Intelligent decision-making method and system for dry magnetic separator

Abstract

The invention discloses an intelligent decision-making method and system for a dry magnetic separator. The method comprises the following steps: acquiring movement tracks of ore particles with different attributes on a separation and separation belt under different control parameters of the dry magnetic separator on the basis of an image recognition technology; on the basis of the different control parameters and the corresponding movement tracks, track prediction models of the ore particles with the different attributes are constructed; a reference trajectory of target ore particles is obtained, and a control parameter optimization model is constructed with the maximum total distance between the plane where the ore separation plate is located and the edges of the magnetic and non-magnetic trajectories as an optimization target; the optimal solution of the control parameter optimization model is solved, sorting of the dry magnetic separator is controlled on the basis of the optimal control parameters corresponding to the optimal solution, the method is high in automation degree and does not depend on manual experience and a large amount of manpower any more, and intelligent support is provided for dry magnetic separation production.

Description

Technical Field

[0001] The present invention relates to the field of intelligent metallurgical control, and in particular to an intelligent decision-making method and system for a dry magnetic separator. Background Art

[0002] The ore is separated under the high-speed rotation of the separation magnetic drum, forming different ejection trajectories. Non-magnetic particles are not affected by the magnetic field force, and are thrown into the non-magnetic material bucket under the combined action of gravity and centrifugal force; while the magnetic particles are adsorbed on the surface of the magnetic drum due to the magnetic field force, and then fall to the non-magnetic area. Depending on whether the particles are magnetic or not, their trajectories have obvious regional separation characteristics. In recent years, with the development of machine vision technology, it has become possible to use machine vision to identify the moving particle trajectories of magnetic separators. Based on this, we can use the extracted particle trajectories to adjust the position of the separation plate in combination with production needs.

[0003] At present, the control of the separation plate of the dry magnetic separator mainly relies on manual visual inspection for operation and adjustment; however, the manual method has great limitations, and its accuracy is easily affected by the subjective experience and physical condition of the workers. It lacks a quantitative standard, and it is difficult to make timely adjustments according to the separation zone, and it is difficult to achieve fine-tuning of small angles (<0.5°), so the separation effect cannot be guaranteed. Summary of the invention

[0004] The present invention provides an intelligent decision-making method and system for a dry magnetic separator, which is used to solve the problems of large errors, low efficiency and high manpower requirements caused by using artificial experience methods to operate and adjust the separation plates in the traditional dry magnetic separation process.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0006] An intelligent decision-making method for a dry magnetic separator comprises the following steps:

[0007] Step S1, obtaining from historical data the movement trajectories of ore particles with different properties on the separation belt under different control parameters of the dry magnetic separator;

[0008] Step S2, constructing trajectory prediction models of ore particles with different attributes based on the different control parameters and their corresponding motion trajectories;

[0009] Step S3, obtaining a reference trajectory of the target ore particle, and taking the minimum difference between the reference trajectory and the predicted trajectory of the trajectory prediction model as the optimization target, constructing a control parameter optimization model;

[0010] Step S4, solving the optimal solution of the control parameter optimization model, and controlling the dry magnetic separator to separate based on the optimal control parameters corresponding to the optimal solution.

[0011] Preferably, the step S1 comprises:

[0012] Obtaining trajectory images of ore particles with different attributes on the separation belt from historical data, and marking the trajectories on the trajectory images;

[0013] A motion trajectory extraction model is constructed based on an image processing algorithm, and the labeled trajectory image is used as training data to train the motion trajectory extraction model. When the trained trajectory matching rate reaches a preset value, the motion trajectory extraction model in this state is used to extract the motion trajectories of ore particles with different properties on the sorting belt.

[0014] 3. The intelligent decision-making method for a dry magnetic separator according to claim 2, characterized in that the trajectory coincidence rate satisfies the following formula:

[0015]

[0016] Among them, rate is the recognition consistency, y std is the ordinate of the standard track, y is the ordinate of the recognition track, and n is the total number of horizontal pixels.

[0017] Preferably, the properties of the ore particles include whether the ore particles are magnetic;

[0018] When the ore particles are magnetic, the trajectory prediction model of the ore particles is:

[0019]

[0020] in,

[0021] μ0 is the vacuum magnetic permeability; k is the magnetic susceptibility of the particle, d is the particle radius, ρ is the particle density, g is the gravitational acceleration; B is the magnetic field strength in the magnetic separator chamber, the unit is A / m 2 ; gradB is the magnetic field gradient, in A / m 2 ; x is the coordinate of the particle on the X-axis of the preset vertical coordinate system; y is the coordinate of the particle on the Y-axis of the preset vertical coordinate system, the preset vertical coordinate system is set on the cross section of the magnetic separator drum, and takes the axis of the magnetic separator drum as the origin, the horizontal direction as the X-axis direction, and the vertical direction as the Y-axis direction; t is time; α is the angle between the normal direction of the particle and the Y-axis;

[0022] When the ore particles are non-magnetic, the trajectory prediction model of the ore particles is:

[0023]

[0024] Among them, g is the acceleration of gravity; h is the height of the feeder outlet from the top of the drum; R is the radius of the drum.

[0025] Preferably, the objective function of the control parameter optimization model is:

[0026]

[0027] Among them, Q represents the output error weighting coefficient matrix, R represents the control increment weighting coefficient matrix, Represents the future output prediction matrix, Y r represents the output reference trajectory, Represents the future input matrix.

[0028] 6. The intelligent decision-making method for a dry magnetic separator according to claim 4, characterized in that the trajectory prediction model is solved by a support vector machine algorithm, and the optimal hyperplane of the support vector machine algorithm satisfies:

[0029]

[0030] in, is the hyperplane normal vector, b is the hyperplane offset;

[0031] The support vector machine algorithm transforms the objective function into a quadratic optimization problem:

[0032]

[0033] Among them, the constraints are: i (ω T x i +b)≥1 i=1,2……n.

[0034] Preferably, the control parameters include an adjustment angle of a sorting plate and a drum speed, and the control of the dry magnetic separator for sorting based on the optimal control parameters corresponding to the optimal solution includes:

[0035] Get the target input y of the dry magnetic separator at the kth moment r (k) and the actual output y(k) on the dry magnetic separator at the kth moment, the control input is the preset adjustment angle of the sorting plate and the drum speed, and the control output is the real-time sorting plate trajectory and sorting plate angle;

[0036] And based on the target input y r (k) and the actual output y(k) to calculate the reference trajectory r(k+i) at the k+ith moment;

[0037] The reference trajectory r(k+i) is subjected to rolling optimization to obtain an optimal control instruction, and the dry magnetic separator is controlled based on the optimal control instruction.

[0038] Preferably, the reference trajectory is obtained by the following formula:

[0039]

[0040] Among them, T ref is the time constant and is the set value.

[0041] Preferably, before rolling optimization is performed on the reference trajectory r(k+i), the following steps are also included:

[0042] Obtain the optimal control instruction u(k) at the kth moment, and calculate the predicted trajectory y corresponding to the optimal control instruction u(k) based on the optimal control instruction u(k) and the trajectory prediction model c (k), calculate the predicted trajectory y c The reference trajectory r(k+i) is corrected according to the difference e(k) between the reference trajectory and the actual trajectory at the corresponding moment.

[0043] A computer system comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the computer program.

[0044] The present invention has the following beneficial effects:

[0045] 1. The intelligent decision-making method for the position of the separation plate of the dry magnetic separator in the present invention adopts the MPC algorithm for the position of the separation plate that directly affects the separation effect, and realizes intelligent predictive control through the process of rolling optimization and feedback correction; and the method has the advantage of strong real-time performance, which can meet the real-time requirements of high-speed dry magnetic separators and improve work efficiency; in addition, the method has a high degree of automation and no longer relies on manual experience and a large amount of manpower, providing intelligent support for dry magnetic separation production.

[0046] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0048] Figure 1 It is a flow chart of the intelligent decision-making method for the position of the dry magnetic separation ore separation plate of the present invention;

[0049] Figure 2 It is the control loop constructed by the present invention.

[0050] Figure 3 This is a force analysis diagram of the magnetic separation process. In the figure, the XOY coordinate system takes the center of the drum as the coordinate origin, the horizontal direction is the X-axis, and the right direction is the positive direction, and the vertical direction is the Y-axis, and the upward direction is the positive direction. DETAILED DESCRIPTION

[0051] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0052] Embodiment 1:

[0053] See also Figure 1-3 The present invention constructs an intelligent decision method for the position of the ore separation plate of a dry magnetic separator, which can be applied to a mineral processing plant to perform intelligent control of the ore separation plate of a dry magnetic separator in a primary circuit system.

[0054] See also Figure 1 In a preferred embodiment of the present invention, an intelligent decision method for the position of a separation plate of a dry magnetic separator is provided, comprising the following steps:

[0055] Step 1: Collect feedforward image information and convert the abstract particle trajectory image into quantitative data, that is, extract the accurate trajectory of the sorting zone.

[0056] In step 1, the image recognition consistency is calculated using the following formula:

[0057]

[0058] Among them, rate is the recognition consistency, y std is the ordinate of the standard track, y is the ordinate of the recognition track, and n is the total number of horizontal pixels.

[0059] In a preferred embodiment of the present invention, a servo control system and an industrial high-speed camera are used to collect dry magnetic separator information, particle trajectory images and environmental background, specifically including: feed rate, drum speed, magnetic field strength; particle trajectory images of different types, sizes, working conditions, frequencies, and magnetic strengths and their corresponding environmental background images; the environmental background image includes ambient lighting and background color data.

[0060] In the present invention, the required accuracy is 90% of the agreement, and the agreement of the samples selected in this embodiment is higher than 92%, which proves that the method can effectively provide the ore band sorting image information required for the control of the ore separation plate.

[0061] Step 2: The controller obtains the predicted trajectory of the controlled variable in the prediction time domain through the system's operating variables, historical values ​​of the controlled variables and the corresponding mathematical model.

[0062] In step 2, the prediction model part is used to predict the future controlled variables based on the given manipulated variables.

[0063] In a preferred embodiment of the present invention, the operating variables include: the adjustment angle of the ore separation plate. The controlled variables include: the optimal position of the ore separation plate.

[0064] In a preferred embodiment of the present invention, the process parameters such as feed rate, material magnetic susceptibility, drum speed, magnetic field strength, etc. are collected in step 1, and the friction force, electromagnetic force, etc. are calculated to obtain the force analysis and acceleration of the magnetic particles in the magnetic separation process. In this way, the motion reference trajectories of the magnetic particles and the non-magnetic particles are further obtained, that is, the mathematical model required for step 2.

[0065] The XOY coordinate system is established with the center of the magnetic separator drum as the origin, and the mathematical model required for step 2 can be expressed as follows:

[0066] Movement trajectory of non-magnetic materials:

[0067]

[0068] Where g is the acceleration due to gravity, which is 9.8 m / s 2 ; h is the height of the feeder outlet from the top of the drum; R is the radius of the drum.

[0069] Movement trajectory of magnetic materials:

[0070]

[0071] Where x is the position of the particle on the X-axis; y is the position of the particle on the Y-axis; n is a natural number 1, 2, ...; t is time, s; α is the angle between the normal direction of the particle and the Y-axis, (°).

[0072] Step 3: Based on the predicted values ​​of step 2, optimize the current operating variables.

[0073] In step 3, the optimal control law in the control time domain (let’s say M) is obtained through optimization, and the first step is executed, that is, the optimal operation value at time T = k is obtained. This is the MPC control process.

[0074] The preferred embodiment of the present invention uses a support vector machine (SVM) to find the optimal position of the scoring plate.

[0075] The above SVM optimal hyperplane can be expressed as follows:

[0076]

[0077] in, is the hyperplane normal vector, and b is the hyperplane offset.

[0078] [In the present invention, the optimal position of the ore separation plate is transformed into a secondary optimization problem:

[0079]

[0080] The constraints are: i (ω T x i +b)≥1i=1,2……n, which means that the separation plate should be as far away from the separation zone as possible.

[0081] The controller is built based on the MPC principle, and its specific structure is as follows Figure 2 The meanings of the symbols are as follows: r represents the set value of the controlled variable; r(k+i) represents the reference trajectory, that is, the current output value y(k) approaches the set value y r The trajectory of the ore separation plate; u(k) represents the calculated optimal control rate, that is, the angle at which the ore separation plate should be adjusted; y c (k) represents the trajectory prediction value obtained by the prediction model; y(k) represents the true value of the parameter in the above sorting process; e(k) represents the feedback prediction error, which is numerically equal to y(k)-y c (k).

[0082] In this embodiment, the setting of the reference trajectory of the controlled variable is free. The reference trajectory approaches the set value in an exponential form with a time constant T ref As a setting parameter, the value of the reference trajectory r(k+i) at each moment in the future is:

[0083]

[0084] The MPC (model predictive control) controller designed in the preferred embodiment of the present invention utilizes an open source optimization tool - the interior point solver IPOPT, which efficiently searches and determines the optimal solution of the objective function J at time k in an online manner, thereby generating a control instruction sequence u(k), u(k+1),...u(k+m) for the system at the current time point.

[0085] The above objective function is as follows:

[0086]

[0087] Among them, Q represents the output error weighting coefficient matrix, R represents the control increment weighting coefficient matrix, Represents the future output prediction matrix, Y r represents the output reference trajectory, Represents the future input matrix.

[0088] In some embodiments, the system's predicted values ​​may differ significantly from the measured values.

[0089] To correct the above differences, the feedback correction module multiplies the prediction error e(k) at time T=k by a certain coefficient at time T=k+1, and uses this to correct the new reference trajectory.

[0090] Step 4: Output the optimal control rate obtained in step 3 to the control system.

[0091] In summary, the intelligent decision-making method for the position of the ore separator of the dry magnetic separator in the preferred embodiment of the present invention proposes a dry magnetic separator ore separator position optimization strategy that integrates visual analysis and intelligent control technology. For the position of the ore separator that directly affects the sorting effect, the MPC algorithm is adopted, and intelligent predictive control is realized through the process of rolling optimization and feedback correction; and this method has the advantage of strong real-time performance, which can meet the real-time requirements of high-speed dry magnetic separators and improve work efficiency; in addition, this method has a high degree of automation and no longer relies on manual experience and a large amount of manpower, laying a foundation for intelligent optimization technology for realizing intelligent manufacturing of metallurgy.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent decision-making method for a dry magnetic separator, characterized in that: The following steps are involved: Step S1, obtaining from historical data the movement trajectories of ore particles with different properties on the separation belt under different control parameters of the dry magnetic separator; Step S2, constructing trajectory prediction models of ore particles with different attributes based on the different control parameters and their corresponding motion trajectories; Step S3, obtaining a reference trajectory of the target ore particle, and taking the minimum difference between the reference trajectory and the predicted trajectory of the trajectory prediction model as the optimization target, constructing a control parameter optimization model; Step S4, solving the optimal solution of the control parameter optimization model, and controlling the dry magnetic separator to separate based on the optimal control parameters corresponding to the optimal solution.

2. The intelligent decision-making method for a dry magnetic separator according to claim 1, characterized in that: The step S1 comprises: Obtaining trajectory images of ore particles with different attributes on the separation belt from historical data, and marking the trajectories on the trajectory images; A motion trajectory extraction model is constructed based on an image processing algorithm, and the labeled trajectory image is used as training data to train the motion trajectory extraction model. When the trained trajectory matching rate reaches a preset value, the motion trajectory extraction model in this state is used to extract the motion trajectories of ore particles with different properties on the sorting belt.

3. The intelligent decision-making method for a dry magnetic separator according to claim 2, characterized in that: The trajectory matching rate satisfies the following formula: Among them, rate is the recognition consistency, y std is the ordinate of the standard track, y is the ordinate of the recognition track, and n is the total number of horizontal pixels.

4. The intelligent decision-making method for a dry magnetic separator according to claim 3, characterized in that: The properties of the ore particles include whether the ore particles are magnetic; When the ore particles are magnetic, the trajectory prediction model of the ore particles is: in, μ0 is the vacuum magnetic permeability; k is the magnetic susceptibility of the particle, d is the particle radius, ρ is the particle density, g is the gravitational acceleration; B is the magnetic field strength in the magnetic separator chamber, the unit is A / m 2 ; gradB is the magnetic field gradient, in A / m 2 ; x is the coordinate of the particle on the X-axis of the preset vertical coordinate system; y is the coordinate of the particle on the Y-axis of the preset vertical coordinate system, the preset vertical coordinate system is set on the cross section of the magnetic separator drum, and takes the axis of the magnetic separator drum as the origin, the horizontal direction as the X-axis direction, and the vertical direction as the Y-axis direction; t is time; α is the angle between the normal direction of the particle and the Y-axis; When the ore particles are non-magnetic, the trajectory prediction model of the ore particles is: Among them, g is the acceleration of gravity; h is the height of the feeder outlet from the top of the drum; R is the radius of the drum.

5. The intelligent decision-making method for a dry magnetic separator according to claim 4, characterized in that: The objective function of the control parameter optimization model is: Among them, Q represents the output error weighting coefficient matrix, R represents the control increment weighting coefficient matrix, Represents the future output prediction matrix, Y r represents the output reference trajectory, Represents the future input matrix.

6. The intelligent decision-making method for a dry magnetic separator according to claim 4, characterized in that: The trajectory prediction model is solved by a support vector machine algorithm, and the optimal hyperplane of the support vector machine algorithm satisfies: in, is the hyperplane normal vector, b is the hyperplane offset; The support vector machine algorithm transforms the objective function into a quadratic optimization problem: Among them, the constraints are: i (ω T x i +b)≥1i=1,2……n.

7. The intelligent decision-making method for a dry magnetic separator according to claim 5, characterized in that: The control parameters include the adjustment angle of the sorting plate and the drum speed, and the control of the dry magnetic separator for sorting based on the optimal control parameters corresponding to the optimal solution includes: Get the target input y of the dry magnetic separator at the kth moment r (k) and the actual output y(k) on the dry magnetic separator at the kth moment, the control input is the preset adjustment angle of the sorting plate and the drum speed, and the control output is the real-time sorting plate trajectory and sorting plate angle; And based on the target input y r (k) and the actual output y(k) to calculate the reference trajectory r(k+i) at the k+ith moment; The reference trajectory r(k+i) is subjected to rolling optimization to obtain an optimal control instruction, and the dry magnetic separator is controlled based on the optimal control instruction.

8. The intelligent decision-making method for a dry magnetic separator according to claim 7, characterized in that: The reference trajectory is obtained by the following formula: Among them, T ref is the time constant and is the set value.

9. The intelligent decision-making method for a dry magnetic separator according to claim 6, characterized in that: Before rolling optimization is performed on the reference trajectory r(k+i), the following steps are also included: Obtain the optimal control instruction u(k) at the kth moment, and calculate the predicted trajectory y corresponding to the optimal control instruction u(k) based on the optimal control instruction u(k) and the trajectory prediction model c (k), calculate the predicted trajectory y c The reference trajectory r(k+i) is corrected according to the difference e(k) between the reference trajectory and the actual trajectory at the corresponding moment.

10. A computer system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

Images