A system for collaborative optimization of energy consumption and quality of a barite powder production line based on digital twinning

Abstract

The application relates to the technical field of industrial process control, and discloses a system for energy consumption and quality collaborative optimization of a barite powder production line based on digital twinning, which collects the grinding feed amount, the mill active power, the feed characteristic particle size, the finished product characteristic particle size and the mill cavity differential pressure, reads the execution instructions of a feeder, a classifier and a circulating fan to construct a digital twin operation frame; the material residence time is determined according to the mill cavity differential pressure and the grinding feed amount, the finished product characteristic particle size is aligned with the historical material state, and the specific grinding energy and the effective grinding constant are obtained; the energy consumption, the particle size and the differential pressure state of the candidate execution vector are predicted through digital twin virtual disturbance, the optimized execution vector is determined within the particle size and differential pressure constraints, and control is implemented. The application can reduce the feedback time lag caused by the grindability of raw materials and the fluctuation of the feed particle size, enable energy consumption adjustment and particle size control to be based on the coordinated execution of the same batch of material states, and improve the stability of continuous production.

Description

Technical Field

[0001] This application relates to the field of industrial process control technology, specifically to a digital twin-based energy consumption and quality synergistic optimization system for a barite powder production line. Background Technology

[0002] The dry production of barite typically involves continuous processes including quantitative feeding, grinding, air classification, finished product collection, and circulating ventilation. Existing production lines' automated control usually adjusts operating conditions based on feed rate, mill load or active power, differential pressure in the grinding chamber or pipelines, classifier speed, fan frequency, and online particle size analysis results. Because feeding, grinding, classification, and pneumatic conveying are interconnected, changes in the same execution quantity can simultaneously affect the mill chamber's material holding capacity, particle refinement, classification and cutting status, and circulating load. Therefore, the control system needs to consider output, energy consumption, equipment operating status, and finished product particle size simultaneously.

[0003] During continuous production, the grindability and feed particle size of barite raw materials fluctuate normally depending on the batch of ore, crushing state, and feed composition. From entering the mill to forming the finished product, the material undergoes grinding, conveying, classification, and coarse powder return. Due to the influence of feed holdup and circulating load, the current energy consumption signal and the current finished product particle size signal may correspond to materials entering the system at different times. Directly adjusting the feed based on the current load and then correcting the classification or airflow based on the lagging particle size results easily leads to control compensation misalignment and repeated adjustments to the execution quantity, causing simultaneous fluctuations in unit product energy consumption and finished product particle size. Therefore, it is difficult to maintain coordinated and stable energy consumption control and quality control in continuous barite powder production. Summary of the Invention

[0004] This application provides a digital twin-based energy consumption and quality collaborative optimization system for a barite powder production line, in order to solve the technical problem of collaborative control mismatch caused by the time discrepancy between energy consumption status and finished product particle size feedback in continuous barite powder production.

[0005] The first aspect of this application provides a digital twin-based energy consumption and quality co-optimization system for a barite powder production line, comprising a controller and a data acquisition device and an execution device respectively connected to the controller. The controller is used to run the digital twin model and includes the following functional modules: The data frame construction module collects the feed rate, mill active power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure, reads the execution instructions of the feeder, classifier and circulating fan, assembles the execution vector in a fixed order, and constructs a digital twin operation frame according to the same timestamp; The material delay calculation module calculates the material delay based on the mill cavity differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and differential pressure delay coefficient. The grinding constant calculation module reads the mill's active power, feed rate, and feed characteristic particle size at historical moments based on material lag time, aligns them with the finished product characteristic particle size, and determines the specific grinding energy and effective grinding constant. The twin response prediction module applies a virtual perturbation to the execution vector, determines the sensitivity matrix based on the effective grinding constant, and obtains the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. The execution vector optimization module determines the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy, under the constraints of the qualified particle size range and the differential pressure safe range. The optimized control execution module breaks down the optimized execution vector into optimized feed rate, optimized classifying speed, and optimized fan frequency, which control the feeder, classifier, and circulating fan respectively.

[0006] The second aspect of this application provides a method for synergistic optimization of energy consumption and quality in a barite powder production line based on digital twins, comprising the following steps: Step S1: Collect the feed rate, mill power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure; read the execution instructions of the feeder, classifier and circulating fan; form an execution vector in a fixed order; and construct a digital twin running frame according to the same timestamp. Step S2: Calculate the material delay time based on the mill chamber differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and pressure difference delay coefficient. Step S3: Based on the mill's active power, feed rate, and feed characteristic particle size at historical moments of material lag time, align them with the finished product characteristic particle size to determine the specific grinding energy and effective grinding constant. Step S4: Apply virtual perturbation to the execution vector, determine the sensitivity matrix based on the effective grinding constant, and obtain the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. Step S5: Under the constraints of the qualified particle size range and the differential pressure safety range, determine the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy; Step S6: The optimized execution vector is split into optimized feed rate, optimized classifying speed and optimized fan frequency, and the feeder, classifier and circulating fan are controlled respectively.

[0007] The third aspect of this application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the aforementioned method for co-optimizing energy consumption and quality in a barite powder production line based on digital twins.

[0008] The beneficial effects of this application are as follows: This application forms digital twin operational data by combining production line data and execution instructions with a unified time base, and determines material lag time based on grinding chamber differential pressure and feed status, enabling the current finished product characteristic particle size to establish a consistent relationship with the energy input and feed particle size status of the corresponding historical materials. Based on this, an effective grinding constant is formed, and the digital twin is used to locally predict the candidate execution states of the feeder, classifier, and circulating fan. Under the condition that the finished product particle size and grinding chamber differential pressure meet the constraints, the execution vector with the lower predicted specific grinding energy is selected. This reduces the interference of raw material property changes and process lag on control judgment, reduces unnecessary repeated grinding and mutual compensation adjustments, and coordinates energy consumption adjustment, particle size control, and equipment operation constraints within the same control cycle, improving continuous operating adaptability and operational stability. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0010] Figure 1 A schematic diagram of the functional modules of the energy consumption and quality collaborative optimization system for a barite powder production line based on digital twins provided in this application embodiment; Figure 2 This is a schematic diagram of the material delay calculation process provided in the embodiments of this application; Figure 3 A schematic diagram illustrating the application scenario of the energy consumption and quality collaborative optimization system for a barite powder production line based on digital twins, provided in this application embodiment. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0012] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0013] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0014] like Figure 1 As shown, a digital twin-based energy consumption and quality co-optimization system for a barite powder production line includes a controller and acquisition and execution devices connected to the controller. The controller is used to run the digital twin model and includes the following functional modules: The data frame construction module A1 collects the feed rate, mill active power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure, reads the execution instructions of the feeder, classifier and circulating fan, assembles the execution vector in a fixed order, and constructs a digital twin operation frame according to the same timestamp. The material delay calculation module A2 calculates the material delay based on the mill cavity differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and differential pressure delay coefficient. The grinding constant calculation module A3 reads the mill's active power, feed rate, and feed characteristic particle size at historical moments based on material lag time, aligns them with the finished product characteristic particle size, and determines the specific grinding energy and effective grinding constant. The twin response prediction module A4 applies a virtual perturbation to the execution vector, determines the sensitivity matrix based on the effective grinding constant, and obtains the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. The execution vector optimization module A5 determines the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy, under the constraints of the qualified particle size range and the differential pressure safe range. The optimized control execution module A6 breaks down the optimized execution vector into optimized feed rate, optimized classifying speed, and optimized fan frequency, which control the feeder, classifier, and circulating fan respectively.

[0015] In one embodiment of this application, the data frame construction module A1 is configured to perform: Step S101, the data acquisition device includes a quantitative feeding metering device, a mill power measuring device, a feed particle size detection device, a finished product online particle size detection device, and a mill cavity inlet and outlet pressure measuring device, which respectively obtain the feed rate into the mill, the active power of the mill, the characteristic particle size of the feed, the characteristic particle size of the finished product, and the differential pressure of the mill cavity. The above parameters are used as the data acquisition data. The characteristic particle size of the feed and the characteristic particle size of the finished product are both d90 particle size. Among them, the feed rate into the mill represents the mass flow rate of barite material entering the mill, which can be continuously obtained by a quantitative feeding metering device equipped with a weighing sensor and a speed detection unit, and the controller reads its actual feed feedback value.

[0016] The active power of the mill represents the active electrical power currently consumed by the mill drive. It can be output by measuring the three-phase electrical parameters by the power measuring device on the mill inlet side, or by reading the data from the mill drive device with active power measurement function.

[0017] The feed characteristic particle size represents the d90 particle size of the material entering the mill. It can be obtained by an online image particle size detection device installed at the mill inlet conveyor or discharge position, which collects representative material images and outputs them after the particle size analysis program is calibrated.

[0018] The characteristic particle size of the finished product represents the d90 particle size of the finished powder particle size distribution, which can be output by the online laser particle size detection device on the finished product sampling branch after sampling and dispersion.

[0019] The differential pressure of the grinding cavity represents the pressure difference between the inlet measuring point and the outlet measuring point of the grinding cavity. It can be obtained directly by connecting two pressure tapping points to a differential pressure transmitter, or by calculating the difference between two pressure measurements within the controller.

[0020] In step S102, the controller reads the execution instructions of the feeder, classifier and circulating fan from the internal instruction register in the current control cycle and forms an execution vector in a fixed order; The execution vector represents the control state data formed by arranging the execution instructions of the feeder, classifier, and circulating fan in a fixed order within the current control cycle. This data is used to maintain a one-to-one correspondence between the physical equipment instructions and the inputs of the digital twin model.

[0021] Specifically, the fixed order of execution instructions is set to feeder, classifier, and circulating fan during system initialization, and remains unchanged during the construction and splitting of execution vectors, candidate execution vectors, and optimized execution vectors.

[0022] Step S103: Write the collected data and execution vector into the digital twin runtime frame according to the same timestamp.

[0023] Specifically, the digital twin operation frame uses a fixed field structure to store a unified timestamp, feed rate, mill active power, feed characteristic particle size, finished product characteristic particle size, mill cavity differential pressure, and execution vector. The controller uses its own synchronous clock as the time reference, and each acquisition device includes the acquisition time when uploading data. For data updated within the same control cycle, it is written into the corresponding operation frame according to the control cycle boundary.

[0024] When the update cycle of the online particle size detection device is longer than that of other acquisition devices, the moment when the device completes a valid measurement is used as the timestamp for the particle size data, and this particle size data is written into the running frame that covers that moment, without backfilling future measurement results to earlier times. Thus, the energy, feed, particle size, and execution status in the historical cache have a traceable temporal relationship.

[0025] Before writing the runtime frame, the controller checks the communication status and data validity flags of the acquisition device. Data that is faulty, exceeds the instrument's effective range, or is diagnosed as invalid by the equipment will not be used as model update data for that control cycle.

[0026] In one embodiment of this application, such as Figure 2 As shown, the material delay calculation module A2 is configured to execute: Step S201: Under the condition that the equipment has completed maintenance and the characteristic particle size of the finished product is within the qualified particle size range, the differential pressure of the grinding chamber and the feed rate into the mill are recorded simultaneously, and the reference differential pressure and the reference feed rate are determined respectively. Specifically, the acceptable particle size range is determined according to the lower and upper limits of the allowable particle size of the finished product specified in the current production order. The benchmark operating condition is within the continuous operating load allowed by the equipment manufacturer, and meets the requirements that there are no protective alarms during feeding, mill operation, classification, and ventilation, and the characteristic particle size of the finished product is continuously within the acceptable particle size range.

[0027] During this stable period, the synchronously acquired differential pressure in the grinding chamber and the feed rate into the mill are averaged over time, and the results are used as the reference differential pressure and reference feed rate, respectively. The length of the stable period should at least cover several effective updates by the online particle size detection device, so that the reference values ​​are not determined by a single instantaneous fluctuation.

[0028] Step S202: Perform a feed step test, record the time from the change in feed to the appearance of a response in the characteristic particle size of the finished product, and determine the reference lag time; Specifically, the reference delay is calibrated through a feed step test. The feed step amplitude is an effective adjustment step size that the feeder can stably execute under the current operating conditions. The feed rate into the mill after the step is still within the allowable load range, and the execution commands of the classifier and circulating fan remain unchanged during the test.

[0029] The stable fluctuation band of the finished product characteristic particle size before the step change is used as the judgment benchmark. When multiple consecutive valid detection results of the finished product characteristic particle size exceed the stable fluctuation band in the same direction after the step change, the time interval between the first occurrence of a sustained response and the feeding step change time is determined as the response time. When equipment conditions permit, feeding increase and feeding decrease tests can be carried out separately, and the average value of each valid response time is used to determine the benchmark lag time to reduce the impact of single measurement fluctuations.

[0030] Step S203: Collect the differential pressure and response time of the grinding cavity under different working conditions within the allowable load range, fit the relationship between the change of differential pressure and the change of response time, and obtain the differential pressure time delay coefficient; Specifically, the differential pressure delay coefficient is calibrated within the allowable load range. Several stable feeding conditions are set up, and the differential pressure and response time of the grinding chamber are repeatedly obtained under each condition. Using the reference differential pressure and reference lag time as normalization references, the linear relationship between the relative change of differential pressure and the relative change of response time is fitted using the least squares method, and the slope is used as the differential pressure delay coefficient.

[0031] The fitting results are only used within the allowable load range covered by the calibration, and the calculated material lag time within this range is always greater than 0. If equipment maintenance causes changes to the mill cavity structure, airflow resistance, or the condition of key components, making the original calibration relationship no longer applicable to the allowable load range after maintenance, the above calibration should be performed again.

[0032] Step S204: Determine the differential pressure load ratio based on the differential pressure in the grinding chamber and the reference differential pressure, correct the reference lag time using the differential pressure delay coefficient and the differential pressure load ratio, and determine the material lag time based on the ratio of the reference feed rate to the feed rate into the mill.

[0033] Specifically, the formula for calculating the differential pressure load ratio is as follows: in, Indicates time The differential pressure load ratio is used to characterize the dimensionless deviation of the current grinding cavity differential pressure relative to the reference differential pressure; The differential pressure in the grinding chamber at any given time; Indicates the reference differential pressure; Indicates the current moment.

[0034] Specifically, the formula for calculating the material lag time is as follows: in, Indicates time Material lag time, i.e., the dynamic time interval from when the material corresponding to the current characteristic particle size of the finished product enters the grinding process to when the particle size response of the finished product is formed, is used to locate the historical energy and feeding state of the same batch of materials; Indicates the reference time delay; Indicates the differential pressure delay coefficient; Indicates the baseline feed rate; Indicates time The feed rate into the mill; in the formula and The meaning remains the same as before.

[0035] It should be noted that the differential pressure in the grinding chamber is composed of the pressure loss formed when the airflow passes through the grinding chamber and the powder within it. Within the calibration range corresponding to the same equipment structure and ventilation control conditions, changes in the material holding state will be reflected in changes in the differential pressure state; simultaneously, changes in the amount of material entering the mill per unit time will alter the time scale of the average material passage process. By using the response time calibrated under stable operating conditions as a benchmark, and then dynamically correcting it using the relative changes in the differential pressure in the grinding chamber and the changes in the feed rate relative to the benchmark value, the material lag time can be updated with the current load state. Therefore, the characteristic particle size of the finished product is no longer mechanically paired with the active power of the mill at the same moment, but rather established with a time correlation with the historical material state corresponding to its formation process, which can reduce the impact of time mismatches between energy feedback and quality feedback on control judgment when raw materials and load change.

[0036] In one embodiment of this application, the grinding constant calculation module A3 is configured to perform: Step S301: The digital twin historical cache performs time interpolation on the historical data according to the material lag time, and reads the mill active power, feed rate and feed characteristic particle size corresponding to the historical time when the current time is subtracted from the material lag time. Specifically, the digital twin history cache stores running frames that have passed validity checks according to the control cycle. The target historical moment typically does not completely coincide with a particular running frame's timestamp. The controller selects two adjacent valid timestamps before and after it, and performs linear time interpolation on the mill's active power, feed rate, and characteristic particle size of the feed. If the target historical moment matches a valid timestamp, the data corresponding to that timestamp is directly read. This avoids the additional alignment error that occurs over a complete control cycle when only the most recent sampling point is used.

[0037] Step S302: Determine the specific mill energy by the ratio of the mill's active power to the feed rate at a historical moment; Specifically, the formula for calculating the specific wear energy is as follows: in, Indicates time The specific grinding energy is the energy consumption per unit mass of ore after aligning with the completion time of the current batch of finished products. This indicates the active power of the mill at a historical moment. This indicates the feed rate into the mill at a historical moment. Indicates time Material delay; This indicates the current time. When the active power of the mill is expressed in kilowatts and the feed rate is expressed in tons per hour, the specific mill energy can be uniformly converted to kilowatt-hours per ton.

[0038] Step S303: Determine the effective grinding constant using the specific grinding energy, the current finished product characteristic particle size, and the historical feed characteristic particle size. The effective grinding constant is the ratio of the specific grinding energy to the difference between the reciprocal of the finished product characteristic particle size and the reciprocal of the feed characteristic particle size.

[0039] Specifically, the formula for calculating the effective grinding constant is as follows: in, Indicates time The effective grinding constant, which is the online process relationship between the specific grinding energy of the same batch of material and the fineness of the feed and finished product particle size, is used to comprehensively reflect the current difficulty of raw material fineness and the effective energy transfer state of the mill. This parameter is updated with the production conditions and is not used as a fixed material property. Indicates time The specific grinding energy; Indicates time The characteristic particle size of the finished product; The characteristic particle size of the feed at a given historical moment; Indicates time Material delay; Indicates the current moment.

[0040] Furthermore, before calculation, the particle size unit is standardized, and it is confirmed that the characteristic particle size of the feed used for batch alignment is greater than the current characteristic particle size of the finished product, the feed rate into the mill is greater than 0, and the relevant collected data are within the effective range of the instrument. Data frames that do not meet the above normal grinding conditions are not used to update the specific grinding energy and effective grinding constant. Updates are only performed after a valid operating frame that meets the conditions appears, in order to avoid online process parameters that have no physical meaning due to material stoppage, instrument malfunction, or abnormal particle size relationships.

[0041] In one embodiment of this application, the twin response prediction module A4 is configured to perform: Step S401: The digital twin model applies virtual disturbances to the execution commands of the feeder, classifier and circulating fan respectively, with the execution vector of the current control cycle as the center. The execution commands corresponding to the virtual disturbances are restricted to their respective allowed execution ranges. Specifically, the permissible execution range is determined according to the continuous operation limits of the feeder speed controller, classifier frequency converter, and circulating fan frequency converter, and is combined with the permissible range given by the equipment manufacturer, the on-site interlock limits, and the control range allowed by the current production plan to take the common permissible part.

[0042] The amplitude of the virtual disturbance is not less than the minimum effective adjustment step size that the corresponding actuator can distinguish and stably execute, while not exceeding the single-cycle allowable change amount already set by the control system. For the current execution vector, one effective disturbance point can be set in both the positive and negative directions for each execution quantity; when effective disturbances in both positive and negative directions cannot be formed due to the boundary constraints of the corresponding allowable execution interval, only disturbance points located within the allowable execution interval are used. Virtual disturbances are calculated only in the digital twin model and are not written into the physical actuator.

[0043] Step S402: For each group of virtual disturbances, the digital twin model calculates the local changes in specific grinding energy, stage reverse diameter correction amount and grinding cavity differential pressure based on the effective grinding constant, and determines the sensitivity matrix through local linear fitting. Specifically, the digital twin model can be a multi-input multi-output local response model established using production line debugging data and historical operating frames. Its inputs are executed by the feeder, classifier, and circulating fan in a fixed sequence, and the effective grinding constant is used as the current grinding state quantity; the outputs include at least the specific grinding energy, the classifying effect state, and the differential pressure in the grinding chamber.

[0044] During model calibration, operational data was selected based on normal data acquisition, no equipment alarms, and coverage of the permissible execution range. Modeling and validation data were then divided chronologically. The least squares method was used to identify local linear response coefficients, and data not used for identification were used to check the direction of output changes and prediction residuals. The identification data covered independent variations of the three execution commands, ensuring that the local response coefficients could be uniquely determined. If the equipment operating conditions exceeded the calibrated coverage range, the local response model was not extrapolated.

[0045] Furthermore, the inverse diameter correction is used to represent the change in particle size distribution caused by the classifier and circulating fan relative to the current operating conditions as an inverse particle size increment. The calibration samples of the local response relationship of the classification are constructed from valid historical operating frames.

[0046] For each sample, the measured finished product characteristic particle size is converted to its inverse particle size. The feed inverse particle size state, characterized by the corresponding feed characteristic particle size, and the grinding refinement amount, characterized by the specific grinding energy and effective grinding constant, are subtracted to obtain the classification effect for that sample. Then, using the classifier, circulating fan execution commands, and mill feed rate as influencing factors, the least squares method is used to fit the local response relationship, and the independent changes of each execution quantity are examined across the sample.

[0047] The digital twin model defines the relative increment at the current execution vector as the baseline value of the graded reverse diameter correction for the current control cycle, which is 0. For virtual disturbance conditions, the change in the finished product cutting state relative to the current condition is determined using the calibrated graded local response relationship, and then converted into the inverse particle size increment relative to the baseline value. Thus, the graded reverse diameter correction only describes the graded increment effect caused by the execution change, and does not repeatedly include the grinding and refining effect characterized by the effective grinding constant.

[0048] Step S403: Establish a local prediction relationship between the candidate execution vector and the predicted specific wear energy, the graded reverse diameter correction amount, and the predicted grinding cavity differential pressure based on the sensitivity matrix.

[0049] Specifically, the calculation formula for the local prediction relationship is as follows: in, This represents the predicted grinding energy corresponding to the candidate execution vector, that is, the unit mass grinding energy consumption predicted by the digital twin model under the action of the candidate execution vector; This represents the hierarchical reverse path correction amount corresponding to the candidate execution vector; This represents the predicted differential pressure in the grinding cavity corresponding to the candidate execution vector, i.e., the pressure difference between the inlet and outlet of the grinding cavity predicted by the digital twin model under the action of the candidate execution vector; Indicates time The specific grinding energy; Indicates time Differential pressure in the grinding cavity; Indicates time The sensitivity matrix; This represents a candidate execution vector, whose instruction order is consistent with the execution vector. This represents the execution vector of the current control cycle; This represents the baseline value for the graded reverse path correction in the current control cycle; Indicates the current moment.

[0050] Furthermore, the sensitivity matrix is ​​a three-row, three-column local mapping matrix. The three columns correspond to the changes in execution commands from the feeder, classifier, and circulating fan, respectively, while the three rows correspond to the specific grinding energy, the correction amount for the classifier's reverse diameter, and the change in grinding chamber differential pressure, respectively. On the current execution vector and its neighboring virtual disturbance samples, using the increments of the three execution commands as independent variables and the changes in the three outputs relative to the current operating condition as dependent variables, three sets of linear relationships are fitted using the local least squares method. The resulting coefficients are arranged in a fixed row and column order to form the sensitivity matrix. This matrix is ​​only used to evaluate candidate execution vectors within the calibrated local range corresponding to the current operating condition and is recalculated in the new control cycle based on the updated operating frame and effective grinding constant.

[0051] Step S404: Add the reciprocal of the characteristic particle size of the feed at the current moment, the ratio of the predicted specific grinding energy to the effective grinding constant, and the classification reverse diameter correction amount, and take the reciprocal to determine the predicted finished particle size.

[0052] Specifically, the formula for calculating the predicted particle size of the finished product is as follows: in, This indicates the predicted finished product particle size corresponding to the candidate execution vector, that is, the finished product d90 particle size obtained under the action of the candidate execution vector based on the current feed particle size state, the predicted grinding action, and the predicted classification action. Indicates time The characteristic particle size of the feed; This represents the predicted specific energy corresponding to the candidate execution vector; Indicates time The effective grinding constant; This represents the hierarchical reverse path correction amount corresponding to the candidate execution vector; Represents the candidate execution vector; Indicates the current moment.

[0053] It should be noted that the feeder, classifier, and circulating fan do not function independently on the production line. Changes in feed will alter the grinding chamber load and energy input per unit mass; changes in classifier speed will change particle classification conditions; and changes in the circulating fan will simultaneously affect particle carrying capacity and grinding chamber pressure. By performing virtual perturbations within a local range centered on the current execution vector and constrained by the corresponding allowable execution interval, and forming a sensitivity matrix, the coupled responses of the three execution variables to energy consumption, classification effect, and grinding chamber differential pressure can be described using the same local mapping. Furthermore, by characterizing the current raw material and grinding state with the effective grinding constant, the deviation generated by the fixed-condition mapping when the raw material properties change can be reduced. This mechanism allows candidate execution states to be locally evaluated without imposing trial actions on the physical equipment, thereby reducing mutual compensation adjustments.

[0054] In one embodiment of this application, the execution vector optimization module A5 is configured to execute: Step S501: Determine the qualified particle size range according to the allowable range of finished product particle size specified in the current production order, and determine the differential pressure safety range according to the mill manufacturer and process debugging. Specifically, the acceptable particle size range directly adopts the allowable range specified in the current production order for the characteristic particle size of the finished product, and is not replaced by the target range generated by the controller. The differential pressure safety range is first determined based on the continuous operating pressure difference limit provided by the mill manufacturer to define the equipment boundary, and then narrowed down by combining the pressure difference range that can maintain stable conveying, classification, and dust removal operations during on-site process debugging, ultimately adopting the continuous range that overlaps with both. When equipment maintenance or airflow structure adjustments cause changes in operating resistance, process debugging is re-confirmed.

[0055] Step S502: Generate candidate execution vectors within the allowable execution range of the feeder, classifier, and circulating fan that are currently available to the controller and match the current production plan; Specifically, candidate execution vectors can be discretized into allowable execution intervals according to the existing instruction resolution of each execution device or the single-cycle adjustment step size allowed by the controller, and the discrete values ​​of the three execution quantities are combined in a fixed order. To avoid unnecessary increases in the number of candidates, candidates are generated only with the current execution vector as the center, within the range where the change of each execution quantity relative to the current value does not exceed the allowable change in the corresponding single cycle. The current production plan is mainly used to limit the executable range corresponding to the feed rate into the mill; the classifier and circulating fan still need to simultaneously meet their respective equipment and process limitations.

[0056] Step S503: Screen candidate execution vectors that predict the finished product particle size to be within the qualified particle size range and the predicted grinding cavity differential pressure to be within the safe differential pressure range. Specifically, during the screening process, it is first checked whether the predicted finished product particle size and the predicted grinding chamber differential pressure simultaneously meet the corresponding ranges, and then the predicted grinding energy is compared among the candidates that meet the conditions.

[0057] Step S504: Determine the optimal execution vector from the selected candidate execution vectors based on the minimum predicted specific wear energy.

[0058] Specifically, the formula for calculating the optimized execution vector is as follows: in, Indicates time The optimized execution vector is one that simultaneously satisfies the finished product particle size constraint and the grinding chamber differential pressure constraint, and predicts the combination of feeder, classifier and circulating fan instructions with the minimum specific grinding energy among feasible candidates. Represents the candidate execution vector; This represents the predicted specific energy corresponding to the candidate execution vector; This indicates the predicted particle size of the finished product corresponding to the candidate execution vector; This represents the predicted differential pressure in the grinding cavity corresponding to the candidate execution vector; Indicates the acceptable particle size range; Indicates the safe range of differential pressure; Indicates the current moment.

[0059] Furthermore, if multiple candidate prediction ratios are the same at the numerical resolution set by the controller, the candidate with the smaller change relative to the current execution vector can be selected first to reduce unnecessary execution adjustments. If no candidate simultaneously satisfies both constraints in the current control cycle, no new optimized execution vector will be issued, the current execution instruction will be maintained, and the over-limit state will be handled by the original interlocking and alarm logic of the equipment; after a new valid operating frame is formed, prediction and screening will be re-executed.

[0060] In one embodiment of this application, the optimization control execution module A6 is configured to execute: Step S601, the execution device includes a feeder speed regulator, a classifier frequency converter and a circulating fan frequency converter. The controller splits and optimizes the execution vector according to the fixed order of the execution instructions of the feeder, classifier and circulating fan in the execution vector to obtain the optimized feed amount, optimized classifier speed and optimized fan frequency. Among them, the optimized feed rate represents the feed control target obtained by decomposing the optimized execution vector and provided to the feeder speed regulator. The optimized classifying speed represents the speed control target obtained by decomposing the optimized execution vector and provided to the classifier frequency converter. The optimized fan frequency represents the frequency control target obtained by decomposing the optimized execution vector and provided to the circulating fan frequency converter.

[0061] Step S602: Write the optimized feed rate into the feeder speed controller, write the optimized classifying speed into the classifier frequency converter, and write the optimized fan frequency into the circulating fan frequency converter. Specifically, before writing to the actuator, the controller checks again whether the three components of the optimized execution vector are within the corresponding allowed execution range, and checks the equipment operation permission, inverter readiness status and existing interlocking conditions.

[0062] When the feeder speed controller receives the setpoint for the feed rate into the mill, it directly writes the optimized feed rate. If the feeder uses rotational speed or frequency as the underlying execution value, the optimized feed rate is converted into the corresponding underlying command based on the feeder's existing calibration relationship. This conversion does not change the mass flow rate target represented by the optimized feed rate. The optimized classifying rotational speed is written as the classifier rotational speed setpoint into the classifier frequency converter, and the optimized fan frequency is written as the circulating fan frequency setpoint into the circulating fan frequency converter.

[0063] The three instructions use the speed ramps, start / stop permissions, and protection interlocks already configured on the production line, without bypassing the equipment protection logic of the mill, classifier, air path, and dust removal system.

[0064] Step S603, after execution, collects the feed rate into the mill, the active power of the mill, the characteristic particle size of the feed, the characteristic particle size of the finished product, and the differential pressure of the mill cavity, and uses it as the data to collect the digital twin running frame of the next control cycle.

[0065] Specifically, after execution, the actual process values ​​measured by each acquisition device are read, rather than the set values ​​in the optimized execution vector. The next control cycle reconstructs the digital twin operation frame with a new unified timestamp, so that the actual response of the physical production line re-enters the material lag time, grinding state, and twin response calculation process, forming a continuous closed-loop update.

[0066] like Figure 3 As shown, in practical applications, corresponding data acquisition devices are connected to the feeding, mill drive, feed detection, finished product detection, and mill cavity pressure measurement points in the barite powder production line. The controller reads the current execution commands of the feeder, classifier, and circulating fan. After the equipment maintenance is completed and stable production is achieved, the benchmark differential pressure and benchmark feed rate are first calibrated under the condition that the finished product characteristic particle size meets the requirements of the current production order. Then, the benchmark lag time is obtained through a feed step test within the allowable load range, and the differential pressure delay coefficient is fitted using data from multiple stable operating conditions. During normal operation, each acquired value and execution command is written into the digital twin operation frame according to the control cycle. The historical cache obtains the historical mill active power, feed rate, and feed characteristic particle size corresponding to the current finished product based on the material lag time, forming the specific grinding energy and effective grinding constant. The digital twin only performs virtual perturbation on the feeder, classifier, and circulating fan commands within the existing allowable execution range of the equipment to obtain the predicted results of energy consumption, particle size, and mill cavity differential pressure for candidate execution states. From the candidate results that meet both the acceptable particle size range and the safe differential pressure range, the one with the lower predicted specific grinding energy is selected, and the corresponding optimized feed rate, optimized classifying speed, and optimized blower frequency are written into the original actuator. The actual data collected after execution enters the next control cycle, thus forming a continuously updated closed-loop control.

[0067] This application provides a method for synergistic optimization of energy consumption and quality in a barite powder production line based on digital twins, comprising the following steps: Step S1: Collect the feed rate, mill power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure; read the execution instructions of the feeder, classifier and circulating fan; form an execution vector in a fixed order; and construct a digital twin running frame according to the same timestamp. Step S2: Calculate the material delay time based on the mill chamber differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and pressure difference delay coefficient. Step S3: Based on the mill's active power, feed rate, and feed characteristic particle size at historical moments of material lag time, align them with the finished product characteristic particle size to determine the specific grinding energy and effective grinding constant. Step S4: Apply virtual perturbation to the execution vector, determine the sensitivity matrix based on the effective grinding constant, and obtain the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. Step S5: Under the constraints of the qualified particle size range and the differential pressure safety range, determine the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy; Step S6: The optimized execution vector is split into optimized feed rate, optimized classifying speed and optimized fan frequency, and the feeder, classifier and circulating fan are controlled respectively.

[0068] This application also provides a computer-readable storage medium storing a computer program thereon; when the computer program is executed by a processor, it implements the energy consumption and quality co-optimization method for a barite powder production line based on digital twins as described in any embodiment of this application.

[0069] A portion of this application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0070] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A system for collaborative optimization of energy consumption and quality in a barite powder production line based on digital twins, characterized in that, The system includes a controller and acquisition and execution devices respectively connected to the controller. The controller is used to run a digital twin model and includes the following functional modules: The data frame construction module collects the feed rate, mill active power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure, reads the execution instructions of the feeder, classifier and circulating fan, assembles the execution vector in a fixed order, and constructs a digital twin operation frame according to the same timestamp; The material delay calculation module calculates the material delay based on the mill cavity differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and differential pressure delay coefficient. The grinding constant calculation module reads the mill's active power, feed rate, and feed characteristic particle size at historical moments based on material lag time, aligns them with the finished product characteristic particle size, and determines the specific grinding energy and effective grinding constant. The twin response prediction module applies a virtual perturbation to the execution vector, determines the sensitivity matrix based on the effective grinding constant, and obtains the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. The execution vector optimization module determines the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy, under the constraints of the qualified particle size range and the differential pressure safe range. The optimized control execution module breaks down the optimized execution vector into optimized feed rate, optimized classifying speed, and optimized fan frequency, which control the feeder, classifier, and circulating fan respectively.

2. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins as described in claim 1, characterized in that, The data frame building module is configured to execute: The data acquisition device includes a quantitative feeding metering device, a mill power measurement device, a feed particle size detection device, a finished product online particle size detection device, and a mill cavity inlet and outlet pressure measurement device, which respectively obtain the mill feed rate, mill active power, feed characteristic particle size, finished product characteristic particle size, and mill cavity differential pressure, wherein the feed characteristic particle size and the finished product characteristic particle size are both d90 particle size. During the current control cycle, the controller reads the execution instructions from the feeder, classifier, and circulating fan from the internal instruction register, assembles them into an execution vector in a fixed order, and writes the collected data and execution vector into the digital twin operation frame with the same timestamp.

3. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins as described in claim 1, characterized in that, The material lag time calculation module is configured to execute: Under the condition that the equipment has been overhauled and the characteristic particle size of the finished product is within the qualified particle size range, the reference differential pressure and the reference feed rate are determined simultaneously; the time from the change in feed to the response of the characteristic particle size of the finished product is recorded through a feed step test, and the reference lag time is determined. Differential pressure and response time of the grinding cavity under different working conditions were collected within the allowable load range and the differential pressure time delay coefficient was obtained by fitting. The differential pressure load ratio is determined based on the differential pressure in the grinding chamber and the reference differential pressure. The reference lag time is corrected using the differential pressure delay coefficient and the differential pressure load ratio. The material lag time is then obtained by correcting the ratio of the reference feed rate to the feed rate into the mill.

4. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins as described in claim 1, characterized in that, The grinding constant calculation module is configured to perform: The digital twin historical cache interpolates historical data based on material lag time, reads the mill active power, feed rate and feed characteristic particle size corresponding to the historical moment when the current moment is subtracted from the material lag time; and determines the specific mill energy by the ratio of mill active power to feed rate at the historical moment. The effective grinding constant is determined by the specific grinding energy, the characteristic particle size of the finished product at the current moment, and the characteristic particle size of the feed at a historical moment. The effective grinding constant is the ratio of the specific grinding energy to the difference between the reciprocal of the characteristic particle size of the finished product and the reciprocal of the characteristic particle size of the feed.

5. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins according to claim 1, characterized in that, The determination of the sensitivity matrix and local prediction relationships in the twin response prediction module includes: The digital twin model centers on the execution vector of the current control cycle and applies virtual disturbances, limited to their respective allowable execution ranges, to the execution commands of the feeder, classifier, and circulating fan. For each set of virtual disturbances, the local changes in specific grinding energy, stage reverse diameter correction, and grinding cavity differential pressure are calculated based on the effective grinding constant, and the sensitivity matrix is ​​determined by local linear fitting. The predicted specific wear energy, graded reverse diameter correction amount, and predicted grinding cavity differential pressure corresponding to the candidate execution vector are determined based on the sensitivity matrix.

6. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins according to claim 1, characterized in that, The determination of the predicted finished product particle size in the twin response prediction module includes: adding the reciprocal of the characteristic particle size of the feed at the current moment, the ratio of the predicted specific grinding energy to the effective grinding constant, and the classification reverse diameter correction amount, and taking the reciprocal to determine the predicted finished product particle size.

7. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins according to claim 1, characterized in that, The execution vector optimization module is configured to execute: The acceptable particle size range is determined by the allowable range of finished product characteristic particle size specified in the current production order, while the differential pressure safety range is determined by the mill manufacturer and process debugging. Generate candidate execution vectors within the allowable execution range of the feeder, classifier, and circulating fan that are currently available to the controller and match the current production plan; Candidate execution vectors are selected based on the predicted finished product particle size being within the acceptable range and the predicted grinding cavity differential pressure being within the safe range. From the selected candidate execution vectors, the optimal execution vector is determined based on the predicted minimum specific grinding energy.

8. The energy consumption and quality synergistic optimization system for a barite powder production line based on digital twins according to claim 1, characterized in that, The optimization control execution module is configured to execute: The execution device includes a feeder speed regulator, a classifier frequency converter, and a circulating fan frequency converter; the controller splits and optimizes the execution vector according to the fixed order of the execution instructions of the feeder, classifier, and circulating fan in the execution vector to obtain the optimized feed rate, optimized classifier speed, and optimized fan frequency; The optimized feed rate, optimized classifying speed, and optimized fan frequency are respectively written into the feeder speed controller, classifier frequency converter, and circulating fan frequency converter; The feed rate, mill power, feed characteristic particle size, finished product characteristic particle size, and mill cavity differential pressure after execution are collected as digital twin operation frame data for the next control cycle.

9. A method for synergistic optimization of energy consumption and quality in a barite powder production line based on digital twins, characterized in that, Implementing the energy consumption and quality co-optimization system for a barite powder production line based on digital twins as described in any one of claims 1 to 8 includes the following steps: Step S1: Collect the feed rate, mill power, feed characteristic particle size, finished product characteristic particle size and mill cavity differential pressure; read the execution instructions of the feeder, classifier and circulating fan; form an execution vector in a fixed order; and construct a digital twin running frame according to the same timestamp. Step S2: Calculate the material delay time based on the mill chamber differential pressure, mill feed rate, reference differential pressure, reference feed rate, reference delay time, and pressure difference delay coefficient. Step S3: Based on the mill's active power, feed rate, and feed characteristic particle size at historical moments of material lag time, align them with the finished product characteristic particle size to determine the specific grinding energy and effective grinding constant. Step S4: Apply virtual perturbation to the execution vector, determine the sensitivity matrix based on the effective grinding constant, and obtain the predicted specific grinding energy, stage reverse diameter correction, predicted grinding cavity differential pressure, and predicted finished particle size of the candidate execution vector. Step S5: Under the constraints of the qualified particle size range and the differential pressure safety range, determine the optimized execution vector from the allowable execution range with the minimum predicted specific wear energy; Step S6: The optimized execution vector is split into optimized feed rate, optimized classifying speed and optimized fan frequency, and the feeder, classifier and circulating fan are controlled respectively.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the energy consumption and quality co-optimization method for a barite powder production line based on digital twins as described in claim 9.

Images