A data-driven multi-rate grinding loop control method and system

Abstract

This invention discloses a data-driven multi-rate grinding loop control method and system. The method includes the following steps: collecting historical data of the grinding loop control system at different time scales; estimating the future trajectory of the control system based on the characteristics of the historical data; predicting future data using historical data known at a set time as a benchmark, with the benchmark set time rolled forward in the time domain to construct a time-varying Hankel matrix of historical output, historical input, future output, and future input; predicting future output data based on the characteristics of the historical data; introducing a correction term for controller calibration, applying the corrected frequency of the vibratory feeder and the opening degree of the water supply valve to the grinding loop control system to track the grinding loop feed rate, water supply rate, and predicted values ​​against the set values. This invention's control method adjusts the control strategy in real time through a data-driven approach, improving control accuracy to adapt to the time-varying and multi-rate characteristics of the grinding loop.

Description

Technical Field

[0001] This invention relates to the field of mineral processing and process control technology, and in particular to a data-driven multi-rate grinding loop control method and system. Background Technology

[0002] In actual ore grinding processes, the grinding loop is a crucial link in achieving ore crushing and classification. Traditional grinding loop control systems typically employ a single-rate control method, meaning that all control and output variables have the same sampling and control period. However, in actual production, due to factors such as equipment aging, changes in the working environment, and variations in operating conditions, the parameters of the grinding loop model exhibit time-varying characteristics. This time-varying characteristic makes the traditional single-rate control method difficult to adapt to actual production needs.

[0003] Furthermore, to prevent equipment damage and ensure process stability, control operations must be limited to safe operating ranges. This includes setting strict upper and lower limits for parameters such as ore feed rate, valve opening, and mill load. However, in the context of the Industrial Internet, to meet higher standards of transient performance, distributed control systems typically set shorter control cycles for faster response and adjustment of system status. However, the detection of ore feed rate and water flow in the basic loop is limited by hardware, resulting in a relatively long output sampling period. This multi-rate characteristic makes the system's dynamic behavior more complex. Continuing to use traditional single-rate control methods not only fails to meet control requirements but may also affect the stable operation of the grinding loop, thus adversely impacting the economic benefits of the concentrator.

[0004] Therefore, there is an urgent need for a grinding loop control system that can effectively handle time-varying parameters and multi-rate characteristics in order to improve control accuracy and system stability, thereby enhancing the economic benefits of the concentrator. Summary of the Invention

[0005] The problem to be solved by the present invention is to provide a data-driven multi-rate grinding loop control method and system, which adjusts the control strategy in real time through data-driven means to improve control accuracy and adapt to the time-varying and multi-rate characteristics of the grinding loop.

[0006] This invention adopts the following technical solution: a data-driven multi-rate grinding loop control method, comprising the following steps:

[0007] Step 1: Data Acquisition and Future Trajectory Estimation: Based on the data acquisition device installed in the concentrator, historical data of the grinding loop control system at different time scales are collected, including: feed rate, water feed rate, frequency of electric vibratory feeder and opening of water supply valve. The future trajectory of the control system is estimated based on the characteristics of the historical data.

[0008] Step 2, Rolling Time Domain Prediction: Using historical data known at a set time as a benchmark, predict future data. The set time as the benchmark is rolled forward in the time domain to construct a time-varying Hankel matrix of historical output, historical input, future output, and future input. Predict future output data based on the characteristics of historical data.

[0009] Step 3, Data-driven control: Based on the future output data predicted in Step 2, a correction term is introduced to correct the controller. The corrected frequency of the vibratory feeder and the opening degree of the water supply valve are applied to the grinding circuit control system to track the grinding circuit feed rate, water supply rate and predicted value against the set value.

[0010] Preferably, step 1, estimating the future trajectory of the control system based on historical data characteristics, includes the following sub-steps:

[0011] Step 1.1: Collect asynchronous historical output and input data, stack them according to the framework cycle, and form reconstructed data with the framework cycle as the new cycle;

[0012] Step 1.2: Solve for historical trajectory patterns, and define the corresponding multi-rate Hankel matrix for historical output and input data based on the reconstructed data to characterize the characteristics of historical data;

[0013] Specifically, methods for solving historical trajectory patterns include:

[0014] Define the pattern of the historical output trajectory, denoted as y. p (m), using the least squares method to solve for the historical output trajectory pattern.

[0015] Define the pattern of the historical input trajectory, denoted as u. p (m), using the least squares method to solve for the historical input trajectory pattern.

[0016] Step 1.3: Define the multi-rate Hankel matrix corresponding to the future output and input data, and estimate the future trajectory based on the characteristics of historical data.

[0017] Specifically, the method for estimating the future trajectory is as follows:

[0018] Define the future output matrix y f (m), including time from (m+1)T0 to (m+N) y )T0+(b-1)T a The data on ore feed and water supply at any given time are arranged in a column vector, N. y This indicates the predicted time domain of the output, with the subscript f indicating future data;

[0019] Define the future output Hankel matrix H[Y] f], including from NT0+T a to (N+N) y -1+j)T0+(b-1)T a The data on ore feed rate and water feed rate at any given time, according to N y Arrange the data in Hankel format using rows b and columns jb; where the first column is NT0+T. a Time to arrive (N+N) y Data on ore feed rate and water feed rate at time T0;

[0020] Estimate the future output matrix y f (m).

[0021] Define the future input matrix u f (m), including time from mT0 to (m+N) u -1)T0+(a-1)T b The data of the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time are arranged in a column vector, N. u This indicates the prediction time domain of the input;

[0022] Define the Hankel matrix H[U] of the future input. f ], including from NT0+T b to (N+N) u -1+j)T0+(a-1)T b The data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time, according to N u Arrange the data in a row and column ja using Hankel format; where the first column is NT0+T. b Time to arrive (N+N) u Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at time T0;

[0023] Estimate the future input matrix u f (m).

[0024] Preferably, step 2, predicting future output data based on historical data characteristics, includes the following sub-steps:

[0025] Step 2.1: Define the time-varying Hankel matrix for historical output. Includes time from (mN-L+1)T0 to (m-L+j-1)T0+(2b-2)T a Real-time data on ore and water supply rates;

[0026] The first column represents the time from (mN-L+1)T0 to (m-1)T0+(b-1)T. a The data on ore feed rate and water feed rate at any given time, where L is the predicted distance;

[0027] Solving for historical trajectory patterns under time-varying Hankel matrices

[0028] Step 2.2: Define the time-varying Hankel matrix for future output. Includes time from (m-L+1)T0 to (m-L+N) y -1+j)T+(2b-2)T a Real-time data on ore and water supply rates;

[0029] The first column represents the time from (m-L+1)T0 to (m-L+N). y )T0+(b-1)T a Real-time data on ore and water supply.

[0030] Estimated future output

[0031] Step 2.3: Define the time-varying Hankel matrix of the historical input. Includes time from (mNL)T0 to ((m-L+j-2)T0+(2a-2)T b The data includes the frequency of the vibratory feeder and the opening degree of the water supply valve at time (mL-1)T0; the first column represents the data from time (mL-1)T0 to (mL-1)T0+(a-1)T0. b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time;

[0032] Solving for historical trajectory patterns under time-varying Hankel matrices

[0033] Step 2.4: Define the time-varying Hankel matrix for future inputs. Includes time from (mL)T0 to (m-L+N) y -2+j)T0+(2a-2)T b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time;

[0034] The first column represents the time from (mL)T0 to (m-L+N) u )T0+(a-1)T b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time;

[0035] Estimate future inputs

[0036] Preferably, in step 3, the data-driven control includes the following sub-steps:

[0037] Step 3.1: Introduce correction terms to perform controller calibration and predict input values; establish prediction equations for output values ​​and estimation equations for input values ​​to perform data-driven prediction of the control system.

[0038] Step 3.2: Define the performance index function, minimize the performance index function, solve for the controller, and obtain the optimal correction term;

[0039] Step 3.3: Obtain the control law and apply it to the control system to track the feed rate, water flow rate and predicted value of the grinding loop against the set value.

[0040] The present invention also provides: a data-driven multi-rate grinding loop control system for implementing any of the above control methods, comprising: a multi-rate data-driven estimation module, a rolling time-domain prediction module, and a controller correction module;

[0041] The multi-rate data-driven estimation module, based on the data acquisition device installed in the concentrator, directly collects historical data of the grinding loop control system at different time scales, including: feed rate, water feed rate, frequency of electric vibratory feeder and opening degree of water supply valve, and processes the data to estimate the future trajectory of the system based on the characteristics of historical data.

[0042] The rolling time-domain prediction module is used to predict future data using the collected data and the data known at a set time as a benchmark. The set time as the benchmark is rolled forward in the time domain to enhance the correlation between data. It constructs a time-varying Hankel matrix of historical output, historical input, future output, and future input, and predicts future output data based on the characteristics of historical data.

[0043] The controller correction module, based on the future output data predicted by the rolling time domain prediction module, introduces a correction term to correct the controller. The corrected frequency of the electric vibratory feeder and the opening degree of the water supply valve are applied to the system to track the feed rate, water supply and predicted value of the grinding circuit against the set value.

[0044] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0045] 1. The multi-rate grinding loop control system of the present invention adjusts the control strategy in real time through data-driven means. It uses known data at a certain moment as a benchmark to predict future data. At the same time, the benchmark moment rolls forward with the time domain, which enhances the correlation between data and effectively improves the prediction accuracy to adapt to the time-varying characteristics and multi-rate characteristics of the grinding loop.

[0046] 2. The multi-rate grinding loop control method of the present invention, by introducing a correction term, ensures that the corrected frequency of the vibratory feeder and the opening degree of the water supply valve, when applied to the system, can effectively track the set value of the feed rate, water supply and their predicted value. Attached Figure Description

[0047] Figure 1This is a flowchart of the data-driven multi-rate grinding loop control method of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the application will be further described in detail below with reference to the accompanying drawings. The described embodiments are only a part of the embodiments involved in this invention. All non-innovative embodiments based on these embodiments by other researchers in the art are within the protection scope of this invention. Furthermore, the step numbers in the embodiments of this invention are only set for ease of explanation and do not limit the order of the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0049] In one embodiment of the present invention, a data-driven multi-rate grinding loop control system, such as... Figure 1 As shown, it includes a multi-rate data-driven estimation module, a rolling time-domain prediction module, and a controller correction module.

[0050] The multi-rate data-driven estimation module is used to directly collect and process data on ore feed rate, water feed rate, electric vibratory feeder frequency and water supply valve opening at different time scales, and estimate the future trajectory of the system based on the characteristics of historical data.

[0051] The rolling time-domain prediction module is responsible for using the collected data and using the known data at a certain moment as a benchmark to predict future data. At the same time, the benchmark moment rolls forward with the time domain to enhance the correlation between data and improve prediction accuracy.

[0052] The controller correction module is responsible for introducing correction terms based on the prediction module mentioned above, so that the corrected frequency of the vibratory feeder and the opening degree of the water supply valve can ensure that the feed rate, water supply and their predicted values ​​track the set values ​​after being applied to the system.

[0053] Specifically, in this embodiment, the control method for the multi-rate grinding circuit includes the following steps:

[0054] Step 1: Data Acquisition and Future Trajectory Estimation

[0055] Data on ore feed rate, water feed rate, electric vibratory feeder frequency, and water supply valve opening are collected using data acquisition devices installed in the ore dressing plant.

[0056] After data collection, there is no need to identify model parameters online. Instead, data from different time scales are directly stacked according to the framework cycle to form reconstructed data with the framework cycle as the new cycle.

[0057] Constructing a multi-rate Hankel matrix to characterize historical data characteristics and estimating the system's future trajectory based on these characteristics involves the following two steps.

[0058] 1.1 Solving for patterns in historical trajectories

[0059] The pattern of historical output trajectories is defined as follows:

[0060] y p (m)=H[Y p g y0 (m)

[0061] Where m represents the number of steps in the frame cycle, y p (m) represents the historical output trajectory matrix, Y p H[Y] represents the output value within a historical fixed-step period. p ] represents the Hankel matrix of output values ​​within a historical fixed-step period, g y0 (m) represents y p (m) and H[Y p The pattern between the output values.

[0062] The least squares method is used to solve for the historical output trajectory pattern:

[0063]

[0064] Where T represents transpose.

[0065] The pattern of historical input trajectories is defined as follows:

[0066] u p (m)=H[U p g u0 (m)

[0067] Among them, u p (m) represents the historical input trajectory matrix, U p H[U] represents the input value within a historical fixed-step period. p ] represents the Hankel matrix of input values ​​within a historical fixed-step period, g u0 (m) represents u p (m) and H[U p The pattern between the input values.

[0068] The least squares method is used to solve for the patterns in historical input trajectories.

[0069]

[0070] 1.2 Estimating future input and output matrices

[0071] Define the future output matrix y f (m), including time from (m+1)T0 to (m+N) y )T0+(b-1)Ta The data on ore feed and water supply at any given time are arranged in a column vector, N. y This indicates the predicted time domain of the output, where the subscript f represents future data, b represents the ratio of the sampling period to the base period, and T0, T... a These represent the control cycles of the frame cycle, the electric vibratory feeder, and the water supply valve, respectively.

[0072] Define the future output Hankel matrix H[Y] f ], including from NT0+T a to (N+N) y -1+j)T0+(b-1)T a The ore feed and water supply data at any given time are arranged in a Hankel format with Nyb rows and jb columns, Y f This represents the output value in the future prediction time domain, where N represents the length of the sequence.

[0073] The first column is NT0+T a Time to arrive (N+N) y Data on ore feed rate and water feed rate at time T0.

[0074] Estimate the future output matrix:

[0075]

[0076] Define the future input matrix u f (m), including time from mT0 to (m+N) u -1)T0+(a-1)T b The data of the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time are arranged in a column vector, N. u This represents the prediction time domain of the input, 'a' represents the multiple relationship between the control period and the base period, and T... b This indicates the sampling period for ore feed and water supply.

[0077] Define the Hankel matrix H[U] of the future input. f ], including from NT0+T b to (N+N) u -1+j)T0+(a-1)T b The data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time, according to N u Arrange the rows and columns in a Hankel pattern, U f This represents the input value in the future control time domain;

[0078] The first column is NT0+T b Time to arrive (N+N) u Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at time T0.

[0079] Estimate the future input matrix:

[0080]

[0081] Step 2, Rolling Time Domain Prediction

[0082] Using known data at a certain moment as a baseline to predict future data, and with the baseline moment rolling forward in the time domain, a time-varying Hankel matrix is ​​constructed consisting of historical output, historical input, future output, and future input. Based on the characteristics of historical data, future output data is predicted, enhancing the correlation between data and improving prediction accuracy.

[0083] Define the time-varying Hankel matrix of historical output Includes time from (mN-L+1)T0 to (m-L+j-1)T0+(2b-2)T a Real-time data on ore and water supply rates. This represents the output value within a fixed step size as the time axis scrolls along it.

[0084] The first column contains the ore feed and water supply data from time (mN-L+1)T0 to time 1, representing the predicted distance.

[0085] Solving for historical trajectory patterns under time-varying Hankel matrices

[0086]

[0087] Define the time-varying Hankel matrix for future output Includes time from (m-L+1)T0 to (m-L+N) y -1+j)T+(2b-2)T a Real-time data on ore and water supply rates. This represents the output value in the time domain for future predictions that scroll along the time axis;

[0088] The first column contains the ore feed and water feed data from time (m-L+1)T0 to time (m-L+1).

[0089] Estimated future output

[0090]

[0091] Define the time-varying Hankel matrix of the historical input. Includes time from (mNL)T0 to ((m-L+j-2)T0+(2a-2)T b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time;

[0092] The first column represents the time from (mNL)T0 to (mL-1)T0+(a-1)T. b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time.

[0093] Solving for historical trajectory patterns under time-varying Hankel matrices

[0094]

[0095] Define the time-varying Hankel matrix for future inputs Includes time from (mL)T0 to (m-L+N) y -2+j)T0+(2a-2)T b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time;

[0096] The first column represents the time from (mL)T0 to (m-L+N) u )T0+(a-1)T b Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time.

[0097] Estimate future inputs

[0098]

[0099] Step 3: Data-driven control

[0100] Based on the above prediction module, a correction term is introduced so that the corrected frequency of the vibratory feeder and the opening degree of the water supply valve can ensure that the feed rate, water supply and their predicted values ​​track the set values ​​after being applied to the system.

[0101] Specifically, it consists of the following two steps:

[0102] 3.1 Conducting a data-driven prediction process:

[0103]

[0104] in, It is by and The historical data matrix is ​​composed of yes and The matrix formed by arranging columns into vectors, z p (m), z f (m) represents historical data and predicted data, respectively.

[0105] Added correction items By adding a correction term, the predicted input value, when applied to the system, can track the output value. Therefore, the prediction process of the output and input values ​​after introducing the correction term can be described as follows:

[0106]

[0107] Predict input values

[0108]

[0109] in, For N u a×N u The identity matrix of a, O u For N u a×N y The zero matrix of b, This indicates a correction term.

[0110] 3.2 Solving the controller

[0111] Define the performance index function:

[0112]

[0113] stz f =z r

[0114] Among them, z r For data that has been actually collected, the equality constraint z f =z r The goal is to make the predicted value equal to the actual collected value, where w represents the setpoint for the feed rate and water supply data. * (m) = [w(m)], where [w(m)] represents a matrix composed of set values. and This is the weighting factor.

[0115] In this embodiment,

[0116]

[0117] Let the performance index function Differentiate U(m), substitute the prediction equations for feed rate and water rate and the estimation equations for input values ​​into the performance index function to minimize the performance index function. By taking the derivative and setting it to 0, the optimal correction term can be obtained.

[0118] The control law can be obtained by substituting the optimal correction term into the estimation equation of the input value:

[0119]

[0120] Will The first element in the algorithm acts on the system. When the time domain rolls to time m+1, the above process is repeated to achieve rolling time domain and online optimization.

[0121] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A data-driven multi-rate grinding loop control method, characterized in that, Includes the following steps: Step 1: Data Acquisition and Future Trajectory Estimation: Based on the data acquisition device installed in the concentrator, historical data of the grinding loop control system at different time scales are collected, including: feed rate, water feed rate, frequency of electric vibratory feeder and opening of water supply valve. The future trajectory of the control system is estimated based on the characteristics of the historical data. Step 2, Rolling Time Domain Prediction: Using historical data known at a set time as a benchmark, predict future data. The set time as the benchmark is rolled forward in the time domain to construct a time-varying Hankel matrix of historical output, historical input, future output, and future input. Predict future output data based on the characteristics of historical data. Step 3, Data-driven control: Based on the future output data predicted in Step 2, a correction term is introduced to correct the controller. The corrected frequency of the vibratory feeder and the opening degree of the water supply valve are applied to the grinding circuit control system to track the grinding circuit feed rate, water supply rate and predicted value against the set value.

2. The data-driven multi-rate grinding loop control method according to claim 1, characterized in that, Step 1, estimating the future trajectory of the control system based on historical data characteristics, includes the following sub-steps: Step 1.1: Collect asynchronous historical output and input data, stack them according to the framework cycle, and form reconstructed data with the framework cycle as the new cycle; Step 1.2: Solve for historical trajectory patterns, and define the corresponding multi-rate Hankel matrix for historical output and input data based on the reconstructed data to characterize the characteristics of historical data; Step 1.3: Define the multi-rate Hankel matrix corresponding to the future output and input data, and estimate the future trajectory based on the characteristics of historical data.

3. The data-driven multi-rate grinding loop control method according to claim 2, characterized in that, In step 1.2, the method for solving the historical trajectory pattern is as follows: The pattern of historical output trajectories is defined as follows: ； in, Indicates the number of steps within the framework cycle. Represents the historical output trajectory matrix. This represents the output value within a historical fixed-step period. The Hankel matrix represents the output values ​​within a historical fixed-step period. express and The pattern between the output values; The least squares method is used to solve the historical output trajectory pattern. : ； Among them, superscript Indicates transpose; The pattern of historical input trajectories is defined as follows: ； in, Represents the historical input trajectory matrix. This represents the input value within a historical fixed-step period. The Hankel matrix represents the input values ​​within a historical fixed-step period. express and The pattern between the input values; The least squares method is used to solve the historical input trajectory pattern. : 。 4. The data-driven multi-rate grinding loop control method according to claim 3, characterized in that, In step 1.3, the method for estimating the future trajectory is as follows: Define the future output matrix , including from Time's up The data on ore feed and water supply at any given time are arranged in a column vector. Indicates the predicted time domain of the output, subscript Represents future data. This indicates the ratio of the sampling period to the base period. , These represent the control cycles of the frame cycle, the vibratory feeder, and the water supply valve, respectively. Define the future output Hankel matrix , including from arrive The data on ore feed rate and water feed rate at any given time, according to OK Arrange the columns in Hankel form. This represents the output value in the future prediction time domain. Indicates the length of the sequence; The first column is Time's up Real-time data on ore and water supply rates; Estimate the future output matrix: ； Define the future input matrix , including from Time's up The data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time are arranged in a column vector. This represents the prediction time domain of the input. This indicates the multiple relationship between the control period and the base period. Indicates the sampling period for ore feed and water supply; Define the Hankel matrix for future inputs. , including from arrive The data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time, according to OK Arrange the columns in Hankel form. This represents the input value in the future control time domain; The first column is Time's up Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time; Estimate the future input matrix: 。 5. The data-driven multi-rate grinding loop control method according to claim 4, characterized in that, Step 2 involves predicting future output data based on historical data characteristics, including the following sub-steps: Step 2.1: Define the time-varying Hankel matrix for historical output. , including from Time's up Real-time data on ore and water supply rates. This represents the output value within a fixed step size as the time axis scrolls along it. The first column is Time's up Real-time data on ore and water supply rates. To predict distance; Solving for historical trajectory patterns under time-varying Hankel matrices : ； Step 2.2: Define the time-varying Hankel matrix for future output. , including from Time's up Real-time data on ore and water supply rates. This represents the output value in the time domain for future predictions that scroll along the time axis; The first column is Time's up Real-time data on ore and water supply rates; Estimated future output : ； Step 2.3: Define the time-varying Hankel matrix of the historical input. , including from Time's up Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time; The first column is Time's up Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time; Solving for historical trajectory patterns under time-varying Hankel matrices : ； Step 2.4: Define the time-varying Hankel matrix for future inputs. , including from Time's up Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time; The first column is Time's up Data on the frequency of the vibratory feeder and the opening degree of the water supply valve at any given time; Estimate future inputs : 。 6. The data-driven multi-rate grinding loop control method according to claim 5, characterized in that, Step 3, the data-driven control, includes the following sub-steps: Step 3.1: Introduce correction terms to perform controller calibration and predict input values; establish prediction equations for output values ​​and estimation equations for input values ​​to perform data-driven prediction of the control system. Step 3.2: Define the performance index function, minimize the performance index function, solve for the controller, and obtain the optimal correction term; Step 3.3: Obtain the control law and apply it to the control system to track the feed rate, water flow rate and predicted values ​​of the grinding loop against the setpoint.

7. The data-driven multi-rate grinding loop control method according to claim 6, characterized in that, In step 3.1, data-driven prediction is performed, using the following method: The data-driven prediction process can be described as follows: ； ； ； in, It is by and The historical data matrix is ​​composed of yes and A matrix formed by arranging columns into vectors. , These represent historical data and forecast data, respectively. Added correction items , The output value is tracked after the predicted input value is applied to the control system. The prediction process for output and input values ​​after introducing correction terms, and the prediction equations for feed rate and water supply values ​​are described as follows: ； Predict input values The estimation equation for the input values ​​is described as follows: ； in, , for The identity matrix, for The zero matrix, This indicates a correction term.

8. The data-driven multi-rate grinding loop control method according to claim 7, characterized in that, In step 3.2, the controller is solved using the following method: Define performance index function : ； in, Equality constraints are based on actual collected data. Used to make the predicted value equal to the actual collected value. The set values ​​represent the ore feed rate and water supply rate data. , This represents a matrix composed of set values. and As a weighting factor; Let the performance index function right Differentiate the equations for predicting feed and water rates and the equations for estimating input values, then substitute these equations into the performance index function to minimize the performance index function. Take the derivative, set the result to 0, and find the optimal correction term. ; Substituting the optimal correction term into the estimation equation for the input value, the control law is obtained: ； Will The first element in the table acts on the system when the time domain rolls to... At time 1, repeat steps 2 to 3 to perform rolling time-domain online optimization.

9. A data-driven multi-rate grinding loop control system for implementing the control method according to any one of claims 1 to 8, characterized in that, include: Multi-rate data-driven estimation module, rolling time-domain prediction module, and controller correction module; The multi-rate data-driven estimation module, based on the data acquisition device installed in the concentrator, directly collects historical data of the grinding loop control system at different time scales, including: data on feed rate, water feed rate, frequency of electric vibratory feeder and opening degree of water supply valve, and processes the data to estimate the future trajectory of the system based on the characteristics of historical data. The rolling time-domain prediction module is used to predict future data using the collected data and data known at a set time as a benchmark. The set time as the benchmark is rolled forward in the time domain to enhance the correlation between data and construct a time-varying Hankel matrix of historical output, historical input, future output, and future input. It then predicts future output data based on the characteristics of historical data. The controller correction module, based on the future output data predicted by the rolling time domain prediction module, introduces a correction term to correct the controller. The corrected frequency of the electric vibratory feeder and the opening degree of the water supply valve are applied to the system to track the feed rate, water supply and predicted value of the grinding circuit against the set value.

Images