A grinding and classification process control particle size optimization system and method

Abstract

The present invention relates to the field of intelligent metallurgical technology, and discloses a grinding and grading process control particle size optimization system and method. The optimization system includes: an input module, a data acquisition module, a control module and an execution module, wherein the input module is used to input a target overflow particle size and transmit the target overflow particle size to the control module; the data acquisition module is used to collect control variables and overflow particle size in the current grinding and grading process, and transmit the control variables and overflow particle size to the control module; the control module is used to input the control variables and overflow particle size into a grading model, obtain grinding adjustment data for the target overflow particle size, and transmit the grinding adjustment data to the execution module; the execution module is used to adjust the control variables in the current grinding and grading process based on the grinding adjustment data, thereby solving the problems of low accuracy and poor stability of the existing grinding and grading control system.

Description

Technical Field

[0001] The present invention relates to the field of intelligent metallurgical technology, and in particular to a grinding classification particle size optimization system and method. Background Art

[0002] Grinding and classification is an important step before flotation in the mineral processing process. The stability of its particle size directly affects the efficiency of flotation and the quality of the final product. However, the existing grinding and classification control system mainly faces the following key problems:

[0003] Nonlinear and large hysteresis characteristics: Traditional linear control systems are unable to effectively cope with this complex dynamic behavior. These characteristics cause the corresponding systems to have problems such as low control accuracy and slow response speed when facing external disturbances and control instructions, thus affecting the quality stability of the product.

[0004] Multivariable coupling: In the grinding and classification process, there is a complex coupling relationship between the control variables (such as feed concentration, pressure and flow rate, etc.). This multivariable coupling effect increases the difficulty of system control. The traditional single variable control method is difficult to optimize multiple control variables at the same time, resulting in unsatisfactory overall control effect of the system.

[0005] External interference: In actual operation, the grinding and grading system is often affected by external interference such as changes in ore properties and equipment wear. These interferences will cause changes in system parameters, thereby affecting control accuracy and system stability. Existing methods often fail to make timely adjustments when dealing with these interferences, making it difficult to maintain stable operation of the system.

[0006] Limitations of data-driven algorithms: Traditional methods mostly rely on manual operations or simple algorithms. Modern control algorithms, such as data-driven algorithms based on neural networks and support vector machines, perform well in static predictions, but have obvious deficiencies in dealing with the large lag and singleness of field data in grinding and grading systems. These algorithms require a large amount of diverse data, while the data in the grinding and grading process is often relatively single and has data imbalance problems, making it difficult for the algorithm to obtain effective data with sufficient coverage. Even if some of the data is obtained, it is difficult to apply it to complex grinding and grading process control due to data imbalance.

[0007] In order to solve the above problems, a new grinding and classification control system is urgently needed to solve the above problems. Summary of the invention

[0008] The invention provides a grinding and classification process control particle size optimization system and method, so as to solve the problems of low precision and poor stability of the existing grinding and classification control system.

[0009] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0010] In a first aspect, the present invention provides a grinding and grading process control particle size optimization system, comprising: an input module, a data acquisition module, a control module and an execution module, wherein the control module stores a grading model based on a first relationship model and a second relationship model and combined with a nonlinear algorithm and a random optimization algorithm; the first relationship model is a relationship model of a control variable, an external disturbance and an overflow particle size, and the second relationship model is a relationship model including a control variable, an external disturbance and a system state;

[0011] The input module is used to input a target overflow particle size and transmit the target overflow particle size to the control module;

[0012] The data acquisition module is used to collect the control variables and overflow particle size in the current grinding and classification process, and transmit the control variables and overflow particle size to the control module;

[0013] The control module is used to input the control variables and the overflow particle size into the classification model, obtain the grinding adjustment data for the target overflow particle size, and transmit the grinding adjustment data to the execution module;

[0014] The classification model predicts the system state at the target overflow particle size based on the second relationship model, and then obtains grinding adjustment data under the influence of the system state based on the first relationship model;

[0015] The classification model obtains grinding adjustment data under the influence of system status based on the first relationship model, which is expressed as:

[0016] ;

[0017] Indicates the target overflow particle size Grinding adjustment data at all times; Indicates the target overflow particle size System status at the moment; express The control variable vector of the system at time ; express The system external disturbance vector at time ; A nonlinear function representing the relationship between overflow particle size and control variables;

[0018] The nonlinear function of the relationship between the overflow particle size and the control variable is expressed by the following formula:

[0019] ;

[0020] in, It shows the relationship between overflow particle size and control variables; represents the system state vector, represents the control variable vector, represents the external interference vector; Indicates the linear effect of overflow particle size on grinding adjustment data, represents the linear effect of the control variable on the grinding adjustment data, Represents the linear effect of external disturbance on grinding adjustment data, The nonlinear characteristics and hysteresis effects of the relationship between overflow particle size and control variables;

[0021] It is expressed by the following formula:

[0022] ;

[0023] in, , , as well as Empirical constants representing coefficients of nonlinear terms; represents the quadratic nonlinear effect of overflow particle size; represents the saturated nonlinear effect of the control variable; It indicates the nonlinear relationship between external disturbance and grinding adjustment data; represents the dynamic response lag of the system to the control variable, where represents the hysteresis coefficient, represents the integral time variable;

[0024] The execution module is used to adjust the control variables in the current grinding and classification process based on the grinding adjustment data.

[0025] Furthermore, the system state at the time of predicting the target overflow granularity based on the second relational model is expressed as:

[0026] ;

[0027] in, A nonlinear function representing the state of the system; express The control variable vector of the system at time ; express The system external disturbance vector at time ;

[0028] The nonlinear function of the system state is expressed by the following formula:

[0029] ;

[0030] in, Indicates the system status; Represents the dynamic characteristics of the system. represents the linear influence of the control variable on the system state, Represents the linear effect of external disturbance on the system state; Represents the nonlinear characteristics and hysteresis effects of the system state;

[0031] It is expressed by the following formula:

[0032] ;

[0033] in, , , as well as Empirical constants representing coefficients of nonlinear terms; It represents the nonlinear influence of the control variables on the grinding and classification effect; Represents the nonlinear effects of system states; represents the nonlinear effect of the control variable; represents the dynamic response hysteresis of the system state, where Indicates the response speed of the system. Represents the time variable, which is the starting time of the integration.

[0034] Furthermore, when the control module obtains the grinding adjustment data through the classification model, the grinding adjustment data is optimized by minimizing the objective function;

[0035] The objective function is expressed by the following formula:

[0036] ;

[0037] ; ;

[0038] Among them, among them, Indicates the target overflow particle size Grinding adjustment data at all times; Indicates the desired target overflow particle size; Indicates The change in the control variable; , as well as Represents the weight coefficient of balancing the grinding coefficient and the tracking error; Represents the weight coefficient of the penalty term or incentive term related to the grinding and classification process logic, Represents the penalty term or incentive term related to the grinding and classification process logic.

[0039] Further, the control variables include feed flow rate, feed concentration and feed pressure;

[0040] The penalty item or incentive item related to the grinding and classification process logic includes a particle size increase incentive item and a particle size decrease penalty item. The particle size increase incentive item is expressed as: ;

[0041] These include, feed flow is too high: ;

[0042] Feed concentration is too high: ;

[0043] Feed pressure is too low: ;

[0044] The granularity reduction penalty term is expressed as: ;

[0045] These include, feed flow is too low: ;

[0046] Feed concentration is too low: ;

[0047] Feed pressure is too high: ;

[0048] in, , , Respectively represent feed concentration, feed pressure, and feed flow rate, , , They respectively represent the influence of feed flow rate on the change of overflow particle size, the influence of feed concentration on the change of overflow particle size, and the influence of feed pressure on the change of overflow particle size.

[0049] Furthermore, the classification model includes the following constraints in the process of optimizing the grinding adjustment data:

[0050] ;

[0051] ;

[0052] in, Indicates the minimum value of the preset control variable; Indicates the maximum value of the preset control variable; Indicates the preset minimum change value of the control variable. Indicates the preset maximum value of the control variable change.

[0053] In a second aspect, the present invention further provides a method for optimizing particle size during grinding and classification process control, which is applied to the above-mentioned system for optimizing particle size during grinding and classification process control, and the method comprises the following steps:

[0054] Step 1: Collect the control variables and overflow particle size in the current grinding and classification process;

[0055] Step 2: Input the control variables and overflow particle size into the classification model to obtain grinding adjustment data for the target overflow particle size;

[0056] Step 3: adjusting the control variables in the current grinding and classification process based on the grinding adjustment data.

[0057] Furthermore, the classification model optimizes the grinding adjustment data, and the optimization includes the following steps:

[0058] Step 201: Initialize the parameters of the classification model;

[0059] Step 202: Calculate the gradient of the objective function of the hierarchical model under the current parameters;

[0060] Step 203: updating the first-order moment estimation and the second-order moment estimation according to the gradient, performing deviation correction on the first-order moment estimation and the second-order moment estimation, introducing an adaptive learning correction term to correct the gradient, and reacquiring parameters according to the corrected gradient;

[0061] Step 204: Update the gradient and parameters, and repeat steps 202 to 203 until the norm of the gradient meets the preset norm or reaches the preset number of iterations.

[0062] Beneficial effects:

[0063] The present invention provides a grinding and classification process control particle size optimization system and method, which can accurately control the nonlinearity and large hysteresis characteristics of the grinding and classification system according to the first relationship model combined with the nonlinear algorithm. Compared with the traditional control method, the system steady-state error is reduced, the control accuracy is improved, and the stability of the grinding particle size is ensured;

[0064] On the basis of the first relationship model, the second relationship model and the nonlinear algorithm are combined to further improve the control accuracy and ensure the stability of the frosting particle size;

[0065] At the same time, nonlinear algorithms also overcome the limitations of static prediction algorithms such as neural networks and support vector machines. They can collect data in real time and dynamically adjust control strategies, making them more adaptable to the dynamic changes and nonlinear characteristics that occur during the grinding and classification process. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 The present invention is a flowchart of a grinding and classification process control particle size optimization system in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0067] The technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by persons having ordinary skills in the field to which the present invention belongs.

[0069] The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "an" do not indicate a quantity limitation, but indicate the existence of at least one. Words such as "connected" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship also changes accordingly.

[0070] The present embodiment provides a grinding and classification process control particle size optimization system, including: an input module, a data acquisition module, a control module and an execution module, the control module stores a classification model based on a first relationship model and a second relationship model combined with a nonlinear algorithm and a random optimization algorithm; the first relationship model is a relationship model of a control variable, an external interference and an overflow particle size, and the second relationship model is a relationship model of a control variable, an external interference and a system state;

[0071] An input module, used for inputting a target overflow particle size and transmitting the target overflow particle size to a control module;

[0072] A data acquisition module is used to collect control variables and overflow particle size in the current grinding and classification process, and transmit the control variables and overflow particle size to the control module;

[0073] Controlled variables include feed flow rate, feed concentration, and feed pressure;

[0074] The data acquisition module transmits real-time data of control variables to PLC (Programmable Logic Controller) or DCS (Distributed Control System) through high-precision sensors and data acquisition cards;

[0075] Use the OPC (OLE for Process Control) protocol to transfer real-time data of control variables from PLC or DCS to the real-time database to ensure the reliability and efficiency of data transmission. The real-time database is used to store and manage all collected data and provide real-time data support for the control module.

[0076] A control module, used for inputting control variables and overflow particle size into a classification model, obtaining grinding adjustment data for a target overflow particle size, and transmitting the grinding adjustment data to an execution module;

[0077] The hierarchical model is used to predict the future system state and output and calculate the optimal control strategy. The optimal input is determined by optimizing the control variables. The grinding adjustment data of the optimal control strategy is sent to the execution module to perform real-time control.

[0078] The execution module is used to adjust the control variables in the current grinding and classification process based on the grinding adjustment data.

[0079] The execution module ensures the stable operation of the system and product quality by controlling the feed concentration, pressure and flow rate of the cyclone. The specific regulation includes the valve adjusting the feed flow rate to control the speed at which the slurry enters the cyclone; the motor controls the operating state of the cyclone to adjust the feed pressure; the pump adjusts the concentration of the slurry to ensure accurate control of the overflow particle size.

[0080] See also Figure 1 , which is a specific flow chart of the particle size optimization system for the grinding and grading process control. The data acquisition module collects the control variables and transmits them to the real-time database. The real-time database stores and manages the real-time data of the control variables. The control module obtains the real-time data of the control variables from the real-time database, obtains the grinding adjustment data in combination with the grading model, and transmits it to the execution module. The execution module controls the actuator to adjust the working state of the grading equipment according to the grinding adjustment data. After the working adjustment, the grading equipment can also feedback data to the real-time database to form a closed-loop control. In addition, an expert experience library can be set up to optimize the control strategy according to the knowledge of the expert experience library, and the control strategy of the control module can be updated as needed.

[0081] Please refer to Table 1. In this embodiment, the sample data of the following control variables are collected by the data collection module, and are evenly collected through the OPC protocol during the production cycle;

[0082] Table 1: Sample data of control variables

[0083]

[0084] See Table 2. Also enter the target overflow size via the input module. is 60.3;

[0085] Table 2: Setting target overflow particle size

[0086]

[0087] In the classification model of the data acquisition module, based on the large hysteresis and nonlinear characteristics of the grinding analysis system, the state equation of the grinding classification system is defined as:

[0088] ;

[0089] in, Indicates the system status; A nonlinear function representing the state of the system; represents the control variable vector, represents the external interference vector;

[0090] Nonlinear function of the system state The specific form is:

[0091] ;

[0092] in, Indicates the system status; represents the system state vector, Represents the dynamic characteristics of the system. represents the linear influence of the control variable on the system state, Represents the linear effect of external disturbance on the system state; Represents the nonlinear characteristics and hysteresis effects of the system state;

[0093] In order to accurately describe the nonlinear and hysteresis effects in the system, the nonlinear function Contains the following items that are highly relevant to the rating system:

[0094] ;

[0095] in, , , as well as The empirical constant representing the coefficient of the nonlinear term, The value is 0.5. The value is 0.2. The value is 0.2. The value is 0.1; It represents the nonlinear influence of the control variables on the grinding and classification effect; Represents the nonlinear effects of system states; represents the nonlinear effect of the control variable; represents the dynamic response hysteresis of the system state, where Indicates the response speed of the system. represents the time variable, which is the starting time of the integration. Represents the integration time variable.

[0096] Based on the relationship between the system output (overflow particle size), state variables and control variables, the following output equation is defined:

[0097] ;

[0098] in, A nonlinear function representing the relationship between overflow particle size and control variables;

[0099] Nonlinear function of the relationship between overflow particle size and control variables It is expressed by the following formula:

[0100] ;

[0101] in, It shows the relationship between overflow particle size and control variables; Indicates the linear effect of overflow particle size on grinding adjustment data, represents the linear effect of the control variable on the grinding adjustment data, Represents the linear effect of external disturbance on grinding adjustment data, The nonlinear characteristics and hysteresis effects of the relationship between overflow particle size and control variables;

[0102] It is expressed by the following formula:

[0103] ;

[0104] in, , , as well as The empirical constant representing the coefficient of the nonlinear term, The value is 0.6. The value is 0.2. The value is 0.1. The value is 0.1; represents the quadratic nonlinear effect of overflow particle size; represents the saturated nonlinear effect of the control variable; It indicates the nonlinear relationship between external disturbance and grinding adjustment data; represents the dynamic response lag of the system to the control variable, where represents the hysteresis coefficient.

[0105] Finally, based on the state equation and output equation defined above, a classification model is constructed to predict the system state at the target overflow particle size based on the second relationship, and then the relevant equations of the grinding adjustment data under the influence of the system state are obtained based on the first relationship;

[0106] The system state when predicting the target overflow granularity based on the second relationship is expressed as:

[0107] ;

[0108] in, Indicates the target overflow particle size The system state at the moment; based on the first relationship, the grinding adjustment data under the influence of the system state is obtained and expressed as;

[0109] ;

[0110] Indicates the target overflow particle size Grinding adjustment data at all times;

[0111] The classification model optimizes the grinding adjustment data by minimizing the objective function;

[0112] The objective function is expressed by the following formula:

[0113] ;

[0114] ; ;

[0115] in, Indicates the target overflow particle size Grinding adjustment data at all times; Indicates the desired target overflow particle size; Indicates The change in the control variable; , as well as Represents the weight coefficient of balancing the grinding coefficient and the tracking error; Represents the weight coefficient of the penalty term or incentive term related to the grinding and classification process logic, Represents the penalty term or incentive term related to the grinding and classification process logic.

[0116] The penalty items or incentive items related to the grinding and classification process logic include the incentive item for increasing particle size and the penalty item for decreasing particle size. The incentive item for increasing particle size is expressed as: ;

[0117] These include, feed flow is too high: ;

[0118] Feed concentration is too high: ;

[0119] Feed pressure is too low: ;

[0120] The penalty term for smaller granularity is expressed as: ;

[0121] These include, feed flow is too low: ;

[0122] Feed concentration is too low: ;

[0123] Feed pressure is too high: ;

[0124] in, , , Respectively represent feed concentration, feed pressure, and feed flow rate, , , They respectively represent the influence of feed flow rate on the change of overflow particle size, the influence of feed concentration on the change of overflow particle size, and the influence of feed pressure on the change of overflow particle size.

[0125] In the process of optimizing the grinding adjustment data by the classification model, the optimal solution is obtained by minimizing the objective function. The following is the expression of the optimal solution of minimizing the objective function:

[0126] ;

[0127] Among them, the control variables must meet the following constraints.

[0128] The classification model includes the following constraints in the process of optimizing grinding adjustment data:

[0129] ;

[0130] ;

[0131] in, Indicates the minimum value of the preset control variable; Indicates the maximum value of the preset control variable; The preset minimum change of the control variable, The preset maximum value of the control variable change.

[0132] Specifically, the constraints are as follows:

[0133] Feed flow :

[0134] ;

[0135] ;

[0136] Feed concentration :

[0137] ;

[0138] ;

[0139] Feed pressure :

[0140] ;

[0141] ;

[0142] Feed concentration: ; ;

[0143] Feed pressure: ; ;

[0144] Feed flow rate: ; .

[0145] Specifically, the optimization solution includes:

[0146] Gradient calculation; objective function For control variables Gradient for:

[0147] ;

[0148] For process logic related items, when or hour:

[0149] ;

[0150] when hour:

[0151] ;

[0152] represents the differential, Used to distinguish and Are they equal? hour, ,otherwise ;

[0153] It represents the sensitivity of the system output to the control variable through the nonlinear function Derived to get .

[0154] Initialization step; setting the initial parameter vector , the first moment estimate and the second moment estimate are all zero vectors. The step size is initialized to , the initial learning rate , the hyperparameters are set to , , .

[0155] Iterative update; each time step Perform the following steps:

[0156] Calculate gradient: Calculate the objective function For current parameters Gradient ;

[0157] Update first-order moment estimate: Update the first-order moment estimate using the current gradient, ;

[0158] Update the second-order moment estimate: Update the second-order moment estimate using the current gradient, ;

[0159] Bias correction: bias correction is performed on the first-order moment and the second-order moment. , ;

[0160] Calculate the adaptive learning rate correction term: Introduce the adaptive learning rate correction term, ;

[0161] Use update parameters: Update parameters using gradients corrected by adaptive learning rate. ;

[0162] Repeat the above iterative update steps until the stopping condition is met, such as the norm of the gradient is small enough, or the maximum number of iterations is reached.

[0163] In the grinding and classification process control particle size optimization system, the following control strategies can also be implemented;

[0164] Implement the first optimal control input:

[0165] ;

[0166] The above process is then repeated at the next time step.

[0167] Implementing optimal control input and applied to the system.

[0168] Time step 1:

[0169] Optimize control parameters: Optimize to get , , .

[0170] Adjust equipment: Adjust the feed concentration, pressure and flow rate of the cyclone according to the above abrasive adjustment data.

[0171] Time step 2:

[0172] New optimized control parameters: , , .

[0173] Perform Adjustment: Adjust the equipment based on the retrieved abrasive adjustment data to keep the system within the target particle size range.

[0174] Finally, the entire mineral classification process is completed according to real-time monitoring and feedback adjustment. See Table 3. The overflow particle size control is performed using the grinding and classification process control particle size optimization system of this embodiment and the conventional control system. The error between the overflow particle size finally obtained by the two systems and the target overflow particle size is compared.

[0175] Table 3: Error comparison table

[0176]

[0177] It can be seen from Table 3 that this embodiment effectively reduces the error of overflow particle size through nonlinear algorithm, random optimization, and the first relationship and the second relationship, realizes precise control of overflow particle size in the grinding and classification system, and ensures stable operation of the system and improvement of product quality.

[0178] Among them, real-time monitoring refers to the use of human-machine interface (HMI) to monitor system status and parameter changes;

[0179] Feedback adjustment refers to continuing to adjust the parameters in the optimization algorithm based on the actual overflow particle size feedback to ensure stable operation of the system.

[0180] This embodiment also provides a method for optimizing particle size during grinding and classification process control, which is applied to the above-mentioned system for optimizing particle size during grinding and classification process control. The method comprises the following steps:

[0181] Step 1: Collect the control variables and overflow particle size in the current grinding and classification process;

[0182] Step 2: Input the control variables and overflow particle size into the classification model to obtain grinding adjustment data for the target overflow particle size;

[0183] Step 201: Initialize the parameters of the classification model;

[0184] Step 202: Calculate the gradient of the objective function of the hierarchical model under the current parameters;

[0185] Step 203: updating the first-order moment estimation and the second-order moment estimation according to the gradient, performing deviation correction on the first-order moment estimation and the second-order moment estimation, introducing an adaptive learning correction term to correct the gradient, and reacquiring parameters according to the corrected gradient;

[0186] Step 204: Update the gradient and parameters, and repeat steps 202 to 203 until the norm of the gradient meets the preset norm or reaches the preset number of iterations.

[0187] Step 3: Adjust the control variables in the current grinding and classification process based on the grinding adjustment data.

[0188] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A grinding and classification process control particle size optimization system, characterized in that: include: An input module, a data acquisition module, a control module and an execution module, wherein the control module stores a hierarchical model constructed based on a first relationship model and a second relationship model in combination with a nonlinear algorithm and a random optimization algorithm; the first relationship model is a relationship model of a control variable, an external disturbance and an overflow granularity, and the second relationship model is a relationship model including a control variable, an external disturbance and a system state; The input module is used to input a target overflow particle size and transmit the target overflow particle size to the control module; The data acquisition module is used to collect the control variables and overflow particle size in the current grinding and classification process, and transmit the control variables and overflow particle size to the control module; The control module is used to input the control variables and the overflow particle size into the classification model, obtain the grinding adjustment data for the target overflow particle size, and transmit the grinding adjustment data to the execution module; The classification model predicts the system state at the target overflow particle size based on the second relationship model, and then obtains grinding adjustment data under the influence of the system state based on the first relationship model; The classification model obtains grinding adjustment data under the influence of system status based on the first relationship model, which is expressed as: ； Indicates the target overflow particle size Grinding adjustment data at all times; Indicates the target overflow particle size System status at the moment; express The control variable vector of the system at time ; express The system external disturbance vector at time ; A nonlinear function representing the relationship between overflow particle size and control variables; The nonlinear function of the relationship between the overflow particle size and the control variable is expressed by the following formula: ； in, It shows the relationship between overflow particle size and control variables; represents the system state vector, represents the control variable vector, represents the external interference vector; Indicates the linear effect of overflow particle size on grinding adjustment data, represents the linear effect of the control variable on the grinding adjustment data, Represents the linear effect of external disturbance on grinding adjustment data, The nonlinear characteristics and hysteresis effects of the relationship between overflow particle size and control variables; It is expressed by the following formula: ； in, , , as well as Empirical constants representing coefficients of nonlinear terms; represents the quadratic nonlinear effect of overflow particle size; represents the saturated nonlinear effect of the control variable; It indicates the nonlinear relationship between external disturbance and grinding adjustment data; represents the dynamic response lag of the system to the control variable, where represents the hysteresis coefficient, represents the integral time variable; The execution module is used to adjust the control variables in the current grinding and classification process based on the grinding adjustment data.

2. The grinding and classification process control particle size optimization system according to claim 1 is characterized in that: The system state at the time of predicting the target overflow granularity based on the second relational model is expressed as: ； in, A nonlinear function representing the state of the system; express The control variable vector of the system at time ; express The system external disturbance vector at time ; The nonlinear function of the system state is expressed by the following formula: ； in, Indicates the system status; Represents the dynamic characteristics of the system. represents the linear influence of the control variable on the system state, Represents the linear effect of external disturbance on the system state; Represents the nonlinear characteristics and hysteresis effects of the system state; It is expressed by the following formula: ； in, , , as well as Empirical constants representing coefficients of nonlinear terms; It represents the nonlinear influence of the control variables on the grinding and classification effect; Represents the nonlinear effects of system states; represents the nonlinear effect of the control variable; represents the dynamic response hysteresis of the system state, where Indicates the response speed of the system. Represents the time variable, which is the starting time of the integration.

3. The grinding and classification process control particle size optimization system according to any one of claims 1-2, characterized in that: When the control module obtains the grinding adjustment data through the classification model, the grinding adjustment data is optimized by minimizing the objective function; The objective function is expressed by the following formula: ； ； ； in, Indicates the target overflow particle size Grinding adjustment data at all times; Indicates the desired target overflow particle size; Indicates The change in the control variable; , as well as Represents the weight coefficient of balancing the grinding coefficient and the tracking error; Represents the weight coefficient of the penalty term or incentive term related to the grinding and classification process logic, Represents the penalty term or incentive term related to the grinding and classification process logic.

4. The grinding and classification process control particle size optimization system according to claim 3 is characterized in that: The control variables include feed flow rate, feed concentration and feed pressure; The penalty item or incentive item related to the grinding and classification process logic includes a particle size increase incentive item and a particle size decrease penalty item. The particle size increase incentive item is expressed as: ; in Including, the feed flow is too high: ; Feed concentration is too high: ; Feed pressure is too low: ; The granularity reduction penalty term is expressed as: ; in Including, the feed flow is too low: ; Feed concentration is too low: ; Feed pressure is too high: ; in, , , Respectively represent feed concentration, feed pressure, and feed flow rate, , , They respectively represent the influence of feed flow rate on the change of overflow particle size, the influence of feed concentration on the change of overflow particle size, and the influence of feed pressure on the change of overflow particle size.

5. The grinding and classification process control particle size optimization system according to claim 4 is characterized in that: The classification model includes the following constraints in the process of optimizing grinding adjustment data: ； ； in, Indicates the minimum value of the preset control variable; Indicates the maximum value of the preset control variable; Indicates the preset minimum change value of the control variable. Indicates the preset maximum value of the control variable change.

6. A method for optimizing particle size control in a grinding and classification process, characterized in that: A grinding and classification process control particle size optimization system applied to any one of claims 1-5, the method comprising the following steps: Step 1: Collect the control variables and overflow particle size in the current grinding and classification process; Step 2: Input the control variables and overflow particle size into the classification model to obtain grinding adjustment data for the target overflow particle size; Step 3: adjusting the control variables in the current grinding and classification process based on the grinding adjustment data.

7. The method for optimizing particle size during grinding and classification process according to claim 6, characterized in that: The classification model optimizes the grinding adjustment data, and the optimization includes the following steps: Step 201: Initialize the parameters of the classification model; Step 202: Calculate the gradient of the objective function of the hierarchical model under the current parameters; Step 203: updating the first-order moment estimation and the second-order moment estimation according to the gradient, performing deviation correction on the first-order moment estimation and the second-order moment estimation, introducing an adaptive learning correction term to correct the gradient, and reacquiring parameters according to the corrected gradient; Step 204: Update the gradient and parameters, and repeat steps 202 to 203 until the norm of the gradient meets the preset norm or reaches the preset number of iterations.

Images