obtaining a feature vector according to water quality state data in the dynamic window; clustering historical feature vectors to generate a clustering cluster, so as to obtain a reference dosing dose and a center coordinate vector; extracting minimum and second minimum distances according to the Euclidean distance between the feature vector and the center coordinate vector, and obtaining a boundary confidence degree according to the minimum distance; obtaining an inertial damping factor according to the reference dosing dose corresponding to the minimum distance and the boundary confidence degree; obtaining a fusion dose according to the minimum and second minimum distances and the corresponding reference dosing dose; and obtaining a smooth dosing dose according to the inertial damping factor, the fusion dose and the reference dosing dose. The present application realizes smooth dosing control.

Description

Technical Field

[0001] This invention relates to the field of chemical dosing control technology. More specifically, this invention relates to an intelligent control method for defluoridation dosing in mine water that adapts to fluctuations in water quality. Background Technology

[0002] In the daily operation of coal mine and industrial mine water treatment plants, the control of defluoridation dosing is a core aspect of ensuring effluent quality compliance and environmental safety in production. Its main task is to monitor the influent status and reaction process of the defluoridation mixing tank around the clock, continuously acquiring real-time water quality data. Based on this data, the controller adjusts actuators such as the defluoridation metering pump, so that when the influent water quality or quantity fluctuates, the dosing instructions can be adjusted promptly and smoothly. This prevents insufficient defluoridation from causing excessive fluoride levels in the effluent, or avoids increased operating costs and secondary water pollution caused by excessive dosing.

[0003] Existing mine water dosing systems typically employ feedback control logic or static condition matching methods, relying on fixed time periods to collect water quality parameters and outputting a baseline dosing dose by comparing the deviation between the current observed value and a preset threshold.

[0004] However, in actual industrial water treatment environments, the influent flow rate of mine water fluctuates frequently due to the scheduling of drainage from underground production. This causes the actual hydraulic residence time of the fluid in a fixed-volume defluoridation mixing tank to change continuously, resulting in distortion of the water quality data acquired by the control end when characterizing the chemical reaction cycle. Under high flow rate conditions, a fixed observation field leads to a significant hysteresis effect in the control signal; while under low flow rate conditions, feature extraction is prone to local bias due to insufficient coverage of the reaction cycle, leading to a shift in the calculation basis of the initial control command. Secondly, the water quality evolution environment in industrial settings is complex. The signals acquired by sensors often contain transient physical disturbances and chemical noise, and water quality characteristics often linger at the boundary between different pollution load conditions. Existing control technologies lack a risk assessment mechanism for water quality boundary conditions, making it difficult to distinguish between high-frequency, short-term water quality fluctuations and long-term, real water quality changes. Summary of the Invention

[0005] To address the technical problems of existing dosing systems exhibiting control benchmark distortion and response lag when facing hydraulic fluctuations, and frequent control signal jumps and actuator oscillations due to the lack of cross-condition boundary identification mechanisms, this invention provides an intelligent control method for mine water defluoridation dosing that adapts to water quality fluctuations. The method includes: collecting the rated volume and real-time influent flow rate of the defluoridation mixing tank; extracting historical data on mine water fluoride ion concentration, pH value, and turbidity within a fixed reference window at each time point, calculating their respective variances, normalizing them, and summing them to obtain the water quality fluctuation intensity at each time point; obtaining the dynamic window length at each time point based on the rated volume, real-time influent flow rate, and water quality fluctuation intensity; and calculating the mean and variance of mine water fluoride ion concentration, pH value, and turbidity within the dynamic window, and normalizing them. The system constructs a feature vector for the current moment; extracts features from historical data to generate historical feature vectors, and clusters them to generate clusters for each operating condition, obtaining the baseline dosage and center coordinate vector for each cluster; calculates the Euclidean distance between the feature vector at each moment and the center coordinate vector of each cluster, and obtains the boundary confidence based on the second smallest distance and the smallest distance; obtains the inertial damping factor based on the difference between the baseline dosage corresponding to the smallest distance at each moment and the baseline dosage at the previous moment, combined with the boundary confidence; obtains the fusion dosage at each moment based on the smallest distance, the second smallest distance, and their corresponding baseline dosages; obtains the smoothed dosage based on the inertial damping factor, the fusion dosage, and the baseline dosage; and completes the dosing operation based on the smoothed dosage.

[0006] This invention addresses the limitations of mine water defluoridation dosing control, including variable hydraulic reaction cycles, blurred boundaries between adjacent operating conditions, and asymmetric dosing risks. It calculates a dynamic window length by integrating real-time influent flow and water quality fluctuation intensity, thus measuring the adaptive time span for capturing mine water quality characteristics. This optimizes data extraction, which traditional fixed-time windows cannot cover the complete physicochemical reaction process. Furthermore, it calculates boundary confidence by extracting a distance feature set between feature vectors and cluster center coordinate vectors, measuring the degree of water quality characteristics' movement at the boundaries of multiple operating conditions and the risk of switching between operating conditions. This improves the control of mine water defluoridation dosing for single operating conditions. The accuracy of boundary identification during the critical period of fuzzy conditions is improved; an inertial damping factor is constructed by fusing boundary confidence to measure the damping adjustment of mine water defluoridation dosing control in the face of increasing and decreasing dosage commands, thereby enhancing the balance between the mechanical stability of defluoridation equipment and the safety of effluent water quality compliance; the inertial damping factor, fused dosage and benchmark dosage are used to perform attenuation adjustment to generate smooth dosage, and an analog control signal is output to the frequency converter of the defluoridation metering pump, which improves the phenomenon of frequent jumps in dosing control commands caused by free interference of water quality characteristics, and improves the smooth transition capability of defluoridation dosing control in the cross-condition switching stage under dynamic water quality environment.

[0007] Preferably, the length of the dynamic window satisfies the following relationship:

[0008] ;

[0009] In the formula, For a moment The dynamic window length; This is the rated volume; For a moment Real-time inflow rate; For a moment The intensity of water quality fluctuations; This is the precision bias constant; This is the window scaling factor; This is the preset minimum window length threshold; This is the rounding function; It is a function for maximizing the value.

[0010] This invention calculates the dynamic window length by integrating real-time mine water inflow and water quality fluctuation intensity, measuring the theoretical hydraulic residence time and adaptive feature extraction span of mine water in a preset defluoridation mixing tank. This provides an objective data benchmark for classifying mine water defluoridation conditions and constructing a dosing control feature set, optimizes the traditional fixed-time-window data extraction mode, improves the dosing control signal sampling collapse phenomenon caused by extreme flow velocity changes, enhances the feature capture agility under high-frequency disturbance conditions, and strengthens the dual adaptive alignment capability of feature extraction span and physical fluid dynamics period.

[0011] Preferably, the boundary confidence level satisfies the following relationship:

[0012] ;

[0013] In the formula, For a moment Boundary confidence level; For a moment The minimum distance; For a moment The second smallest distance.

[0014] This invention calculates boundary confidence by extracting the distance difference between the feature vector and the coordinate vector of the cluster center of the working condition, and measures the degree of free movement of mine water quality characteristics and the risk of control switching at the boundary of multiple working conditions. It provides a trigger indicator for judging whether the defluoridation dosing control is in a single working condition or a fuzzy critical period. It optimizes the identification logic of conventional distance judgment methods that cannot distinguish between short-term water quality fluctuations and long-term water quality changes, reduces the probability of abnormal jumps in dosing benchmarks caused by fuzzy working condition boundaries, and improves the matching accuracy of mine water defluoridation dosing control commands under the free state of water quality characteristics.

[0015] Preferably, the inertial damping factor satisfies the following relationship:

[0016] ;

[0017] In the formula, For a moment The inertial damping factor; For a moment The baseline dosage corresponding to the minimum distance; for The baseline dosage at any given time; It is an asymmetric adjustment base; For smoothing adjustment coefficient; For a moment Boundary confidence level; To prevent constants with a denominator of zero; sgn(x) is a symbolic function. When x > 0, sgn(x) = 1; when x = 0, sgn(x) = 0; when x < 0, sgn(x) = -1; where x is a dummy variable.

[0018] This invention constructs an inertial damping factor by integrating boundary confidence and asymmetric reagent risk logic. This factor measures the damping stiffness of mine water defluoridation dosing control in the face of increasing demand and decreasing trends. It provides a regulatory basis for outputting control commands that balance the stable operation of defluoridation metering equipment and the safety of water quality compliance. It optimizes the balance weight between mechanical operation stability and dosing response speed, improves the metering compensation lag caused by frequent oscillations in dosing commands in the ambiguous water quality boundary zone, reduces the risk of fluoride exceeding the standard caused by untimely dosing compensation, and enhances the mechanical tolerance of dosing control equipment to frequent dosing reduction fluctuations.

[0019] Preferably, the smooth drug dosing satisfies the following relationship:

[0020] ;

[0021] In the formula, For a moment Smooth drug dosing; For a moment The inertial damping factor; for The baseline dosage at any given time; For a moment The fusion dose.

[0022] This invention integrates the fused dosage obtained by cross-weighting with the baseline dosage, and uses the inertial damping factor to generate a smooth dosage by adjusting the time dimension of the execution. It measures the smooth transition of the control actuator during cross-condition switching and its alignment with real-time reagent demand, providing an objective control closed loop for outputting accurate analog control signals to the defluorination metering pump. This overcomes the control gap that cannot be matched with the real reagent demand in a fuzzy state by the baseline dosage under a single operating condition, optimizes the control command transition curve under cross-condition switching, reduces the probability of instantaneous flow oscillation of the defluorination metering pump, and improves the stability of reagent dosing control under dynamic water quality characteristics.

[0023] Preferably, the step of calculating the mean and variance of mine water fluoride concentration, pH value, and turbidity within a dynamic window, and constructing the feature vector for the current moment after normalization, includes: extracting mine water quality status data containing mine water fluoride concentration, mine water pH value, and mine water turbidity over a time span of the dynamic window; calculating the mean and variance of mine water fluoride concentration, mine water pH value, and mine water turbidity within the dynamic window length; and after normalizing the calculated mean and variance by maximum and minimum values, arranging and concatenating them to generate the feature vector for the current moment.

[0024] Preferably, obtaining the fusion dose at each time point based on the minimum distance, the second smallest distance, and their corresponding baseline dosage at each time point includes: calculating the product of the baseline dosage corresponding to the second smallest distance and the minimum distance, and the product of the baseline dosage corresponding to the minimum distance and the second smallest distance, respectively; using the sum of these two products as the numerator and the sum of the minimum distance and the second smallest distance as the denominator, and obtaining the fusion dose at the current time point by dividing by the product.

[0025] Preferably, the step of extracting features from historical data to generate historical feature vectors, clustering them to generate clusters for each operating condition, and obtaining the baseline dosage and center coordinate vector corresponding to each operating condition cluster includes: obtaining multiple historical feature vectors from historical data; clustering based on multiple historical feature vectors using the K-Means clustering algorithm to generate multiple operating condition clusters; arithmetically averaging the baseline dosages corresponding to the historical feature vectors contained in each operating condition cluster to obtain the baseline dosage for each operating condition cluster; and extracting the center coordinate vector of each operating condition cluster.

[0026] Preferably, the step of extracting historical data on fluoride concentration, pH value, and turbidity of mine water within a fixed reference window before each time point, calculating their respective variances, normalizing them, and then summing them to obtain the water quality fluctuation intensity at each time point includes: based on a preset fixed reference window, extracting historical data on fluoride concentration, pH value, and turbidity of mine water before the current time point; calculating the variances of the historical data on fluoride concentration, pH value, and turbidity of mine water respectively, and then summing them after normalization using the maximum and minimum values ​​to obtain the water quality fluctuation intensity at the current time point.

[0027] Preferably, the step of completing the dosing operation based on the smooth dosing dosage includes: using a controller to convert the smooth dosing dosage into an analog control signal and sending it to the frequency converter of the defluorination metering pump to control the dosing operation.

[0028] The beneficial effects of this invention are as follows: Addressing the limitations of mine water defluoridation control, such as variable hydraulic reaction cycles, ambiguous boundaries between adjacent operating conditions, and asymmetric dosing risks, this invention optimizes the data extraction mechanism that traditional fixed-time-window methods cannot cover the complete physicochemical reaction process by constructing a dynamic window length to measure the adaptive time span for extracting mine water quality characteristics. Furthermore, by calculating boundary confidence levels to measure the degree of water quality characteristics' free movement and the risk of cross-operating-condition switching at the boundary, this invention provides objective trigger indicators for defining single operating conditions or ambiguous zones, improving the accuracy of the control system in identifying operating condition changes. Finally, by fusing non-symmetrical... The system utilizes an inertial damping factor, based on a risk logic, to measure the adjustment stiffness of dosing control in the face of dosage increase and decrease commands. This provides a regulatory basis for outputting dosing commands that balance equipment stability and water quality safety, enhancing the balance between mechanical stability and control response. By using the inertial damping factor and fusion dosage to perform attenuation adjustment, a smooth dosing dosage is generated, and an analog control signal is output to the frequency converter of the defluoridation metering pump. This improves the frequent jumps in dosing control commands caused by water quality fluctuations, reduces the risk of fluoride exceedance due to lag in defluoridation agent metering response, and improves the stability and accuracy of dosing control during cross-condition switching. Attached Figure Description

[0029] Figure 1 This is a flowchart of an intelligent control method for defluoridation dosing in mine water that adapts to fluctuations in water quality;

[0030] Figure 2 This is a comparison chart of the dosage changes between the traditional and the strategies of this invention in the process of defluoridation of mine water. Detailed Implementation

[0031] 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.

[0032] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] This invention discloses an intelligent control method for defluoridation dosing in mine water that adapts to water quality fluctuations, referring to... Figure 1 This includes steps S1 to S4:

[0034] S1. Collect the rated volume and real-time influent flow rate of the defluoridation mixing tank, calculate the water quality fluctuation intensity to obtain the dynamic window length, and extract water quality status data to construct feature vectors and distance feature sets.

[0035] It should be noted that the mine water inflow rate fluctuates due to underground drainage, and the actual hydraulic reaction cycle within the defluorination mixing tank changes with the inflow velocity. These physical characteristics lead to a data extraction environment where water quality characteristics are distorted in mine water defluorination dosing control. This makes it impossible for traditional fixed time windows to cover the complete physicochemical reaction process, resulting in deviations in the determination of the baseline dosing dosage. Therefore, this invention combines the rated volume and real-time inflow rate to obtain a dynamic window length, performing adaptive adjustments during the feature extraction stage of mine water defluorination dosing control, providing a data foundation for classifying mine water defluorination operating conditions.

[0036] Specifically, the rated volume of the defluoridation mixing tank and the real-time influent flow rate at the current moment are collected; based on a preset fixed reference window, historical data on mine water fluoride concentration, pH value, and turbidity before the current moment are extracted; the variances of the historical data on mine water fluoride concentration, pH value, and turbidity are calculated respectively, and after normalization using the maximum and minimum values, they are summed to obtain the water quality fluctuation intensity at the current moment; the dynamic window length at the current moment is obtained based on the rated volume, real-time influent flow rate, water quality fluctuation intensity, and preset window scaling factor.

[0037] Using the dynamic window length as the time span, mine water quality status data including mine water fluoride concentration, mine water pH value, and mine water turbidity are extracted; the mean and variance of mine water fluoride concentration, mine water pH value, and mine water turbidity within the dynamic window length are calculated respectively; after normalizing the calculated mean and variance by maximum and minimum values, they are arranged and spliced ​​to generate the feature vector of the current time.

[0038] For example, the feature vector is (mean fluoride ion concentration, variance fluoride ion concentration, mean pH value, variance pH value, mean turbidity, variance turbidity).

[0039] The same feature extraction rules as described above are applied to historical data. The mean and variance of each item in the historical data are calculated, normalized, and concatenated to construct multiple historical feature vectors. Based on these historical feature vectors, the K-Means clustering algorithm is used to generate multiple operating condition clusters. The arithmetic mean of the baseline dosing doses corresponding to the historical feature vectors contained in each operating condition cluster is calculated to obtain the baseline dosing dose for each operating condition cluster. The center coordinate vector and the corresponding baseline dosing dose of each operating condition cluster are extracted. The Euclidean distance between the feature vector at the current moment and the center coordinate vector of each operating condition cluster is calculated to obtain the distance feature set.

[0040] Specifically, the dynamic window length satisfies the expression:

[0041] ;

[0042] In the formula, For a moment The dynamic window length; This is the rated volume; For a moment Real-time inflow rate; For a moment The intensity of water quality fluctuations; This is the precision bias constant; This is the window scaling factor; This is the preset minimum window length threshold; This is the rounding function; It is a function for maximizing the value.

[0043] in, The higher the value, the greater the real-time inflow rate. The smaller the value, the longer the theoretical hydraulic retention time of the mine water in the pre-set defluoridation mixing tank, resulting in a longer dynamic window length for the calculation output. The longer the time, the more information on the evolution of mine water quality is included by expanding the time search range, thus suppressing the one-sidedness of feature extraction caused by the narrow sampling range; The smaller the value, the higher the real-time inflow rate. The larger the value, the faster the mine water passes through the pre-set defluoridation mixing tank, thus increasing the dynamic window length of the calculation output. The shorter the window, the more adaptively the feature extraction span of mine water defluoridation dosing control shrinks as the real-time influent flow rate increases, effectively improving the feature capture agility under high flow rate conditions and reducing the signal lag effect caused by excessively long dynamic window length.

[0044] The data-driven water quality fluctuation feedback correction term is used to characterize water quality fluctuations when there are severe fluctuations within a fixed reference window. Increasing the value makes the dynamic window length... Adaptive shrinkage enhances the sensitivity of feature capture under high-frequency disturbance conditions; when water quality is stable... The value decreases, resulting in a smaller dynamic window length. Adaptive stretching expands the time search range to include richer evolutionary information.

[0045] In this embodiment, the precision bias constant is 0.001. This can be modified later based on the actual situation.

[0046] For example, the window scaling factor determines the redundancy of statistical feature extraction, with an empirical range of [1.0, 1.5]. In this embodiment, the window scaling factor is set to 1.2. Implementers can set the window scaling factor according to the severity of fluctuations in mine water quality. For instance, when the frequency of water quality fluctuations is high, the window scaling factor can be appropriately reduced to improve the timeliness of feature capture; when water quality changes are relatively gradual, the window scaling factor can be appropriately increased to enhance the robustness of statistical features.

[0047] For example, the minimum window length threshold determines the minimum data sample size for feature extraction under extreme high flow rate conditions, with an empirical range of [5, 15]. In this embodiment, the minimum window length threshold is set to 10. Implementers can set the minimum window length threshold based on the frequency of extreme mine water flow rates and the sensor noise level. For instance, when the system sampling frequency is low and high control sensitivity is required, the minimum window length threshold can be appropriately reduced to decrease computational lag in data truncation; when the system background noise is high or there is interference from water quality spikes, the minimum window length threshold can be appropriately increased to avoid variance calculation failure caused by an extremely short window.

[0048] For example, the fixed benchmark observation window represents the fixed number of samples participating in the variance calculation, which determines the breadth of the reference field for calculating the water quality fluctuation intensity. The empirical range is [30, 60] sample points. In this embodiment, the fixed benchmark observation window is set to 50 sample points. Implementers can set the number of samples for the fixed benchmark observation window according to the evolution cycle characteristics of the overall mine water quality. For example, when the water quality is in a period of high-frequency and violent fluctuations and the system is required to have extremely fast trend perception, the number of samples can be appropriately reduced to improve the refresh rate and timeliness of fluctuation intensity assessment; when the water quality has long-wavelength slow drift or large background noise, the number of samples can be appropriately increased to enhance the representativeness and noise resistance of the fluctuation benchmark.

[0049] S2. Extract the minimum and second minimum distances from the distance feature set and calculate the boundary confidence of the mine water quality status at the current moment.

[0050] It should be noted that after mine water quality characteristics are clustered according to operating conditions, transient fluctuations often cause feature detachment at the boundaries of multiple clusters. Conventional distance determination methods cannot distinguish between short-term water quality fluctuations and long-term water quality changes. This characteristic leads to blurred boundaries between adjacent operating conditions in mine water defluoridation dosing control, making it impossible to measure the degree of detachment of water quality characteristics, thus causing abnormal jumps in control signals during dosing benchmark matching. Therefore, this invention calculates boundary confidence based on the distance feature set to measure the fluctuation risk of mine water quality status, providing a basis for judgment to suppress benchmark dosing dose jumps.

[0051] Specifically, the minimum and second minimum distances are extracted from the distance feature set; the boundary confidence at the current time is obtained based on the second minimum and minimum distances.

[0052] Specifically, the boundary confidence satisfies the expression:

[0053] ;

[0054] In the formula, For a moment Boundary confidence level; For a moment The minimum distance; For a moment The second smallest distance.

[0055] in, The larger the value, the deeper the feature vector at the current moment lies within a specific working condition cluster, indicating that the distribution of mine water quality characteristics is highly deterministic, thus increasing the boundary confidence. The closer the value is to 1, the clearer the current classification of mine water quality becomes, and the more likely the mine water defluoridation dosing control will remain under a single operating condition, thus ensuring the stability of the dosing control.

[0056] The smaller the value, the more the feature vector at the current moment is located in the middle boundary area between the two working condition clusters, thus increasing the boundary confidence. The closer the value is to 0, the more ambiguous the identification of the control characteristics of mine water defluorination dosing becomes, indicating that the mine water defluorination dosing control is in a critical risk period of operating condition switching, thus providing an accurate trigger indicator for the subsequent introduction of an inertial damping factor to suppress the jump in the benchmark dosing dose.

[0057] S3. Based on boundary confidence and asymmetric drug risk logic, calculate the risk bias term under the conditions of increased or decreased dosage, and generate the inertial damping factor to suppress boundary jumps.

[0058] It should be noted that the defluoridation reaction in mine water oscillates with the dosing command in the water quality boundary zone, presenting an asymmetric risk of insufficient dosage leading to excessive fluoride levels and excessive dosage in actual dosing. This physical characteristic results in a trade-off between mechanical stability and water quality compliance in mine water defluoridation dosing control, thus inducing lag in defluoridation agent metering response and excessive fluoride levels. Therefore, this invention generates an inertial damping factor based on a confidence sensitivity term, providing a regulatory basis for outputting dosing commands that balance equipment stability and water quality safety.

[0059] Specifically, the inertial damping factor at the current moment is calculated based on the preset asymmetric adjustment base, the preset smooth adjustment coefficient, the baseline dosage corresponding to the minimum distance at the current moment, the baseline dosage recorded by the system at the previous moment, and the boundary confidence level.

[0060] In one embodiment, the inertial damping factor satisfies the following expression:

[0061] ;

[0062] In the formula, For a moment The inertial damping factor; For a moment The baseline dosage corresponding to the minimum distance; for The baseline dosage at any given time; It is an asymmetric adjustment base; For smoothing adjustment coefficient; For a moment Boundary confidence level; To prevent constants with a denominator of zero; sgn(x) is a symbolic function. When x > 0, sgn(x) = 1; when x = 0, sgn(x) = 0; when x < 0, sgn(x) = -1; where x is a dummy variable.

[0063] in, Characterizing risk bias; when the baseline dosage is... Greater than the baseline dosage When this occurs, it indicates a current need for increased dosage, the sign function outputs 1, increasing the value of the risk bias term and thus the inertial damping factor. The decay causes the stiffness of the defluoridation dosing control in mine water to soften rapidly, making it more responsive to dosage increases and thus mitigating the risk of fluoride over-limits due to untimely dosage compensation; when the baseline dosage... Less than the baseline dosage When this occurs, it indicates a current trend towards reducing medication dosage, and the sign function outputs -1, decreasing the risk bias term and amplifying the damping effect, thus increasing the inertial damping factor. Maintaining a high level of dosing will make the control stiffness of defluoridation in mine water more rigid, thus locking in the high-dose dosing inertia to suppress frequent dosing fluctuations caused by water quality fluctuations.

[0064] in, By nonlinearly amplifying the damping effect under boundary ambiguity, the system achieves suppression and rapid response to water quality fluctuations. It ensures that the inertial damping factor is significantly reduced when the boundary confidence is high to improve the response speed, and that the inertial damping factor is significantly increased when the boundary confidence is low to suppress and control fluctuations.

[0065] The larger the value, the more stable the current operating conditions. The closer the value is to 0, The larger the value, the more... The closer the value is to 0, the damping of the mine water defluorination dosing control is released, thereby ensuring that the mine water defluorination dosing control is highly consistent with the real-time characteristics of the mine water under stable operating conditions. The smaller the value, the more likely the current mine water quality is to be in the fuzzy zone at the boundary of the working condition clusters, and the squared boundary confidence score. The closer it gets to 0, The smaller the value, the more... The closer the value is to 1, the greater the control damping of the mine water defluorination dosing control, thereby forcing the mine water defluorination dosing control to retain the baseline dosing dose at the previous moment, preventing frequent jumps in dosing commands caused by boundary oscillations.

[0066] In this embodiment, the constant to prevent the denominator from being zero is 0.001, which can be modified according to the actual situation.

[0067] For example, the asymmetric adjustment base determines the difference in risk preference between increasing and decreasing dosage for mine water defluoridation control. The empirical range is [0.1, 0.5]. In this embodiment, the asymmetric adjustment base is set to 0.3. Implementers can set the asymmetric adjustment base according to the stringency of the mine water effluent fluoride emission standards. For instance, when the emission index is close to the critical value and the penalty for exceeding the standard is extremely severe, the asymmetric adjustment base can be appropriately increased to accelerate the response speed of increasing dosage; when the reagent cost sensitivity is high, the asymmetric adjustment base can be appropriately decreased to balance defluoridation reagent consumption.

[0068] For example, the smoothing adjustment coefficient determines the nonlinear mapping strength of the inertial damping factor to the boundary confidence level, with an empirical value range of [1,2]. In this embodiment, the smoothing adjustment coefficient is set to 1.5. Implementers can set the smoothing adjustment coefficient according to the mechanical tolerance level of the defluorination agent metering pump. For instance, when the defluorination agent metering pump is susceptible to frequent adjustment losses, the smoothing adjustment coefficient can be appropriately increased to enhance the stiffness of the mine water defluorination dosing control; when extremely high real-time response capability is required, the smoothing adjustment coefficient can be appropriately decreased.

[0069] In another embodiment, the inertial damping factor satisfies the expression:

[0070] ;

[0071] In the formula, For a moment The inertial damping factor; For a moment The baseline dosage corresponding to the minimum distance; for The baseline dosage at any given time; It is an asymmetric adjustment base; For smoothing adjustment coefficient; For a moment Boundary confidence level; The function is a sign function. When x>0, sgn(x)=1; when x=0, sgn(x)=0; when x<0, sgn(x)=-1.

[0072] in, A higher value indicates that the current operating conditions may be approaching stability, leading to a decrease in the inertial damping factor. This reduction allows for adjustments to the dosing control stiffness to better suit the current water quality characteristics. The smaller the value, the more likely the current water quality is in a fuzzy zone at the boundary of the operating condition clusters, leading to an inertial damping factor. Move closer to the value 1 to suppress the jumping of the drug administration command.

[0073] In contrast to another embodiment of the present invention, In this embodiment, the structure, within the low confidence interval, is linearly mapped to... As it increases slightly from 0, its mapping value exhibits a constant linear growth, leading to... Reducing this weakens the ability to suppress fluctuations at the operating condition boundary; while By utilizing the low slope property of the squared term structure near the zero point, Maintaining a high level in the ambiguous zone enhances tolerance to data noise;

[0074] In the high confidence interval of this embodiment, due to the linear mapping in When it reaches its maximum value of 1, the value of its mapping term is fixed at 1, resulting in the damping factor... Unable to approach zero, it retains a certain degree of control inertia, resulting in response lag; and By using the collapse structure at the denominator, the value of the mapping term tends to infinity when the confidence level approaches 1, driving... The convergence to 0 achieved effective release of damping, reducing the risk of excessive fluoride levels due to delayed drug addition.

[0075] S4. The dual-condition characteristic dose of the spatial dimension is integrated and the inertial damping factor is introduced to smoothly adjust the execution time, calculate the smooth drug dosage, and drive the drug delivery mechanism.

[0076] It should be noted that when the characteristics of mine water quality become fluid at the boundary of multiple operating condition clusters, the baseline dosage for a single operating condition cannot match the actual reagent demand, which is in an ambiguous state. This physical characteristic leads to a command jump environment in defluoridation dosing control when switching between operating conditions, causing a control gap between two operating points, thereby inducing instantaneous flow oscillations in the defluoridation metering pump and inaccurate reagent dosing. Therefore, this invention provides a control basis for outputting a smooth dosing dosage that balances response speed and operational stability by performing smooth adjustment of the inertial damping factor in the time dimension.

[0077] Specifically, the baseline dosage corresponding to the minimum distance and the baseline dosage corresponding to the second minimum distance are extracted. Combining the minimum distance, the second minimum distance, the baseline dosage corresponding to the minimum distance, and the baseline dosage corresponding to the second minimum distance, an inverse distance weighted algorithm is used. The sum of the minimum and second minimum distances is used as the denominator, and the second minimum distance is used as the cross-weight numerator of the baseline dosage corresponding to the minimum distance. The minimum distance is also used as the cross-weight numerator of the baseline dosage corresponding to the second minimum distance. The two baseline dosages are then cross-weighted and summed to obtain the fused dosage at the current moment. When the defluorination dosing control is in its initial state, the fused dosage is assigned to the baseline dosage. When the defluorination dosing control is not in its initial state, the baseline dosage recorded by the system at the previous moment is obtained. Based on the inertial damping factor, the baseline dosage, and the fused dosage, the smoothed dosage at the current moment is obtained. The smoothed dosage is converted into an analog control signal using a controller and sent to the inverter of the defluorination metering pump to control the dosing operation.

[0078] Specifically, the smoothed dosing rate satisfies the expression:

[0079] ;

[0080] In the formula, For a moment Smooth drug dosing; For a moment The inertial damping factor; for The baseline dosage at any given time; For a moment The fusion dose.

[0081] in, The higher the value, the more likely the water quality characteristics are in a highly ambiguous boundary zone at the current moment, indicating that the mine water's classification under different operating conditions is unclear, which affects the smoothing of chemical dosing dosage. The more it tends to retain the baseline dosage weight from the previous moment, the more it suppresses the mechanical oscillation of the defluorination metering pump by locking the dosing inertia, thus achieving a smooth transition. The smaller the value, the more likely the water quality characteristics are within a single operating condition at the current moment, resulting in higher certainty in determining the mine water state and facilitating smooth dosage adjustments. The more it tends to match the fusion dose weight at the current moment, the more sensitive the control end is to real-time water quality characteristics, so as to ensure that the dosage of the reagent matches the actual defluorination needs in real time, and achieve smooth and stable control that adapts to water quality fluctuations.

[0082] The higher the value, the heavier the pollution load of the mine water, and the more likely it is to be in a high-fluoride condition. This requires adding more defluoridating agents to ensure the effluent meets standards, thus... Increase the dosage; The smaller the value, the better the current mine water quality or the lower the operating load, the less defluoridating agent is needed, thus reducing the need for defluoridation agents. The dosage was reduced to a lower level to avoid waste caused by excessive drug administration.

[0083] Figure 2 The figure shows a comparison of dosage changes between the traditional and the strategies of this invention during the defluoridation process of mine water. The traditional dosing control strategy uses a fixed time window to extract water quality characteristics, making it highly susceptible to fluctuations in influent water. Its dosing commands not only exhibit dense oscillations during stable periods but also tend to experience abnormal jumps and sharp drops during transitions between operating conditions. In contrast, this invention uses adaptive feature extraction to filter out high-frequency interference during stable operation, ensuring smooth and stable dosing commands and reducing equipment wear and fluctuating chemical consumption. During the ramp-up phase when demand increases, it achieves a rapid response with low lag, reducing the risk of excessive fluoride levels in mine water due to untimely compensation. In the ambiguity zone between dosage reduction and operating conditions, it identifies boundary fluctuations and adaptively amplifies control damping, exhibiting a gentle descent tail characteristic by locking the high-dose dosing inertia. Operational results show that this invention effectively balances steady-state noise reduction and transient prevention of exceeding standards, suppresses cross-operating-condition command jumps, and enhances the safety and stability control of the defluoridation dosing process while ensuring stable effluent water quality compliance.

Claims

1. A smart control method for defluoridation dosing in mine water that adapts to fluctuations in water quality, characterized in that, include: The rated volume and real-time influent flow rate of the defluoridation mixing tank are collected; historical data of mine water fluoride concentration, pH value, and turbidity within a fixed reference window before each time point are extracted, their respective variances are calculated, normalized, and then summed to obtain the water quality fluctuation intensity at each time point; the dynamic window length at each time point is obtained based on the rated volume, real-time influent flow rate, and water quality fluctuation intensity; the mean and variance of mine water fluoride concentration, pH value, and turbidity within the dynamic window are calculated, normalized, and then used to construct the feature vector for the current time point. Historical feature vectors are generated by extracting features from historical data and clustering them into clusters for each operating condition. The baseline dosage and center coordinate vector for each cluster are obtained. The Euclidean distance between the feature vector at each time step and the center coordinate vector of each cluster is calculated, and the boundary confidence score is obtained based on the second smallest and smallest distances. The inertial damping factor is obtained based on the difference between the baseline dosage corresponding to the minimum distance at each time step and the baseline dosage at the previous time step, combined with the boundary confidence score, satisfying the following: ； For a moment The inertial damping factor; For a moment The baseline dosage corresponding to the minimum distance; for The baseline dosage at any given time; It is an asymmetric adjustment base; For smoothing adjustment coefficient; For a moment Boundary confidence level; To prevent constants with a denominator of zero; sgn(x) is a symbolic function. When x > 0, sgn(x) = 1; when x = 0, sgn(x) = 0; when x < 0, sgn(x) = -1, where x is a dummy variable. The fusion dose at each time moment is obtained based on the minimum distance, the second smallest distance, and their corresponding baseline dosages. This includes: calculating the product of the baseline dosages corresponding to the second smallest distance and the minimum distance, and the product of the baseline dosages corresponding to the minimum distance and the second smallest distance, respectively; using the sum of these two products as the numerator and the sum of the minimum distance and the second smallest distance as the denominator, and dividing by the sum to obtain the fusion dose at the current time. A smooth dosing dose is obtained based on the inertial damping factor, fusion dose, and baseline dosing dose, satisfying the following: ； For a moment Smooth drug dosing; For a moment Fusion dose; Complete the dosing operation according to the smooth dosing dosage.

2. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The length of the dynamic window satisfies the following relationship: ； In the formula, For a moment The dynamic window length; This is the rated volume; For a moment Real-time inflow rate; For a moment The intensity of water quality fluctuations; This is the precision bias constant; This is the window scaling factor; This is the preset minimum window length threshold; This is the rounding function; It is a function for maximizing the value.

3. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The boundary confidence scores satisfy the following relationship: ； In the formula, For a moment Boundary confidence level; For a moment The minimum distance; For a moment The second smallest distance.

4. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The process of calculating the mean and variance of mine water fluoride concentration, pH value, and turbidity within a dynamic window, and constructing a feature vector for the current moment after normalization, includes: extracting mine water quality status data containing mine water fluoride concentration, pH value, and turbidity with the dynamic window length as the time span; calculating the mean and variance of mine water fluoride concentration, pH value, and turbidity within the dynamic window length; and after normalizing the calculated mean and variance by maximum and minimum values, arranging and concatenating them to generate the feature vector for the current moment.

5. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The process of extracting features from historical data to generate historical feature vectors, clustering them to generate clusters for each operating condition, and obtaining the baseline dosage and center coordinate vector corresponding to each operating condition cluster includes: obtaining multiple historical feature vectors from historical data; clustering based on multiple historical feature vectors using the K-Means clustering algorithm to generate multiple operating condition clusters; arithmetically averaging the baseline dosages corresponding to the historical feature vectors contained in each operating condition cluster to obtain the baseline dosage for each operating condition cluster; and extracting the center coordinate vector of each operating condition cluster.

6. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The process of extracting historical data on fluoride concentration, pH value, and turbidity of mine water within a fixed reference window prior to each time point, calculating their respective variances, normalizing them, and then summing them to obtain the water quality fluctuation intensity at each time point includes: based on a preset fixed reference window, extracting historical data on fluoride concentration, pH value, and turbidity of mine water prior to the current time point; calculating the variances of the historical data on fluoride concentration, pH value, and turbidity of mine water, and then normalizing them using the maximum and minimum values ​​before summing them to obtain the water quality fluctuation intensity at the current time point.

7. The intelligent control method for defluoridation dosing in mine water adapting to water quality fluctuations according to claim 1, characterized in that, The process of completing the dosing operation based on the smooth dosing dosage includes: using a controller to convert the smooth dosing dosage into an analog control signal and sending it to the frequency converter of the defluorination metering pump to control the dosing operation.

Images