Method and system for parameter optimization of xrt ore sorter imaging structure

Abstract

The application discloses a kind of parameter optimization method and system of XRT ore sorting machine imaging structure, including obtaining the structure parameter information of target XRT ore sorting machine;In simulation software, the simulation model of target XRT ore sorting machine is constructed;The influence of each element in X-ray imaging structure on image signal-to-noise ratio is studied and the final influencing factor is determined;With signal-to-noise ratio maximum and X-ray source power minimum as optimization goal, with actual working condition and process condition as constraint, the parameter optimization model of XRT ore sorting machine imaging structure is constructed and solved, and the parameter optimization of target XRT ore sorting machine imaging structure is completed.The application is accurately modeled and simulated to target XRT ore sorting machine, and the construction and solution of targeted parameter optimization model are carried out after determining the influencing factor of parameter, not only realize the parameter optimization of target XRT ore sorting machine imaging structure, but also the scheme of the application is relatively simple, more accurate, and better effect.

Description

Technical Field

[0001] This invention belongs to the field of mining equipment technology, specifically relating to a parameter optimization method and system for the imaging structure of an XRT ore sorting machine. Background Technology

[0002] XRT ore separators are intelligent mineral processing equipment that uses XRT technology to separate ores. In an XRT ore separator, the ore is evenly distributed at the beginning of a conveyor belt by a vibrating feeder and retaining rubber sheets. It is then transported by the conveyor belt to a position above a detector. An X-ray source emits X-rays that penetrate the ore, illuminating the detector and forming an image. This image information is then transmitted to a signal processing center, which analyzes the ore grade and distribution location, controlling the air nozzles to blow away the tailings, thus completing the ore separation. In the operation of an XRT ore separator, the most critical aspect—ore imaging—relies on the X-ray sensor detection system, which is a prerequisite for ore separation. The imaging structure is a key component responsible for the ore imaging part. A schematic diagram of the X-ray imaging structure is shown below. Figure 1 As shown in the figure, the labels are: 1 is the lead cylinder, 2 is the lead-antimony collimator, 3 is the protective box, 4 is the protective partition, 5 is the lead plate under the conveyor belt, 6 is the detector card, 7 is the tungsten target, and 8 is the beryllium filter. The imaging principle is: the conveyor belt transports the ore to the area to be detected. The X-rays generated by the electron bombardment of the tungsten target penetrate the ore and irradiate the detector. The detector captures the X-ray signal and converts it into an electrical signal to form an image of the ore.

[0003] The signal-to-noise ratio (SNR) of X-ray imaging images is one of the key indicators for evaluating image quality. It reflects the detectability of useful information in an image amidst noise. A high SNR means that more useful information can be identified in the image, enabling more accurate identification of useful minerals and waste rock, thereby effectively improving the sorting efficiency and accuracy of XRT ore sorting machines.

[0004] Current research mainly focuses on X-ray image processing and X-ray intensity enhancement. X-ray image processing typically employs various algorithms (such as numerical simulation, wavelet transform, transient theory noise model, and coherence analysis) to process the obtained images and obtain images with higher signal-to-noise ratios; however, these methods are often extremely complex. While X-ray intensity enhancement seems relatively simple, excessively high-intensity X-rays are difficult for ores to effectively attenuate, making the information they carry difficult to identify. Furthermore, excessively high-intensity X-ray sources can cause excessive X-ray leakage, increasing the radiation protection costs of XRT ore sorting machines. Summary of the Invention

[0005] One of the objectives of this invention is to provide a parameter optimization method for the imaging structure of an XRT ore sorting machine that is relatively simple, more accurate, and has better results.

[0006] The second objective of this invention is to provide a system for optimizing the parameters of the imaging structure of the XRT ore sorting machine.

[0007] The parameter optimization method for the imaging structure of the XRT ore sorting machine provided by this invention includes the following steps:

[0008] S1. Obtain the structural parameter information of the target XRT ore separator;

[0009] S2. Based on the structural parameter information obtained in step S1, and combining the Monte Carlo simulation scheme, Boltzmann equation, cross-validation scheme and polynomial fitting scheme, construct a simulation model of the target XRT ore separator in the simulation software.

[0010] S3. Based on the simulation model obtained in step S2, perform single-factor analysis using the control variable method to study the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, and determine the final influencing factors.

[0011] S4. Based on the influencing factors obtained in step S3, with the maximum signal-to-noise ratio and the minimum X-ray source power as optimization objectives, and with actual working conditions and process conditions as constraints, construct a parameter optimization model for the imaging structure of the XRT ore sorting machine.

[0012] S5. Based on the genetic algorithm, solve the model constructed in step S4 to complete the parameter optimization of the target XRT ore sorting machine imaging structure.

[0013] Step S2 includes the following steps:

[0014] S2.1. Based on the structural parameter information obtained in step S1, construct a three-dimensional structural simulation model of the target XRT ore separator in the simulation software;

[0015] S2.2. Based on the Monte Carlo simulation scheme and Boltzmann equation, the propagation process and interaction of X-rays in the target XRT ore separator are simulated;

[0016] S2.3. Based on the cross-validation scheme and the polynomial fitting scheme, establish the mathematical relationship between X-ray energy and quantity and image signal-to-noise ratio.

[0017] Step S2.1 specifically includes the following steps:

[0018] Taking the X-ray imaging structure of the target XRT ore separator as the research object, and taking the ray emission point as the origin, a simulation model of the X-ray imaging structure of the target XRT ore separator was established using Geant4 simulation software.

[0019] The constructed X-ray imaging structure simulation model includes a lead cylinder, a lead-antimony collimator, a protective box and protective partition, a lead plate under the belt, and a scintillator for the detector card.

[0020] Step S2.2 specifically includes the following steps:

[0021] Based on the Monte Carlo simulation scheme, a set of state parameters is set for each particle to record all its current motion states; the state parameters include the particle's spatial coordinates. ,speed The unit vector corresponding to the direction of motion Particle weight values and time ;

[0022] Ignoring the interactions between particles, the Boltzmann equations are used to record the motion of the particles, as follows:

[0023]

[0024] In the formula The particle distribution function; The spatial gradient of the particle distribution function; For the particle to be in And the speed magnitude is The total cross section at that time; The magnitude of the particle's velocity before the collision; The direction of motion of the particle before the collision; The collision time; Let be the scattering function, used to represent a velocity at position r. And the direction is The particles, after the collision time Then, it is scattered to the velocity And the direction is The conditional probability within a unit velocity and a unit solid angle; For the particle to be in And the speed magnitude is The total cross section of particle scattering at that time; The source term is used to represent time. In position At a point, within a unit time, unit volume, or unit solid angle, the energy produced has velocity. and direction The number of particles;

[0025] The process of simulating the particle output problem based on the Monte Carlo simulation scheme includes the following steps:

[0026] (1) Using normalized source term distribution Sampling is performed to determine the initial state of the particles. The initial weights of the particles are calculated using the following formula. for ;in, The total volume of the computational domain;

[0027] (2) Calculate the particle collision coordinates: The distance the particle travels is calculated using the following formula. :

[0028]

[0029] In the formula This indicates that the particle is in a state of The total cross section at that time; To solve for the distance the particle traveled The integral variable in the integral expression; The free path of particle transport;

[0030] The particle collision coordinates were calculated. for ;

[0031] (3) Determine if the particle is within the simulation area: Set the particle's operating area as... When the particle leaves the region When the particle is in the region, the simulation stops; Then continue the simulation;

[0032] (4) Calculate the particle collision time for ;

[0033] (5) Update the cumulative value of the target physical quantity using the following formula:

[0034]

[0035] In the formula This represents the current cumulative target physical quantity; This refers to the previously accumulated target physical quantity; The global scale factor is set. The weighting function for the particles;

[0036] (6) Calculate the velocity and corresponding unit direction vector of the particle after the collision, and use them to iteratively update the particle;

[0037] The scattering effect produced after particle collision is due to Confirmation by distributed sampling;

[0038] Calculate the time of particle scattering effect for This is to achieve iterative updates of particle states; among which, This is the delay time after the collision, and according to Determined by distribution sampling;

[0039] (7) The particle weights after the scattering effect are calculated using the following formula:

[0040]

[0041] In the formula The particle weights after the scattering effect occurs; It is the scattering cross section; This is the total cross section after the scattering effect occurs;

[0042] Update the particle's state parameters;

[0043] (8) Repeat steps (2) to (7) for iterative calculation until the number of calculated particles reaches the total number of particles; end the iterative calculation.

[0044] Step S2.3 specifically includes the following steps:

[0045] Under the same parameter conditions, several sets of data on average X-ray energy, number of X-rays and image signal-to-noise ratio were obtained through simulation.

[0046] Based on the obtained data, a five-fold cross-validation scheme was used to obtain the difference between the training error and the validation error in first-order, second-order, and third-order polynomials.

[0047] A polynomial fitting scheme is used for fitting, and the following formula is used as the final fitting function:

[0048]

[0049] In the formula Image signal-to-noise ratio; These are the first fitted parameters; The average energy of the X-rays; These are the second fitting parameters; The number of X-rays; The third fitting parameter; The fourth fitting parameter; This is the fifth fitting parameter.

[0050] Step S3 specifically includes the following steps:

[0051] Based on the simulation model obtained in step S2, a single-factor analysis using the controlled variable method was conducted to study the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, and to determine the final influencing factors, including tube voltage. tube current Target angle beryllium filter thickness , width of lead cylinder opening , lead-antimony collimator opening width , width of protective partition opening and the width of the lead plate opening under the belt .

[0052] Step S4 specifically includes the following steps:

[0053] The following formula is used as the sub-objective function for the signal-to-noise ratio. :

[0054]

[0055] In the formula It is a constant value; This is the first weight value; This is the second weight value; This is the third weight value; Let be the value of the i-th influencing factor; Let be the value of the j-th influencing factor; Corresponding to the tube voltage in sequence tube current Target angle beryllium filter thickness , width of lead cylinder opening , lead-antimony collimator opening width , width of protective partition opening and the width of the lead plate opening under the belt The value of ;

[0056] The following formula is used as the sub-target function for X-ray source power. :

[0057]

[0058] With the optimization objectives of maximizing the signal-to-noise ratio and minimizing the X-ray source power, the objective function is constructed as follows: ;

[0059] Using actual operating conditions and process conditions as constraints, the constraints are constructed as follows:

[0060]

[0061] In the formula This represents the minimum value of the i-th influencing factor, set according to actual working conditions and process conditions. This represents the maximum value of the i-th influencing factor, set according to actual working conditions and process conditions.

[0062] Step S5 specifically includes the following steps:

[0063] The NSGA-II algorithm is used to solve the model constructed in step S4 based on the obtained Pareto curve, thereby completing the parameter optimization of the imaging structure of the target XRT ore sorting machine.

[0064] This invention also provides a system for optimizing the parameters of the imaging structure of the XRT ore separator, comprising a structural parameter acquisition module, a simulation model construction module, an influencing factor determination module, an optimization module construction module, and a parameter optimization module; the structural parameter acquisition module, simulation model construction module, influencing factor determination module, optimization module construction module, and parameter optimization module are connected in series; the structural parameter acquisition module is used to acquire the structural parameter information of the target XRT ore separator and upload the data information to the simulation model construction module; the simulation model construction module is used to construct a simulation model of the target XRT ore separator in simulation software based on the received data information, the acquired structural parameter information, and combining Monte Carlo simulation scheme, Boltzmann equation, cross-validation scheme, and polynomial fitting scheme, and upload the data information... The module for determining influencing factors uploads data to the XRT ore sorting machine imaging structure. This module uses the received data and simulation model to perform single-factor analysis based on the controlled variable method, studying the impact of each element in the XRT imaging structure on the image signal-to-noise ratio (SNR) and determining the final influencing factors. The data is then uploaded to the optimization module. The optimization module, based on the received data and influencing factors, constructs a parameter optimization model of the XRT ore sorting machine imaging structure, with the maximum SNR and minimum XRT source power as optimization objectives and actual operating conditions and process constraints as constraints. This model is then uploaded to the parameter optimization module. The parameter optimization module, based on the received data and a genetic algorithm, solves the constructed model to optimize the parameters of the target XRT ore sorting machine imaging structure.

[0065] The parameter optimization method and system for the imaging structure of the XRT ore sorting machine provided by this invention, through precise modeling and simulation of the target XRT ore sorting machine, and after determining the influencing factors of the parameters, constructing and solving a targeted parameter optimization model, not only achieves parameter optimization of the imaging structure of the target XRT ore sorting machine, but also the solution of this invention is relatively simple, more accurate, and has better results. Attached Figure Description

[0066] Figure 1This is a schematic diagram of the X-ray imaging structure of an existing XRT ore sorting machine; where, Figure 1 (a) is a frontal three-dimensional schematic diagram of the X-ray imaging structure. Figure 1 (b) is a three-dimensional sectional view of the X-ray imaging structure from the side.

[0067] Figure 2 This is a schematic diagram of the method flow of the present invention.

[0068] Figure 3 This diagram illustrates the proportions of the three effects under different X-ray energies and atomic numbers of different substances in an embodiment of the method of the present invention.

[0069] Figure 4 This is a schematic diagram showing the variation curves of training error and verification error under different orders in an embodiment of the method of the present invention.

[0070] Figure 5 This is a schematic diagram of the fitting result surface of an embodiment of the method of the present invention.

[0071] Figure 6 This is a schematic diagram of the residual distribution in an embodiment of the method of the present invention.

[0072] Figure 7 This is a schematic diagram illustrating the signal-to-noise ratio variation under different parameters in an embodiment of the method of the present invention; wherein, Figure 7 (a) is a schematic diagram of the signal-to-noise ratio variation curves under different transistor voltages. Figure 7 (b) is a schematic diagram of the signal-to-noise ratio variation curves under different tube currents. Figure 7 (c) is a schematic diagram of the signal-to-noise ratio variation curves under different target angles. Figure 7 (d) is a schematic diagram showing the signal-to-noise ratio variation curves under different beryllium filter thicknesses. Figure 7 (e) is a schematic diagram showing the signal-to-noise ratio variation curves under different lead cylinder opening widths. Figure 7 (f) is a schematic diagram showing the signal-to-noise ratio variation curves under different lead-antimony collimator aperture widths. Figure 7 (g) is a schematic diagram of the signal-to-noise ratio variation curves under different opening widths of the protective partition. Figure 7 (h) is a schematic diagram of the signal-to-noise ratio variation curves under different belt opening widths of the lead plate.

[0073] Figure 8 This is a schematic diagram of the Pareto front changing with the number of iterations in an embodiment of the method of the present invention; wherein, Figure 8 (a) is a schematic diagram of the curve after one iteration. Figure 8 (b) is a schematic diagram of the curve after 10 iterations. Figure 8 (c) is a schematic diagram of the curve after 50 iterations. Figure 8 (d) is a schematic diagram of the curve after 100 iterations.

[0074] Figure 9 This is a schematic diagram showing the changes in image signal-to-noise ratio before and after optimization under different X-ray source powers in an embodiment of the method of the present invention.

[0075] Figure 10 This is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0076] like Figure 2 The diagram shown is a flowchart of the method of the present invention: The parameter optimization method for the imaging structure of the XRT ore sorting machine disclosed in this invention includes the following steps:

[0077] S1. Obtain the structural parameter information of the target XRT ore separator.

[0078] S2. Based on the structural parameter information obtained in step S1, and combining the Monte Carlo simulation scheme, Boltzmann equation, cross-validation scheme, and polynomial fitting scheme, construct a simulation model of the target XRT ore separator in the simulation software.

[0079] The specific implementation includes the following steps:

[0080] S2.1. Based on the structural parameter information obtained in step S1, construct a three-dimensional structural simulation model of the target XRT ore separator in the simulation software; specifically including the following steps:

[0081] Taking the X-ray imaging structure of the target XRT ore separator as the research object, and taking the ray emission point as the origin, a simulation model of the X-ray imaging structure of the target XRT ore separator was established using Geant4 simulation software.

[0082] The constructed X-ray imaging structure simulation model includes a lead cylinder, a lead-antimony collimator, a protective box and protective partition, a lead plate under the belt, and a scintillator for the detector card.

[0083] S2.2. Based on the Monte Carlo simulation scheme and Boltzmann equations, the propagation process and interaction of X-rays in the target XRT ore separator are simulated; specifically, the following steps are included:

[0084] In the simulation of X-ray imaging structures, electrons originate from the cathode of the X-ray source, are accelerated by the electric field of the tube voltage, and bombard the tungsten target, generating continuous X-rays and characteristic X-rays, as well as the interaction between X-rays and the imaging structure in the subsequent imaging process. These are all general problems of particle transport processes. Therefore, the Monte Carlo method is used to simulate the propagation process and interaction of electrons and X-rays in the X-ray imaging structure.

[0085] Based on the Monte Carlo simulation scheme, a set of state parameters is set for each particle to record all its current motion states; the state parameters include the particle's spatial coordinates. ,speed The unit vector corresponding to the direction of motion Particle weight values and time ;

[0086] Ignoring the interactions between particles, the Boltzmann equations are used to record the motion of the particles, as follows:

[0087]

[0088] In the formula The particle distribution function; The spatial gradient of the particle distribution function; For the particle to be in And the speed magnitude is The total cross section at that time; The magnitude of the particle's velocity before the collision; The direction of motion of the particle before the collision; The collision time; Let be the scattering function, used to represent a velocity at position r. And the direction is The particles, after the collision time Then, it is scattered to the velocity And the direction is The conditional probability within a unit velocity and a unit solid angle; For the particle to be in And the speed magnitude is The total cross section of particle scattering at that time; The source term is used to represent time. In position At a point, within a unit time, unit volume, or unit solid angle, the energy produced has velocity. and direction The number of particles;

[0089] The process of simulating the particle output problem based on the Monte Carlo simulation scheme includes the following steps:

[0090] (1) Using normalized source term distribution Sampling is performed to determine the initial state of the particles. The initial weights of the particles are calculated using the following formula. for ;in, The total volume of the computational domain;

[0091] (2) Calculate the particle collision coordinates: The distance the particle travels is calculated using the following formula. :

[0092]

[0093] In the formula This indicates that the particle is in a state of The total cross section at that time; For integration variables; The free path of particle transport;

[0094] The particle collision coordinates were calculated. for ;

[0095] (3) Determine if the particle is within the simulation area: Set the particle's operating area as... When the particle leaves the region When the particle is in the region, the simulation stops; Then continue the simulation;

[0096] (4) Calculate the particle collision time for ;

[0097] (5) Update the cumulative value of the target physical quantity using the following formula:

[0098]

[0099] In the formula This represents the current cumulative target physical quantity; This refers to the previously accumulated target physical quantity; The global scale factor is set (related to system geometry and simulation architecture). The weighting function for the particles;

[0100] (6) Calculate the velocity and corresponding unit direction vector of the particle after the collision, and use them to iteratively update the particle;

[0101] The collision of particles produces a scattering effect, and the velocity... Things have changed, from Confirmation by distributed sampling;

[0102] Calculate the time of particle scattering effect for This is to achieve iterative updates of particle states; among which, This is the delay time after the collision, and according to Determined by distribution sampling;

[0103] (7) The particle weights after the scattering effect are calculated using the following formula:

[0104]

[0105] In the formula The particle weights after the scattering effect occurs; It is the scattering cross section; This is the total cross section after the scattering effect occurs;

[0106] Update the particle's state parameters;

[0107] (8) Repeat steps (2) to (7) for iterative calculation until the number of calculated particles reaches the total number of particles; end the iterative calculation;

[0108] Finally, the simulation model needs to be set up to collect particle data during the simulation process. This requires defining the entire process from the initial emission of particles to each particle collision event and finally to the decay of the particles. Data collection and processing first requires judging the particle information, and then collecting and processing the particle information that needs to be statistically analyzed. When a particle is first emitted, the initialization behavior of the particle is edited in G4UserRunAction. Based on the core of Monte Carlo theory, the random number generator of the particle is edited to simulate the uncertainty in the physical process and ensure the accuracy of the simulation results. The judgment and collection of particle information for each collision event simulation needs to be edited and judged in (G4UserEventAction). When the particle entering the geometry of the detector card is an X-ray, the energy and quantity of the X-ray are recorded according to the X-ray data recorded by the Monte Carlo method. The particle termination information is edited and output in G4UserEventAction. The data processing adopts the root method to record the energy of each X-ray entering the detector card, accumulate the X-ray quantity, and then output it in the form of a histogram to obtain the distribution map of X-ray energy, thus completing the establishment of the entire simulation model.

[0109] S2.3. Based on the cross-validation scheme and the polynomial fitting scheme, establish the mathematical relationship between X-ray energy and quantity and image signal-to-noise ratio; specifically including the following steps:

[0110] The model built by Geant4 simulation software cannot simulate the subsequent circuit structure and cannot directly obtain the signal-to-noise ratio of the image. It can only obtain the energy and quantity of X-rays received by the detector card. In actual imaging, the detector card also receives X-ray energy and intensity and emits fluorescence. After photoelectric conversion and digital information transmission through the circuit structure, a digital image is finally formed. Therefore, to obtain the signal-to-noise ratio of the image, it is necessary to combine experiments and simulations. By using the X-ray energy obtained from the successfully verified X-ray imaging structure model and the image signal-to-noise ratio value obtained from experiments with the same parameters, a mathematical relationship between X-ray energy and quantity and image signal-to-noise ratio can be established. This allows for the evaluation of the differences in image quality formed by the optimized imaging structure parameters.

[0111] Under the same parameter conditions, several sets of data on average X-ray energy, number of X-rays and image signal-to-noise ratio were obtained through simulation.

[0112] Based on the obtained data, a five-fold cross-validation scheme was used to obtain the difference between the training error and the validation error in first-order, second-order, and third-order polynomials.

[0113] A polynomial fitting scheme is used for fitting, and the following formula is used as the final fitting function:

[0114]

[0115] In the formula Image signal-to-noise ratio; These are the first fitted parameters; The average energy of the X-rays; These are the second fitting parameters; The number of X-rays; The third fitting parameter; The fourth fitting parameter; This is the fifth fitting parameter.

[0116] S3. Based on the simulation model obtained in step S2, perform single-factor analysis using the controlled variable method to study the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, and determine the final influencing factors; specifically including the following steps:

[0117] Based on the simulation model obtained in step S2, a single-factor analysis was performed using the controlled variable method to study the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, and to determine the final influencing factors. The determined final influencing factors include tube voltage. tube current Target angle beryllium filter thickness , width of lead cylinder opening , lead-antimony collimator opening width , width of protective partition opening and the width of the lead plate opening under the belt .

[0118] S4. Based on the influencing factors obtained in step S3, with the maximum signal-to-noise ratio and minimum X-ray source power as optimization objectives, and with actual working conditions and process constraints as constraints, construct a parameter optimization model for the imaging structure of the XRT ore separator; specifically including the following steps:

[0119] The following formula is used as the sub-objective function for the signal-to-noise ratio. :

[0120]

[0121] In the formula It is a constant value; This is the first weight value; This is the second weight value; This is the third weight value; Let be the value of the i-th influencing factor; Let be the value of the j-th influencing factor; Corresponding to the tube voltage in sequence tube current Target angle beryllium filter thickness , width of lead cylinder opening , lead-antimony collimator opening width , width of protective partition opening and the width of the lead plate opening under the belt The value of ;

[0122] The following formula is used as the sub-target function for X-ray source power. :

[0123]

[0124] With the optimization objectives of maximizing the signal-to-noise ratio and minimizing the X-ray source power, the objective function is constructed as follows: ;

[0125] Using actual operating conditions and process conditions as constraints, the constraints are constructed as follows:

[0126]

[0127] In the formula This represents the minimum value of the i-th influencing factor, set according to actual working conditions and process conditions. This represents the maximum value of the i-th influencing factor, set according to actual working conditions and process conditions.

[0128] S5. Based on the genetic algorithm, solve the model constructed in step S4 to complete the parameter optimization of the target XRT ore sorting machine imaging structure; specifically including the following steps:

[0129] The solution to a multi-objective optimization problem is not a single solution, but a set of optimal solutions, which is a set of solutions where several objective functions reach a balance. These solutions form the Pareto curve. Solutions outside this set will compromise the values ​​of other objective functions while achieving the optimization of another objective function. Therefore, the selection of the optimal solution for a specific multi-objective optimization problem needs to be based on the actual engineering situation and selected on the Pareto curve.

[0130] There are various strategies for solving multi-objective optimization models, including common strategies such as weighted methods, ideal point methods, NSGA-II, and multi-objective particle swarm optimization. Both NSGA-II and multi-objective particle swarm optimization are effective methods for obtaining Pareto front solution sets in multi-objective optimization problems. However, NSGA-II can explore a wider search space, has stronger adaptability, can maintain the diversity of solutions, avoids convergence to local optima, and is suitable for finding uniform distributions of Pareto fronts. Therefore, NSGA-II is chosen.

[0131] Therefore, the NSGA-II algorithm was finally adopted, and the model constructed in step S4 was solved based on the obtained Pareto curve to complete the parameter optimization of the target XRT ore sorting machine imaging structure.

[0132] The method of the present invention will be further described below with reference to an embodiment:

[0133] Taking the X-ray imaging structure of the 1200-type XRT ore separator as the research object, a simulation model was first established in Geant4 simulation software.

[0134] The X-ray source used in the 1200-type XRT ore separator is a Gaomei integrated source. The electron emission energy and quantity are set according to the tube voltage and current of the source, and the electron emission direction is at a 25° angle to the target surface. The internal X-ray tube of the Gaomei source is made of cermet, which possesses excellent high-temperature resistance and thermal conductivity, capable of withstanding the high heat generated by X-rays and meeting heat dissipation requirements. The tube is kept in a vacuum environment to prevent electron attenuation during air propagation, thus avoiding energy waste, and to prevent filament burnout. X-ray tube parameters are shown in Table 1.

[0135]

[0136] The lead cylinder is an important component of the X-ray source. Its function is to shield and encapsulate the X-ray tube, prevent X-ray leakage in unnecessary directions, and ensure the mechanical stability and directionality of the X-ray source. The parameter values ​​of the geometric model of the lead cylinder are shown in Table 2.

[0137]

[0138] The opening width of the lead-antimony collimator is used to collimate X-rays, forming a parallel beam with strong directionality and low divergence, avoiding interference from stray scattered lines, controlling the direction of the X-ray beam, and ensuring that a clear image is formed in the detection area. The lead-antimony collimator is rectangular in shape, and its modeling parameter values ​​are shown in Table 3. The shape of the lead-antimony opening is trapezoidal, and its modeling parameter values ​​are shown in Table 4.

[0139]

[0140]

[0141] The protective box and protective partition are used to locally shield and isolate certain areas of the system to prevent X-rays from leaking into non-target areas. The protective box contains X-rays that penetrate through the lead-antimony collimator. The opening width of the protective partition can control the range and direction of the X-ray beam when the X-rays pass through. The modeling parameter values ​​of the protective box and protective partition are shown in Table 5.

[0142]

[0143] The lead plate under the belt provides mechanical support and radiation protection for the detection belt, which is used to transport ore. The size of its opening is used to reduce the scattering of X-rays, while allowing X-rays that penetrate the ore to enter the detector card. The modeling parameter values ​​of the lead plate under the belt are shown in Table 6.

[0144]

[0145] The detector card is used to collect X-rays that penetrate the ore and convert them into electrical signals. After subsequent processing, a grayscale image of the ore is formed. The part of the detector card that receives X-rays and emits fluorescence is the scintillator. Therefore, when modeling, only the scintillator model needs to be constructed. The modeling parameter values ​​of the scintillator are shown in Table 7.

[0146]

[0147] The above are geometric modeling diagrams of various parts of the X-ray imaging structure. These diagrams affect the entire X-ray propagation process and its interaction with matter, thus impacting the amount of information carried in the image, causing changes in the image signal-to-noise ratio, and consequently altering the sorting accuracy and efficiency of the XRT ore sorting machine.

[0148] The interaction between X-rays and matter mainly occurs in three ways: the photoelectric effect, the Compton effect, and the electron-electron pair effect. These interactions exhibit different behaviors at different energies and atomic numbers, such as... Figure 3 As shown: when energy is in At MeV, the photoelectric effect dominates; at energies of... At MeV, the Compton effect dominates; at energies of... At MeV, the electron-electron pair effect dominates. In the imaging structure of an XRT ore separator, the tube voltage of the X-ray source ranges from 100 to 180 kV, resulting in X-ray energies ranging from 100 to 180 keV. Therefore, the main physical processes occurring in the imaging structure are the X-ray photoelectric effect and the Compton effect.

[0149] Then, the particle motion process is simulated based on the Monte Carlo method.

[0150] Determine the factors affecting image signal-to-noise ratio:

[0151] The experiments were conducted, and the average X-ray energy and number of X-rays obtained from the simulation under the same parameter conditions, as well as the signal-to-noise ratio of the experimental images, are shown in Tables 8 and 9:

[0152]

[0153]

[0154] To determine the optimal mathematical function for fitting the average X-ray energy E, the number of X-rays N, and the signal-to-noise ratio (SNR), five-fold cross-validation was used to obtain the differences between the training error and the validation error in first-, second-, and third-order polynomials. This was implemented using MATLAB code, and the results are as follows: Figure 4 As shown, when using a second-order polynomial, the mean squared error of the training set is closest to the mean squared error of the validation set. Therefore, the second-order polynomial is selected as the fitting function.

[0155] Since the signal-to-noise ratio (SNR) is also zero when the average energy and quantity of X-rays are both zero, the SNR fitting function must pass through the origin. Therefore, this fitting function has no constant term and can be expressed as:

[0156]

[0157] Finally, the fitting results are as follows: Figure 5 As shown, the fitting function is given by the following equation, and the goodness of fit R² for this equation is 0.991.

[0158]

[0159] To verify the reasonableness of the fitting function, a residual distribution plot was drawn based on the fitting function, as shown below. Figure 6 As shown, the residuals are uniformly distributed on both sides of the zero line, without showing any specific structure or pattern, and all residual points are distributed within the range of 3δ, with no outliers, indicating that the residuals are randomly distributed and the fitting function is reasonable; therefore, the obtained fitting formula can be used to calculate the image signal-to-noise ratio using the average energy and quantity of X-rays, and to more intuitively judge the influence of each parameter of the imaging structure on the image signal-to-noise ratio.

[0160] Univariate analysis to determine the final influencing factors:

[0161] The impact of X-ray imaging structures on image signal-to-noise ratio (SNR) manifests in two aspects: firstly, during the X-ray generation stage, the X-ray source parameters directly determine the energy distribution of the generated X-rays; secondly, before the X-rays propagate to the detector card, the limitations and attenuation caused by various imaging structures on the energy distribution range of the X-rays. Factors affecting the X-ray source include tube voltage, tube current, and target angle. These three factors directly influence the energy intensity and emission direction of the generated X-rays, thus affecting the subsequent image SNR. Beryllium filters filter out X-rays with excessively low energy, enhancing the overall average energy of the X-rays, but they also affect the number of X-rays, impacting the final image signal-to-noise ratio. The signal-to-noise ratio (SNR) of X-ray imaging structures is studied by examining the opening widths of the lead cylinder, lead-antimony collimator, protective partition, and lead plate under the belt. These factors, on the one hand, limit the energy distribution of X-rays, reducing leakage and scattered radiation; on the other hand, excessive shielding can reduce X-ray utilization, wasting some X-rays and lowering the image's SNR. Therefore, eight factors—tube voltage, tube current, target angle, lead cylinder opening width, beryllium filter thickness, lead-antimony collimator opening width, protective partition opening width, and lead plate under the belt opening width—are selected to investigate the influence of X-ray imaging structures on image SNR. The controlled variable method is used to study the influence of each factor on the image SNR. The resulting image is shown below. Figure 7 As shown.

[0162] Figure 7 (a) shows that the signal-to-noise ratio increases significantly with increasing tube voltage. Increasing tube voltage leads to an increase in the upper energy limit of emitted electrons, a higher upper energy limit for X-rays generated by bremsstrahlung, and an increase in the number of characteristic X-rays generated by characteristic radiation, thus increasing the overall energy of X-rays. Simultaneously, increasing tube voltage leads to an increase in the kinetic energy of electrons emitted from the cathode in the X-ray source, resulting in more bremsstrahlung and characteristic radiation from electrons colliding with the anode tungsten target, thus increasing the number of X-rays generated.

[0163] Figure 7 (b) shows that the signal-to-noise ratio is significantly enhanced with the increase of tube current. The increase in tube current increases the number of electrons emitted from the X-ray tube cathode, resulting in more bremsstrahlung and characteristic radiation, and an increase in the number of continuous and characteristic X-rays. The increased total number of X-rays means more X-rays are received by the detector card, leading to a stronger signal.

[0164] Figure 7(c) indicates that within the 5-60° range, the signal-to-noise ratio (SNR) continuously increases with the target angle, peaking at 60°. Then, with further increases in the target angle, the SNR gradually decreases. When the target angle increases from 0° to 60°, the effective focal area of ​​the electron beam on the target gradually decreases, the effective focal area becomes more concentrated, the directionality of X-rays is enhanced, the number of X-rays reaching the detector increases, and the proportion of high-energy X-rays increases, improving signal strength and increasing the SNR. When the target angle increases from 60° to 90°, the effective bombardment area of ​​the electron beam on the target rapidly decreases, the collision depth between electrons and the target increases, and some energy is wasted on non-productive heat loss. Furthermore, the directionality of X-rays generated at larger angles deteriorates, resulting in the generation of more low-energy X-rays and scattered radiation, increasing the impact of background noise. Under these dual influences, the SNR gradually decreases. When the target angle approaches 90°, the geometry of the target hinders the release of some X-rays, further weakening the effective signal strength.

[0165] Figure 7 (d) indicates that the signal-to-noise ratio decreases continuously as the thickness of the beryllium filter decreases. When the thickness of the beryllium filter is greater than 2.6 mm, the signal-to-noise ratio no longer decreases with the increase of the thickness of the beryllium filter. This is because when the thickness of the beryllium filter increases to 2.6 mm, too many low-energy X-rays have been effectively filtered out. The source of the signal-to-noise ratio signal is mainly high-energy X-rays. Since the beryllium filter cannot effectively filter out high-energy X-rays, the signal strength no longer changes, and the signal-to-noise ratio remains unchanged.

[0166] Figure 7 (e) indicates that: in the range of 0-35mm, the signal-to-noise ratio (SNR) increases with the increase of the lead tube opening width; the SNR reaches its peak when the lead tube opening width increases to 35mm; after 35mm, further increases in opening width do not change the SNR. When the lead tube opening width is between 0-35mm, the increase in the lead tube opening width increases the number of X-rays passing through, increases the number of X-rays received by the detector card, enhances the signal strength, and increases the image SNR; when the lead tube opening width is between 35mm-45mm, the number of X-rays received by the detector card saturates, the SNR reaches its peak, and the optimal balance between signal and noise is achieved; further increases in opening width may increase the total number of X-rays passing through the lead tube, but the number of X-rays within the X-ray distribution range that the detector card can receive no longer changes, the number of received X-rays saturates, the signal strength stabilizes, and the SNR no longer changes.

[0167] Figure 7(f) indicates that when the aperture width of the lead-antimony collimator is 5-25 mm, the signal-to-noise ratio (SNR) increases with the increase of the aperture width. However, when the aperture width is 25-30 mm, the SNR no longer increases with the increase of the aperture width. Increasing the aperture width reduces the restriction and absorption of X-rays, leading to an increase in the number of X-rays within the X-ray distribution range that the detector card can receive, resulting in enhanced signal intensity and a higher SNR, peaking at 25 mm. When the aperture width is between 25-30 mm, the total number of X-rays passing through the collimator continues to increase, but within the range that the detector card can receive, the number of X-rays no longer changes, and the signal intensity remains constant. Therefore, the SNR tends to stabilize after the aperture width of the lead-antimony collimator exceeds 25 mm.

[0168] Figure 7 (g) indicates that when the opening width of the protective partition is between 1-4 mm, the signal-to-noise ratio (SNR) increases significantly with the increase of the opening width; when the opening width is between 4-9 mm, the SNR increases slowly with the increase of the opening width; and when the opening width is greater than 9 mm, the SNR no longer increases. When the opening width gradually increases from 1 mm to 4 mm, the number of X-rays passing through the protective partition increases significantly, and the number of X-rays within the receiving range of the detector card increases significantly. This directly increases the number of X-rays reaching the detector, significantly enhancing the signal strength and significantly increasing the SNR. When the opening width continues to increase to 4-9 mm, the total number of X-rays passing through the protective partition increases, but the number of X-rays within the receiving range of the detector card only increases slightly, approaching saturation. The signal strength increases slowly, and the SNR increases slightly to remain constant. When the opening width is greater than 9 mm, the total number of X-rays continues to increase, but within the receiving range of the detector card, the number of X-rays no longer changes, the signal strength does not increase, and the SNR remains unchanged.

[0169] Figure 7 (h) indicates that when the opening width of the lead plate under the belt is between 1 and 2.4 mm, the signal-to-noise ratio (SNR) increases with the increase of the opening width. When the opening width is greater than 2.4 mm, the SNR no longer changes. When the opening width increases from 1 mm to 2.4 mm, the number of X-rays passing through the lead plate under the belt increases, and the number of X-rays within the receiving range of the detector increases, thus improving the signal strength of the detector and increasing the SNR. When the opening width exceeds 2.4 mm, the number of X-rays passing through the lead plate under the belt and irradiating the detector no longer increases, and the signal strength no longer changes, indicating a stable SNR.

[0170] Model construction and solution:

[0171] The optimization of the X-ray imaging structure of the XRT ore sorting machine has two objectives. One is to improve the signal-to-noise ratio (SNR) of the X-ray imaging structure to enhance the sorting machine's recognition accuracy, with the maximum SNR as the optimization objective. The other is to improve the efficiency of the X-ray source by reducing its power, with the minimum X-ray source power as the optimization objective. When constructing the optimization objectives, tube voltage (U), tube current (I), target angle (α), beryllium filter thickness (d), lead cylinder opening width (k1), lead-antimony collimator opening width (k2), protective partition opening width (k3), and belt under-lead plate opening width (k4) are used as decision variables. The optimization objectives are to maximize the SNR and minimize the X-ray source power. The constraints are limited by actual operating conditions and process conditions, as shown in Table 10.

[0172]

[0173] The image signal-to-noise ratio (SNR) in XRT ore sorting determines the recognition accuracy, thus affecting the overall sorting performance of the machine. A low SNR makes it difficult for the machine to accurately identify the ore grade, impacting the sorting capacity. The objective function is constructed with the maximum SNR value as the goal. However, the NSGA-II algorithm optimizes by minimizing the objective function value. Therefore, the original maximum objective function value is transformed into its minimum value by adding a negative sign. The objective function is then:

[0174]

[0175] The X-ray source power of an XRT ore separator determines the emitted X-ray intensity. Excessive X-ray source power increases the risk of X-ray radiation leakage and the cost of X-ray protection, thus reducing the economic efficiency of the XRT ore separator. An objective function is constructed with the goal of minimizing the X-ray source power. ;

[0176] Based on the above objective function, structural parameters, and constraints, the multi-objective optimization model for the XRT ore sorting machine's X-ray imaging structure is as follows:

[0177]

[0178] The multi-objective optimization problem of X-ray imaging structures was solved using NSGA-II, and the algorithm was implemented using MATLAB. Pareto curves were plotted. The core parameters of the algorithm include population size, crossover probability, mutation probability, and number of iterations. The parameter settings are shown in Table 11.

[0179]

[0180] Solving the multi-objective optimization model of X-ray imaging structures and plotting the Pareto optimal solution set reveals different Pareto optimal solution set curves for different iteration numbers. When the iteration number is 100, the Pareto optimal solution set curve is stable, and the points within the optimal solution set are evenly distributed on the curve, as shown below. Figure 8 As shown.

[0181] In the optimization process of the NSGA-II algorithm, the optimal solutions on the Pareto front obtained after 100 iterations are the result of the coordination between the two optimization objectives after the optimization of X-ray imaging structural parameters. There is no distinction between superior and inferior solutions. Researchers can choose the most suitable solution from them according to the actual working conditions.

[0182] Based on experience, the X-ray source power of an XRT ore separator should not exceed 800W. Based on this, the maximum signal-to-noise ratio (SNR) of the image before optimization was 75.89, and the maximum SNR of the image after optimization was 94.26, an improvement of 24.21%. The specific parameters after optimization are shown in Table 12.

[0183]

[0184] At the same X-ray source power, the optimized results show a significant improvement compared to the unoptimized results, such as... Figure 9 As shown, this illustrates that when the X-ray source power is within the permissible range, optimizing the X-ray imaging structural parameters can improve the efficiency of X-ray source power utilization and significantly enhance the image signal-to-noise ratio. Engineers of XRT ore sorting machines can select the appropriate X-ray source model based on the actual working conditions and process requirements, reducing unnecessary power waste.

[0185] To ensure the credibility of the optimization scheme, a simulation model was set based on the parameters in Table 12. The optimized signal-to-noise ratio (SNR) value was obtained after simulation calculation. This value was compared with the actual optimization results, as shown in Table 13.

[0186]

[0187] As can be seen from Table 13, the relative error between the simulation values ​​and the optimized results is small, proving that the optimization scheme obtained by solving the multi-objective optimization model of X-ray imaging structure by the NSGA-II algorithm has high reliability.

[0188] like Figure 10The diagram shows the functional modules of the system of the present invention: The system disclosed in this invention for optimizing the parameters of the imaging structure of the XRT ore separator includes a structural parameter acquisition module, a simulation model construction module, an influencing factor determination module, an optimization module construction module, and a parameter optimization module; the structural parameter acquisition module, simulation model construction module, influencing factor determination module, optimization module construction module, and parameter optimization module are connected in series; the structural parameter acquisition module is used to acquire the structural parameter information of the target XRT ore separator and upload the data information to the simulation model construction module; the simulation model construction module is used to construct a simulation model of the target XRT ore separator in the simulation software based on the received data information, the acquired structural parameter information, and in combination with the Monte Carlo simulation scheme, Boltzmann equation, cross-validation scheme, and polynomial fitting scheme. The model is constructed by first obtaining the model and then uploading the data to the influencing factor determination module. The influencing factor determination module, based on the received data and the obtained simulation model, performs single-factor analysis using the controlled variable method to study the impact of each element in the X-ray imaging structure on the image signal-to-noise ratio (SNR) and determines the final influencing factors. This data is then uploaded to the optimization module. The optimization module, based on the received data and the obtained influencing factors, constructs a parameter optimization model of the XRT ore sorting machine imaging structure, with the optimization objectives of maximizing SNR and minimizing X-ray source power, and constrained by actual working conditions and process conditions. This model is then uploaded to the parameter optimization module. Finally, the parameter optimization module, based on the received data and the obtained influencing factors, solves the constructed model using a genetic algorithm to complete the parameter optimization of the target XRT ore sorting machine imaging structure.

Claims

1. A method of parameter optimization of an XRT ore sorter imaging structure, characterized in that The method comprises the following steps: S1. Obtain the structure parameter information of the target XRT ore sorter; S2. According to the structure parameter information obtained in step S1, combine the Monte Carlo simulation scheme, the Boltzmann equation, the cross-validation scheme and the polynomial fitting scheme, and construct a simulation model of the target XRT ore sorter in a simulation software; S3. According to the simulation model obtained in step S2, perform single-factor analysis based on the control variable method, study the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, and determine the final influencing factors; specifically comprising the following steps: Based on the simulation model obtained according to step S2, single factor analysis is carried out based on the control variable method, the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio is studied, and the final influencing factors are determined, including tube voltage , tube current , target angle , beryllium filter thickness , lead cylinder opening width , lead-antimony collimator opening width , protective baffle opening width and lead plate opening width under the belt ; S4. According to the influencing factors obtained in step S3, taking the maximum signal-to-noise ratio and the minimum X-ray source power as the optimization objectives, and taking the actual working conditions and process conditions as the constraints, a parameter optimization model of the XRT ore sorter imaging structure is constructed; specifically comprising the following steps: The following equation is used as a sub-objective function of the signal-to-noise ratio value : wherein is a constant value; is a first weight value; is a second weight value; is a third weight value; is a value of the ith influencing factor; is a value of the jth influencing factor; corresponding to tube voltage , tube current , target angle , beryllium filter thickness , lead cylinder opening width , lead-antimony collimator opening width , protective septum opening width , and under-belt lead plate opening width , respectively. The following equation is used as the X-ray source power sub-objective function : With the maximum signal-to-noise ratio and the minimum X-ray source power as the optimization objectives, a target function is constructed as ; Taking the actual working conditions and process conditions as the constraint conditions, the constraint conditions are constructed, which is expressed as: In the formula is the minimum value of the i-th influencing factor set according to the actual working condition and process condition; is the maximum value of the i-th influencing factor set according to the actual working condition and process condition; S5. Based on the genetic algorithm, the model constructed in step S4 is solved, and the parameter optimization of the imaging structure of the target XRT ore sorter is completed.

2. The method of parameter optimization of XRT ore sorter imaging structures according to claim 1, characterized in that The step S2 comprises the following steps: S2.

1. According to the structure parameter information obtained in step S1, a three-dimensional structure simulation model of the target XRT ore sorter is constructed in the simulation software; S2.

2. Based on the Monte Carlo simulation scheme and the Boltzmann equation, the propagation process and interaction of X-rays of the target XRT ore sorter are simulated; S2.

3. Based on the cross-validation scheme and the polynomial fitting scheme, a mathematical relationship between the X-ray energy and quantity and the image signal-to-noise ratio is established.

3. The method of parameter optimization of an XRT ore sorter imaging structure according to claim 2, characterized in that The step S2.1 specifically comprises the following steps: Taking the X-ray imaging structure of the target XRT ore sorter as the research object, taking the ray emitting point as the origin, using the Geant4 simulation software, an X-ray imaging structure simulation model of the target XRT ore sorter is established; The constructed X-ray imaging structure simulation model includes a lead cylinder, a lead antimony collimator, a protection box and a protection partition, a lead plate under the belt, and a scintillator of a detection card.

4. The method of parameter optimization of an XRT ore sorter imaging structure according to claim 3, characterized in that The step S2.2 specifically comprises the following steps: Based on the Monte Carlo simulation scheme, a set of state parameters is set for each particle, which is used to record all the motion states of the particle at present; the state parameters include the spatial coordinates , the velocity , the unit vector corresponding to the motion direction , the particle weight value and the time ; Ignoring the interaction between particles, the Boltzmann equation is used to record the operation process of the particles, which is expressed as: In the formula The particle distribution function; The spatial gradient of the particle distribution function; For the particle to be in And the speed magnitude is The total cross section at that time; The magnitude of the particle's velocity before the collision; The direction of motion of the particle before the collision; The collision time; Let be the scattering function, used to represent a velocity at position r. And the direction is The particles, after the collision time Then, it is scattered to the velocity And the direction is The conditional probability within a unit velocity and a unit solid angle; For the particle to be in And the speed magnitude is The total cross section of particle scattering at that time; As the source term, it is used to represent time. In position At a point, within a unit time, unit volume, or unit solid angle, the energy produced has velocity. and direction The number of particles; Based on the Monte Carlo simulation scheme, the process of simulating the particle output problem is simulated, comprising the following steps: (1) The source term distribution is normalized The initial state of the particles is determined by sampling The initial weight of the particles is calculated using the following equation is ; where is the total volume of the computational domain; (2) Calculate the particle collision coordinates: the following formula is used to calculate the distance of the particle movement : wherein represents the total cross section of the particles in the state when; is the integral variable; is the particle transport free path; The particle collision coordinates are calculated To ; (3) Determine whether the particle is in the simulation region: set the running region of the particle as ; when the particle leaves the region , the simulation stops; if the particle is in the region , the simulation continues; (4) Calculate particle collision time For ; (5) The cumulative value of the target physical quantity is updated by using the following formula: wherein is the current accumulated target physical quantity; is the previous accumulated target physical quantity; is a set global scale factor; is a weight function for the particle; (6) The velocity and corresponding unit direction vector of the particle after collision are calculated, which are used for iterative updating of the particle; The scattering effect is generated after the particle collision, and the scattering effect is generated by Distribution sampling confirmation; Time at which the particle undergoes a scattering effect is calculated To Implement an iterative update of the state of the particle; wherein, is the delay time after the collision, and is determined according to the distribution sampling; (7) The particle weight after the particle scattering effect is calculated by using the following formula: wherein is the weight of the particle after the scattering effect has occurred; is the scattering cross section; is the total cross section after the scattering effect has occurred; The state parameters of the particle are updated; (8) Repeat steps (2) to (7) to perform iterative calculation until the number of calculated particles reaches the total number of particles; end the iterative calculation.

5. The method of parameter optimization of an XRT ore sorter imaging structure according to claim 4, characterized in that The step S2.3 specifically comprises the following steps: Under the same parameter conditions, a plurality of sets of data of the X-ray average energy, the X-ray quantity and the image signal-to-noise ratio are simulated; According to the obtained plurality of sets of data, the difference values of the training error and the verification error in the first-order, second-order and third-order polynomials are obtained by using the five-fold cross-validation scheme; The polynomial fitting scheme is adopted for fitting, and the following formula is used as the final fitting function formula: wherein is the image signal-to-noise ratio; is the first fitting parameter; is the X-ray mean energy; is the second fitting parameter; is the X-ray number; is the third fitting parameter; is the fourth fitting parameter; is the fifth fitting parameter.

6. The method of parameter optimization of an XRT ore sorter imaging structure according to claim 5, characterized in that The step S5 specifically includes the following steps: The NSGA-II algorithm is adopted, and the model constructed in the step S4 is solved according to the obtained Pareto curve, so that the parameter optimization of the imaging structure of the target XRT ore separator is completed.

7. A system for implementing the method of optimizing parameters of the imaging structure of the XRT ore sorter of one of claims 1 to 6, characterized in that The structure parameter acquisition module, the simulation model construction module, the influence factor determination module, the optimization module construction module and the parameter optimization module are connected in sequence; the structure parameter acquisition module is used for acquiring the structure parameter information of the target XRT ore separator and uploading the data information to the simulation model construction module; the simulation model construction module is used for constructing the simulation model of the target XRT ore separator in the simulation software according to the received data information, the obtained structure parameter information, the Monte Carlo simulation scheme, the Boltzmann equation, the cross-validation scheme and the polynomial fitting scheme, and uploading the data information to the influence factor determination module; The influence factor determination module is used for performing single factor analysis based on the control variable method according to the received data information and the obtained simulation model, researching the influence of each element in the X-ray imaging structure on the image signal-to-noise ratio, determining the final influence factor, and uploading the data information to the optimization module construction module; The optimization module construction module is used for constructing the parameter optimization model of the imaging structure of the XRT ore separator according to the received data information and the obtained influence factor, taking the maximum signal-to-noise ratio and the minimum X-ray source power as the optimization objectives, and taking the actual working condition and the process condition as the constraints, and uploading the data information to the parameter optimization module; The parameter optimization module is used for solving the constructed model based on the genetic algorithm according to the received data information, so that the parameter optimization of the imaging structure of the target XRT ore separator is completed. ​

Images