An ore impurity sorting system, method and storage medium

Abstract

The application discloses an ore impurity sorting system, method and storage medium, relates to the intelligentization technical field of mineral processing, and solves the technical problems that the prior art lacks consideration of the influence of the environment and characteristic quality on the sorting result, the sorting result is poor, the sorting scheme is relatively single, cannot adapt to the change of working condition data, the effect of the sorting scheme is poor, and the accuracy and efficiency of the impurity sorting system are low; decision data is generated according to ore data and environment data; the decision data is input into an impurity prediction joint model to obtain impurity categories and impurity contents and generate a sorting scheme, the decision weights of multi-modal data are dynamically adjusted along with the change of the environment and multi-modal data quality, the model pays more attention to data with high weights when discriminating, the accuracy of impurity discrimination is improved, the sorting scheme is adjusted in real time according to working condition data, the sorting scheme meets enterprise targets, and the accuracy and efficiency of the impurity sorting system are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of mineral processing intelligence, and particularly relates to an ore impurity sorting system and method and a storage medium. BACKGROUND

[0002] Ore impurity sorting is a technical process that, by means of physical, chemical or physical-chemical means, according to the significant differences in density, magnetism and other properties between target minerals and impurities in ores, the impurity components in ores are accurately and effectively separated through a series of carefully designed process flows. This process is crucial for improving the grade and purity of ores, and can provide more high-quality and pure raw materials for subsequent smelting or processing links, thereby significantly improving production efficiency and product quality, reducing production cost, and playing an indispensable important role in the field of mining production.

[0003] In the prior art, when ore impurity sorting is performed, a plurality of features are maintained in a fixed form for sorting, and the influence of the environment and feature quality on the sorting result is not considered, which results in poor sorting results, and the sorting scheme is relatively single and cannot adapt to changes in working condition data, which results in poor effects of the sorting scheme and further leads to low accuracy and efficiency of the impurity sorting system. Therefore, the ore impurity sorting system still needs to be further improved. SUMMARY

[0004] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application provides an ore impurity sorting system, method and storage medium, which are used to solve the technical problem that the prior art lacks consideration of the influence of the environment and feature quality on the sorting result, which results in poor sorting results, and the sorting scheme is relatively single and cannot adapt to changes in working condition data, which results in poor effects of the sorting scheme and further leads to low accuracy and efficiency of the impurity sorting system.

[0005] To achieve the above-mentioned purpose, a first aspect of the present application provides an ore impurity sorting system, comprising: a data acquisition module, a data analysis module and a database; the data acquisition module and the data analysis module are electrically and / or communicatively connected; the database is electrically and / or communicatively connected with the data acquisition module and the data analysis module, respectively;

[0006] The data acquisition module: obtains environment data, ore data and target data through a data acquisition device;

[0007] The data analysis module: generates decision data according to the ore data and the environment data; inputs the decision data into an impurity prediction joint model to obtain an impurity category and an impurity content; the impurity prediction joint model is constructed by an artificial intelligence model; generates a sorting scheme according to the impurity category, the impurity content and the target data;

[0008] The database is used to store the data of each module and store the historical data required for training the model.

[0009] Through the above steps, the application can dynamically and flexibly adjust the decision weight corresponding to the multi-modal data according to the environmental conditions and the change of the quality of the multi-modal data. In this way, when the model discriminates, it will focus more on the data with higher weight, thereby significantly improving the accuracy of impurity discrimination. At the same time, the sorting scheme is optimized and adjusted in real time according to the working condition data, ensuring that the sorting scheme can highly match the actual target of the enterprise, and finally effectively improving the accuracy and operating efficiency of the impurity sorting system.

[0010] Further, the decision data is generated according to the ore data and the environmental data, comprising:

[0011] Obtaining ore data and environmental data; the ore data includes ore parameters and multi-modal data; the environmental data includes temperature, humidity, dust concentration and vibration intensity;

[0012] Generating a modal quality score according to the multi-modal data;

[0013] Generating an environmental influence coefficient according to the environmental data;

[0014] A weight coefficient function QZF(MZP, HYX) is constructed through a nonlinear relationship between the modal quality score and the environmental influence coefficient and the modal weight coefficient; wherein MZP represents the modal quality score, and HYX represents the environmental influence coefficient;

[0015] The modal quality score and the environmental influence coefficient corresponding to the multi-modal data are substituted into the weight coefficient function to calculate the weight coefficient corresponding to the multi-modal data;

[0016] The weight coefficient and the corresponding multi-modal data are combined to obtain the decision data.

[0017] Further, the modal quality score is generated according to the multi-modal data, comprising:

[0018] Obtaining multi-modal data; the multi-modal data includes X-ray fluorescence spectrum XRF, laser-induced breakdown spectroscopy LIBS, X-ray diffraction XRD and hyperspectral imaging HSI;

[0019] Extracting a plurality of index parameters ZC corresponding to the multi-modal data j ;

[0020] A score function PF(ZC is constructed according to the nonlinear relationship between the plurality of index parameters and the modal quality score corresponding thereto j ); wherein j represents the number of the plurality of index parameters corresponding to the modal data;

[0021] corresponding to the multi-modal data is obtained by substituting the corresponding several index parameters in the multi-modal data into the corresponding scoring function;

[0022] The normalized modal quality score is obtained by maximum-minimization operation on the several modal quality scores.

[0023] Further, the environmental data generates an environmental impact coefficient, including:

[0024] Obtain environmental data; the environmental data refers to several modal impact data MYS i ;

[0025] The impact function YF(MYS i ) is constructed through the nonlinear relationship between the several modal impact data and the environmental impact coefficient of the corresponding multi-modal data; wherein i represents the number of the several modal impact data corresponding to the multi-modal data;

[0026] The environmental impact coefficient corresponding to the multi-modal data is obtained by substituting the corresponding several modal impact data of the multi-modal data into the corresponding impact function.

[0027] Further, the impurity prediction joint model is constructed by an artificial intelligence model, including:

[0028] Obtain several historical decision data and their corresponding historical impurity categories and historical impurity contents;

[0029] The several historical decision data and their corresponding historical impurity categories and historical impurity contents are divided into a training set, a test set and a validation set;

[0030] Select an artificial intelligence model as a base model;

[0031] The base model is trained through the training set, and the learning rate or other hyperparameters are adjusted on the validation set to obtain a pre-trained model;

[0032] The pre-trained model is verified on the test set, and finally an impurity prediction joint model with input of decision data and output of impurity category and impurity content is obtained.

[0033] The impurity prediction joint model is trained in several corresponding historical data, which can determine the impurity category and calculate the impurity content of the ore through the model, greatly improve the accuracy and speed of identification, and improve the efficiency of the ore impurity sorting system.

[0034] Further, the sorting scheme is generated according to the impurity category and the impurity content and the target data, including:

[0035] Obtain impurity categories, impurity contents, target data and working condition data;

[0036] Construct a plurality of target sub-functions MZF according to the target data;

[0037] Generate an optimization function based on the plurality of target sub-functions MZF ; the optimization function satisfies the following formula:

[0038] ; wherein k represents the number corresponding to the plurality of target sub-functions, represents the optimization weight corresponding to the plurality of target sub-functions; the optimization weight is generated by the working condition data;

[0039] Solve the optimization function by a parameter optimization algorithm to obtain an optimal parameter value, and integrate the optimal parameter value to obtain a sorting scheme.

[0040] The application constructs an optimization function with the aid of target data, wherein each target parameter corresponds to a dedicated target sub-function, so that the system can comprehensively consider multiple target parameters in the process of generating a sorting scheme, thereby generating a sorting scheme that fully meets the needs of the enterprise; Furthermore, the application can flexibly adjust the optimization weight corresponding to the target sub-function according to the real-time changes of the working condition data; in this way, the sorting scheme is no longer single and fixed, but has rich diversity, which can accurately adapt to the actual target of enterprise production, thereby significantly improving the operation efficiency of the sorting system.

[0041] Further, the plurality of target sub-functions MZF are constructed according to the target data, comprising:

[0042] Obtain target data; the target data includes purity, recovery rate, cost and energy consumption;

[0043] Construct a purity target sub-function CMZF through a nonlinear relationship between purity and purity optimization;

[0044] Construct a recovery rate target sub-function HMZF through a nonlinear relationship between recovery rate and recovery rate optimization;

[0045] Construct a cost target sub-function CBMZF through a nonlinear relationship between cost and cost optimization;

[0046] Construct an energy consumption target sub-function NMZF through a nonlinear relationship between energy consumption and energy consumption optimization;

[0047] The target sub-function includes a purity target sub-function CMZF, a recovery rate target sub-function HMZF, a cost target sub-function CBMZF and an energy consumption target sub-function NMZF.

[0048] Further, the optimization weight is generated through the working condition data, comprising:

[0049] Obtaining working condition data; the working condition data comprises working condition parameter values GC corresponding to a plurality of target sub-functions;

[0050] According to the nonlinear relationship between the plurality of working condition parameter values and the priority score, a priority score function YPF(GC) is constructed;

[0051] A plurality of target sub-functions corresponding to the priority YJ is calculated by substituting the plurality of working condition parameter values into the priority score function YPF(GC);

[0052] According to the nonlinear relationship between the priority and the optimization weight, an optimization weight calculation function YQF(YJ) is constructed;

[0053] The plurality of priorities is substituted into the optimization weight calculation function YQF(YJ) to calculate the optimization weight MQ corresponding to the plurality of target sub-functions.

[0054] The second aspect of the present application provides a method for sorting ore impurities, comprising:

[0055] Obtaining environmental data, ore data and target data;

[0056] Generating decision data according to the ore data and the environmental data;

[0057] Inputting the decision data into the impurity prediction joint model to obtain the impurity category and the impurity content; the impurity prediction joint model is constructed by an artificial intelligence model;

[0058] Generating a sorting scheme according to the impurity category, the impurity content and the target data.

[0059] Another aspect of the present application provides a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program is executed by a processor to realize the ore impurity sorting system according to the first aspect of the present application.

[0060] Compared with the prior art, the beneficial effects of the present application are:

[0061] 1、The present application generates decision data according to ore data and environmental data; inputs the decision data into the impurity prediction joint model to obtain the impurity category and the impurity content; generates a sorting scheme according to the impurity category, the impurity content and the target data, dynamically adjusts the decision weight corresponding to the multi-modal data with the change of the environment and the multi-modal data quality, so that the model pays more attention to the data with high weight when making a judgment, improves the accuracy of impurity judgment, and simultaneously adjusts the sorting scheme in real time according to the working condition data, so that the sorting scheme is more in line with the enterprise target, and the accuracy and efficiency of the impurity sorting system are improved.

[0062] 2、The application carries out double analysis on multi-modal data affecting the results when determining impurity categories and calculating impurity contents, not only considers the data quality of multi-modal data itself, but also considers the inhibitory effect of environment on multi-modal data, both of which jointly determine the decision degree of multi-modal data when determining impurity categories and calculating impurity contents, thereby improving the accuracy of impurity category determination and impurity content calculation.

[0063] 3、The application uses a plurality of data indexes corresponding to multi-modal data to evaluate the quality of modal data, simultaneously uses modal influence data having inhibitory effect on multi-modal data to evaluate the influence degree of environment on multi-modal data, and carries out targeted analysis on multi-modal data, so that the quantification of data quality and environmental influence of multi-modal data is more accurate, and more accurate data support is provided for subsequent allocation of corresponding weight coefficients of multi-modal data.

[0064] 4、The application constructs an optimization function through target data, each target parameter corresponds to a target sub-function, which can generate a sorting scheme that meets the enterprise demand in multiple target parameters when generating a sorting scheme, and simultaneously adjusts the optimization weight corresponding to the target sub-function according to the working condition data, so that the sorting scheme has diversity, meets the production target of the enterprise, and improves the efficiency of the sorting system. BRIEF DESCRIPTION OF DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0066] Figure 1 It is a schematic diagram of an ore impurity sorting system of the present application;

[0067] Figure 2 It is a decision data generation flowchart of the present application;

[0068] Figure 3 It is a flowchart of an ore impurity sorting method of the present application. DETAILED DESCRIPTION

[0069] The technical solutions of the present application will be described in detail below in conjunction with the embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0070] Referring to Figure 1 The first aspect of the present application provides a mineral impurity sorting system, comprising: a data acquisition module, a data analysis module and a database; the data acquisition module and the data analysis module are electrically and / or communicatively connected; the database is electrically and / or communicatively connected with the data acquisition module and the data analysis module, respectively;

[0071] The data acquisition module: obtains environmental data, mineral data and target data through a data acquisition device; the data acquisition device includes a plurality of sensors, such as temperature / humidity sensors; the environmental data refers to the surrounding environmental data during mineral sorting; the mineral data refers to the mineral data that needs to be sorted; the target data refers to the data that needs to be sorted towards the target during mineral sorting;

[0072] The data analysis module: generates decision data according to the mineral data and the environmental data; the decision data refers to the decisive data for determining the impurity category and impurity content of the mineral; inputs the decision data into an impurity prediction joint model to obtain the impurity category and impurity content, which refers to the impurity parameters in the mineral; the impurity prediction joint model is constructed through an artificial intelligence model; generates a sorting scheme according to the impurity category and impurity content and the target data; the sorting scheme refers to the optimal sorting scheme under the current state;

[0073] The database is used to store the data of each module and store the historical data required for training the model.

[0074] Referring to Figure 2 In the present embodiment, the decision data is generated according to the mineral data and the environmental data, which includes:

[0075] Obtain the mineral data and the environmental data; the mineral data includes mineral parameters and multi-modal data; the environmental data includes temperature, humidity, dust concentration and vibration intensity, etc.;

[0076] Generate a modal quality score according to the multi-modal data;

[0077] Generate an environmental impact coefficient according to the environmental data;

[0078] Construct a weight coefficient function QZF(MZP, HYX) through the nonlinear relationship between the modal quality score and the environmental impact coefficient and the modal weight coefficient; wherein, MZP represents the modal quality score, and HYX represents the environmental impact coefficient;

[0079] The weight coefficient function QZF(MZP, HYX) satisfies the following formula:

[0080] ; wherein, n represents the number corresponding to the multi-modal data; and respectively represent a data quality sensitivity parameter and an environmental influence suppression parameter, and and are both greater than 0, and specific values are set according to experience, and in the embodiment, the and are respectively set to 1.5 and 0.8, the advantages of high-quality modal data are amplified through a power function, when > 1, the quality difference is nonlinearly amplified; when 0 < 1, the difference is smoothed; an exponential decay function is used to reduce the weight of a high environmental impact coefficient, and the weight decays exponentially with each unit increase in environmental impact, and the decay rate is controlled by , and by adjusting and , the sensitivity of quality and environment can be flexibly controlled; the higher the modal quality score corresponding to the multi-modal data, the more easily the effective features of the multi-modal data are expressed, which is more beneficial for subsequent impurity discrimination, and the greater the environmental impact coefficient corresponding to the multi-modal data, the greater the inhibitory effect on the multi-modal data, therefore, the weight coefficient increases with the increase of the modal quality score and the decrease of the environmental impact coefficient;

[0081] The modal quality score and the environmental impact coefficient corresponding to the multi-modal data are substituted into the weight coefficient function to calculate the weight coefficient corresponding to the multi-modal data.

[0082] The weight coefficient and the corresponding multi-modal data are combined to obtain decision data.

[0083] In the implementation, when carrying out impurity category determination and impurity content calculation, multi-modal data that will affect the results are subjected to double-depth analysis; on the one hand, the data quality possessed by the multi-modal data itself is fully considered; on the other hand, the inhibitory effect of environmental factors on the multi-modal data is also taken into account, and the two factors interact and jointly determine the key degree of the multi-modal data in the process of impurity category determination and impurity content calculation, thereby improving the accuracy of impurity category determination and the accuracy of impurity content calculation.

[0084] In the implementation, the modal quality score is generated according to the multi-modal data, including:

[0085] The multi-modal data is obtained; the multi-modal data includes X-ray fluorescence spectrum XRF, laser-induced breakdown spectroscopy LIBS, X-ray diffraction XRD, and hyperspectral imaging HSI;

[0086] A plurality of index parameters ZC j corresponding to the multi-modal data are extracted;

[0087] According to the nonlinear relationship between the several index parameters and the corresponding modal quality scores, a scoring function PF(ZC j ) is constructed; wherein j represents the number of several index parameters corresponding to the modal data;

[0088] The embodiment extracts several index parameters corresponding to XRF, including signal-to-noise ratio SNR, which is used to measure signal purity; peak area ratio FMB, which is expressed as the ratio of target element peak area to reference peak area, and the reference peak area is set according to experience; and half-peak width ratio BFB, which is expressed as the ratio of target half-peak width to reference half-peak width, and the reference half-peak width is set according to experience;

[0089] The scoring function corresponding to XRF satisfies the following formula:

[0090] ; wherein is the reference signal-to-noise ratio, and the specific value is set according to experience;

[0091] The embodiment extracts the line intensity ratio PQB, the line broadening matching degree PZPD and the plasma temperature stability DWX corresponding to LIBS;

[0092] The scoring function corresponding to LIBS satisfies the following formula:

[0093] ; three indicators are coupled in the form of product, and the corresponding quality score is compressed to the interval [0, 1] by using the Sigmoid function;

[0094] The embodiment extracts the peak background ratio FBB, the angle matching degree JPD and the peak sharpness FRD corresponding to XRD;

[0095] The scoring function corresponding to XRD satisfies the following formula:

[0096] ;

[0097] The embodiment extracts the average signal-to-noise ratio PSNR corresponding to HSI, which is expressed as the average signal-to-noise ratio of all wavebands; dynamic range DF, which is expressed as the ratio of actual dynamic range to theoretical maximum value, and the theoretical maximum value is set according to experience; and waveband offset degree BPD, which is expressed as the waveband center wavelength offset;

[0098] The scoring function corresponding to HSI satisfies the following formula:

[0099] ; wherein is the maximum theoretical signal-to-noise ratio of the device, is the allowed waveband offset standard deviation, and the specific value is set according to experience, and in the embodiment, the is set to 1nm;

[0100] The corresponding several index parameters in the multi-modal data are substituted into the corresponding scoring function to obtain the modal quality score corresponding to the multi-modal data;

[0101] The normalized modal quality score is obtained by maximum-minimum operation on the several modal quality scores.

[0102] The environmental data in the embodiment generates an environmental impact coefficient, including:

[0103] Obtain environmental data; the environmental data refers to several modal influence data MYS i that affect the multi-modal data;

[0104] An influence function YF(MYS i ) is constructed through the nonlinear relationship between the several modal influence data and the environmental impact coefficient of the corresponding multi-modal data; wherein i represents the number of the several modal influence data corresponding to the multi-modal data;

[0105] The several modal influence data corresponding to XRF are extracted, including temperature WD, humidity SD and electromagnetic interference DG;

[0106] The influence function corresponding to XRF satisfies the following formula:

[0107] ; wherein, represents the temperature tolerance range, BW represents the standard temperature, and the specific value is set according to experience, and in the embodiment, BW is set to 25 , is set to 5 ; BS represents a safe humidity threshold, represents a humidity penalty coefficient, represents electromagnetic interference sensitivity, and the specific value is set according to experience, and in the embodiment, BS, and are respectively set to 40%, 0.02 and 0.1;

[0108] The several modal influence data corresponding to LIBS are extracted, including vibration intensity ZQ, humidity SD and dust concentration FN;

[0109] The influence function corresponding to LIBS satisfies the following formula:

[0110] ; wherein, BS represents a safe humidity threshold, represents a humidity attenuation coefficient, represents a vibration sensitivity coefficient, represents a dust penalty coefficient, and the specific value is set according to experience, and in the embodiment, BS, 、 and are respectively set to 30%, 0.05 / %, 0.1 and 0.01; the vibration influence is nonlinearly amplified, and the humidity and dust concentration suppression effects are significant;

[0111] The embodiment extracts a plurality of modal influence data corresponding to XRD, including temperature fluctuation , vibration intensity ZQ and air flow KL;

[0112] The influence function corresponding to XRD satisfies the following formula:

[0113] ; wherein, MW represents the maximum allowable temperature difference, represents the vibration sensitivity, represents the air flow attenuation coefficient, and the specific value is set according to experience. In the embodiment, MW, and are respectively set to 2 , 0.2 and 0.05; the temperature term uses a cosine function, which slowly decays within the allowable range and sharply rises when exceeding the threshold;

[0114] The embodiment extracts a plurality of modal influence data corresponding to HSI, including light intensity GQ, temperature WD and vibration intensity ZQ;

[0115] The influence function corresponding to HSI satisfies the following formula:

[0116] ; wherein, represents the light penalty coefficient, BG represents the standard light intensity, and BW represents the standard temperature, represents the temperature sensitivity, represents the vibration penalty coefficient, and the specific value is set according to experience. In the embodiment, BG and BW are respectively set to 500 lux and 20 , and , and are respectively set to 0.005, 0.03 and 0.1; the light intensity and vibration intensity have linear suppression on it, and the temperature deviation has exponential increase;

[0117] The plurality of modal influence data corresponding to the plurality of modal influence data is substituted into the corresponding influence function to obtain the environmental influence coefficient corresponding to the plurality of modal data.

[0118] The embodiment applies a plurality of key data indexes corresponding to multi-modal data to accurately evaluate the quality of the modal data; meanwhile, modal influence data capable of producing an inhibitory effect on the multi-modal data is introduced to scientifically evaluate the influence degree of the environment on each modal data; through this double analysis method, the multi-modal data is analyzed in a targeted manner, and the data quality quantification and environmental influence quantification of the multi-modal data are more accurate, which provides more reliable and accurate data basis for subsequent reasonable allocation of corresponding weight coefficients of the multi-modal data, and helps to improve the scientificity and effectiveness of the entire data processing and analysis process.

[0119] The impurity prediction joint model in the embodiment is constructed by an artificial intelligence model, including:

[0120] A plurality of historical decision data, historical impurity categories and historical impurity contents corresponding thereto are obtained;

[0121] The plurality of historical decision data, historical impurity categories and historical impurity contents corresponding thereto are divided into a training set, a test set and a validation set; the ratio between the training set, the validation set and the test set is set to 7:2:1;

[0122] An artificial intelligence model is selected as a base model; the artificial intelligence model includes a BP model and the like;

[0123] The base model is trained through the training set, and a pre-training model is obtained by adjusting a learning rate or other hyperparameters on the validation set;

[0124] The pre-training model is verified on the test set, and finally an impurity prediction joint model with the input of decision data and the output of impurity categories and impurity contents is obtained.

[0125] The generation of a sorting scheme according to the impurity categories and the impurity contents and target data in the embodiment includes:

[0126] The impurity categories, impurity contents, target data and working condition data are obtained;

[0127] A plurality of target sub-functions MZF are constructed according to the target data;

[0128] An optimization function is generated based on the plurality of target sub-functions MZF ; the optimization function satisfies the following formula:

[0129] ; wherein k represents the number corresponding to the plurality of target sub-functions, represents the optimization weight corresponding to the plurality of target sub-functions; the optimization weight is generated through the working condition data;

[0130] The optimal parameter value is obtained by solving the optimization function through a parameter optimization algorithm, and the sorting scheme is obtained by integrating the optimal parameter value, and the parameter optimization algorithm includes NSGA-II algorithm and the like.

[0131] In the embodiment, a plurality of target sub-functions MZF are constructed according to target data, including:

[0132] Target data is obtained, and the target data includes purity, recovery rate, cost and energy consumption;

[0133] A purity target sub-function CMZF is constructed through a nonlinear relationship between the purity and the purity optimization, and the purity target sub-function satisfies the following formula:

[0134] ; wherein h represents the number corresponding to the impurity category, represents the content of the hth impurity, represents the total content of impurities;

[0135] A recovery rate target sub-function HMZF is constructed through a nonlinear relationship between the recovery rate and the recovery rate optimization, and the recovery rate target sub-function satisfies the following formula:

[0136] ; wherein, represents the feed amount, i.e., the mass of the ore entering the system, represents the mass of the final product;

[0137] A cost target sub-function CBMZF is constructed through a nonlinear relationship between the cost and the cost optimization, and the cost target sub-function satisfies the following formula:

[0138] ; wherein, represents the unit price of the rth reagent, represents the consumption of the rth reagent;

[0139] An energy consumption target sub-function NMZF is constructed through a nonlinear relationship between the energy consumption and the energy consumption optimization, and the energy consumption target sub-function satisfies the following formula:

[0140] ; DGL represents the motor power, and YS represents the motor running time;

[0141] The target sub-function includes a purity target sub-function CMZF, a recovery rate target sub-function HMZF, a cost target sub-function CBMZF and an energy consumption target sub-function NMZF.

[0142] The optimization weight in the embodiment is generated through working condition data, including:

[0143] Acquire operating condition data; the operating condition data includes the operating condition parameter values ​​(GC) corresponding to several objective sub-functions;

[0144] A priority scoring function YPF(GC) is constructed based on the nonlinear relationship between several operating condition parameter values ​​and priority scores; the priority scoring function YPF(GC) satisfies the following formula:

[0145] ;in, This is represented by the importance level corresponding to the kth objective sub-function, and the importance level is preset based on experience. This is represented as the baseline value of the working condition parameters corresponding to the kth objective sub-function, which is set based on experience; This is represented as the threshold value of the working condition parameter corresponding to the k-th objective sub-function, which is set based on experience; Expressed as response sensitivity, >1, the specific value is set based on experience, and in this embodiment it will be Set to 1.6;

[0146] Substituting the current operating condition parameter values ​​into the priority scoring function YPF(GC) yields the priority YJ corresponding to several target sub-functions;

[0147] The optimization weight calculation function YQF(YJ) is constructed based on the nonlinear relationship between priority and optimization weight; the optimization weight calculation function YQF(YJ) satisfies the following formula:

[0148] ;in, This is represented as the priority of the k-th sub-objective function; Expressed as steepness coefficient, >1, the specific value is set based on experience, and in this embodiment it will be Set to 2; In this embodiment, the operating condition parameter value that affects the weight coefficient of the cost subfunction is the drug inventory. If the drug inventory is low, then when optimizing, it is necessary to reduce its corresponding weight coefficient in order to achieve the goal of saving costs.

[0149] Substituting several priorities into the optimization weight calculation function YQF(YJ) yields the optimization weights MQ corresponding to several objective sub-functions.

[0150] This embodiment analyzes the operating condition parameter values ​​that affect each target sub-function to obtain the priority of the current target sub-function. Target sub-functions with higher priorities are given higher optimization weights, so that when generating sorting schemes, the focus is on optimizing target sub-functions with higher optimization weights. This makes the sorting schemes more comprehensive and can be dynamically adjusted according to operating condition data, thereby improving the efficiency of the sorting system.

[0151] Please seeFigure 3 The second aspect embodiment of the present application provides a mineral impurity sorting method, comprising:

[0152] obtaining environment data, mineral data and target data;

[0153] generating decision data according to the mineral data and the environment data;

[0154] inputting the decision data into an impurity prediction joint model to obtain an impurity category and an impurity content; the impurity prediction joint model is constructed by an artificial intelligence model;

[0155] generating a sorting scheme according to the impurity category, the impurity content and the target data.

[0156] Another aspect embodiment of the present application provides a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement a mineral impurity sorting system according to the first aspect of the present application.

[0157] Some data in the above formula is calculated by removing the dimension, the formula is obtained by software simulation of a large amount of collected data to obtain a formula closest to the real situation; the preset parameters and the preset threshold in the formula are set by a person skilled in the art according to the actual situation or obtained by a large amount of data simulation.

[0158] The working principle of the present application is as follows: environment data, mineral data and target data are obtained; decision data is generated according to the mineral data and the environment data; an impurity category and an impurity content are obtained by inputting the decision data into an impurity prediction joint model; a sorting scheme is generated according to the impurity category, the impurity content and the target data, the decision weights of the multi-modal data are dynamically adjusted with the change of the environment and the multi-modal data quality, so that the model pays more attention to the data with high weights during judgment, the accuracy of impurity judgment is improved, the sorting scheme is adjusted in real time according to the working condition data, so that the sorting scheme is more in line with the enterprise target, the accuracy and efficiency of the impurity sorting system are improved, the problems that the existing technology lacks consideration of the influence of the environment and the feature quality on the sorting result, the sorting result is poor, the sorting scheme is relatively single and cannot adapt to the change of the working condition data, the effect of the sorting scheme is poor, and the accuracy and efficiency of the impurity sorting system are low are avoided.

[0159] The above embodiments are only used to illustrate the technical method of the present application and not to limit it, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. An ore impurity sorting system characterised in that, The application relates to a mineral separation method and device. The application comprises a data acquisition module and a data analysis module connected with each other. The data acquisition module acquires environmental data, ore data and target data through a data acquisition device. The data analysis module generates decision data according to the ore data and the environmental data, inputs the decision data into an impurity prediction joint model to obtain an impurity category and an impurity content, and constructs the impurity prediction joint model through an artificial intelligence model. A separation scheme is generated according to the impurity category, the impurity content and the target data. The decision data is generated according to the ore data and the environmental data, and the method comprises the following steps. Ore data and environmental data are acquired, the ore data comprises ore parameters and multi-modal data, and the environmental data comprises temperature, humidity, dust concentration and vibration intensity. Modal quality scores are generated according to the multi-modal data. Environmental influence coefficients are generated according to the environmental data. A weight coefficient function QZF (MZP, HYX) is constructed through a nonlinear relationship between the modal quality scores and the environmental influence coefficients and modal weight coefficients, wherein MZP represents the modal quality scores, and HYX represents the environmental influence coefficients. The weight coefficients corresponding to the multi-modal data are calculated by substituting the modal quality scores and the environmental influence coefficients corresponding to the multi-modal data into the weight coefficient function. The weight coefficients and the corresponding multi-modal data are combined to obtain the decision data.

2. An ore impurity separation system according to claim 1, characterised in that, The modal quality scores are generated according to the multi-modal data, and the method comprises the following steps. The multi-modal data comprises X-ray fluorescence spectrum XRF, laser-induced breakdown spectroscopy LIBS, X-ray diffraction XRD and hyperspectral imaging HSI. extracting a plurality of index parameters ZC corresponding to the multi-modal data j ; According to the nonlinear relationship between the several index parameters and the corresponding modal quality scores, a scoring function PF(ZC j ) is constructed; wherein j represents the number of several index parameters corresponding to the modal data; Several index parameters in the multi-modal data are substituted into corresponding scoring functions to calculate the modal quality scores corresponding to the multi-modal data. The normalized modal quality scores are obtained by maximizing and minimizing the several modal quality scores.

3. An ore impurity separation system according to claim 1, characterised in that, The environmental influence coefficients are generated according to the environmental data, and the method comprises the following steps. acquiring environmental data; the environmental data refers to a plurality of modality influence data MYS which have an influence on the multi-modal data i ; An influence function YF(MYS i ) is constructed by a nonlinear relationship between the several modal influence data and the environmental influence coefficient of the corresponding multimodal data; wherein i represents the number of several modal influence data corresponding to the multimodal data; Several modal influence data corresponding to the multi-modal data are substituted into corresponding influence functions to calculate the environmental influence coefficients corresponding to the multi-modal data.

4. An ore impurity separation system according to claim 1, characterised in that, The impurity prediction joint model is constructed through an artificial intelligence model, and the method comprises the following steps. Several historical decision data, historical impurity categories and historical impurity contents are acquired. The several historical decision data, the historical impurity categories and the historical impurity contents are divided into a training set, a test set and a verification set. An artificial intelligence model is selected as a base model. The base model is trained through the training set, and a pre-training model is obtained by adjusting a learning rate or other hyperparameters on the verification set. The pre-training model is verified on the test set, and finally the impurity prediction joint model with the decision data as input and the impurity category and the impurity content as output is obtained.

5. An ore impurity separation system according to claim 1, characterised in that, The separation scheme is generated according to the impurity category, the impurity content and the target data, and the method comprises the following steps. The impurity category, the impurity content, the target data and working condition data are acquired. Several target sub-functions MZF are constructed according to the target data. Generating an optimization function based on a number of objective sub-functions MZF ; the optimization function satisfies the following equation: ; wherein k represents a number corresponding to a number of target sub-functions, represents an optimization weight corresponding to a number of target sub-functions; the optimization weight is generated by working condition data; Optimal parameter values are obtained by solving an optimization function through a parameter optimization algorithm, and the separation scheme is obtained by integrating the optimal parameter values.

6. An ore impurity separation system according to claim 5, characterised in that, The several target sub-functions MZF are constructed according to the target data, and the method comprises the following steps. Obtaining target data; the target data includes purity, recovery rate, cost and energy consumption; Constructing a purity target sub-function CMZF through a nonlinear relationship between purity and purity optimization; Constructing a recovery rate target sub-function HMZF through a nonlinear relationship between recovery rate and recovery rate optimization; Constructing a cost target sub-function CBMZF through a nonlinear relationship between cost and cost optimization; Constructing an energy consumption target sub-function NMZF through a nonlinear relationship between energy consumption and energy consumption optimization; The target sub-function includes a purity target sub-function CMZ, a recovery rate target sub-function HMZF, a cost target sub-function CBMZF and an energy consumption target sub-function NMZF.

7. An ore impurity separation system according to claim 5, characterised in that, The optimization weight is generated through working condition data, including: Obtaining working condition data; the working condition data includes working condition parameter values GC corresponding to a plurality of target sub-functions; Constructing a priority score function YPF(GC) according to a nonlinear relationship between a plurality of working condition parameter values and priority scores; Calculating a plurality of target sub-functions corresponding priority levels YJ by substituting a plurality of working condition parameter values into the priority score function YPF(GC); Constructing an optimization weight calculation function YQF(YJ) according to a nonlinear relationship between priority levels and optimization weights; Calculating a plurality of target sub-functions corresponding optimization weights MQ by substituting a plurality of priority levels into the optimization weight calculation function YQF(YJ).

8. A method for sorting ore impurities, applied to the ore impurity sorting system according to any one of claims 1-7, characterized in that, Including: Obtaining environment data, ore data and target data; Generating decision data according to ore data and environment data; The decision data is generated according to the ore data and the environment data, including: Obtaining ore data and environment data; the ore data includes ore parameters and multi-modal data; the environment data includes temperature, humidity, dust concentration and vibration intensity; Generating a modal quality score according to multi-modal data; Generating an environmental impact coefficient according to environment data; Constructing a weight coefficient function QZF(MZP, HYX) through a nonlinear relationship between the modal quality score and the environmental impact coefficient and the modal weight coefficient; wherein MZP represents the modal quality score, and HYX represents the environmental impact coefficient; Substituting the modal quality score and the environmental impact coefficient corresponding to the multi-modal data into the weight coefficient function to calculate the weight coefficient corresponding to the multi-modal data; Combining the weight coefficient and the corresponding multi-modal data to obtain decision data; Inputting the decision data into an impurity prediction joint model to obtain impurity categories and impurity contents; the impurity prediction joint model is constructed through an artificial intelligence model; Generating a sorting scheme according to the impurity categories, the impurity contents and the target data.

9. A computer readable storage medium for use with a mineral impurity sorting system according to any one of claims 1 to 7, wherein, The computer program is stored on the computer readable storage medium.

Images