Ore resource dynamic allocation optimization method and system

Abstract

The invention provides an ore resource dynamic allocation optimization method and system, and the method comprises the steps: predicting and obtaining the ore configuration of an initial population individual in a next working condition environment based on the ore configuration in a historical working condition environment through a point set registration method; and after an excellent initial population is obtained through prediction, optimization solution is carried out on the initial population based on the multi-target optimization model, and an ore configuration scheme corresponding to a target optimal value under global optimization is obtained. According to the method, distribution experience under historical working conditions is integrated, dynamic optimization is carried out in combination with an optimization target, reasonable ore distribution under a complex working condition scene is achieved, and the ore output rate and the high-quality concentrate content are overall improved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to artificial intelligence, and particularly relates to a method and system for optimizing the dynamic allocation of ore resources. Background Art

[0002] The statements in this part merely provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In the context of the current increasingly scarce resources and continuously improving environmental protection requirements, the sustainable development of resources has become a crucial issue. For the mining industry, with the growth of the global economy and technological progress, the demand for mineral resources has been continuously rising, which not only accelerates the consumption of resources but also exerts great pressure on the ecological environment. To achieve the sustainable development goal, the mining industry is transforming from the traditional stock development mode to the demand-driven development mode, emphasizing the rational allocation of raw ore resources according to the actual inventory, equipment status, and market demand. This transformation requires the mine to be able to respond to changes in the market and equipment efficiency in real time and dynamically adjust the processing volume of different raw ores to maximize the utilization efficiency of raw ore.

[0004] Raw ore allocation is used to separate tailings from raw ore to recover valuable minerals. The main optimization objectives of this process include the concentrate yield, i.e., the product output of the concentrator, and the concentrate grade, i.e., the proportion of valuable components in the concentrate. These two indicators respectively reflect the quantity and quality of the concentrate. However, in the mineral processing process, due to different roasting processes and differences in the physical properties of ores, the optimization objectives of the concentrate yield and grade exhibit significant non-linear characteristics; in addition, the inventory content of different types of ores and the upper and lower limit constraints of available resources are complex and diverse, and it is necessary to ensure that the concentrate grade meets a specific demand range; moreover, the efficiency decline or abnormal interruption of equipment due to wear during the smelting process leads to dynamic changes in equipment constraints and model parameters, making it difficult for traditional optimization methods to adapt to this complex real situation. Therefore, an efficient method for dealing with complex dynamic working conditions is needed to generate a more reasonable raw ore allocation plan. Summary of the Invention

[0005] To overcome the deficiencies of the above-mentioned prior art, the present invention provides a method and system for optimizing the dynamic allocation of ore resources, integrating the allocation experience under historical working conditions, dynamically optimizing in combination with optimization objectives, realizing the reasonable allocation of ores in complex working condition scenarios, and overall improving the ore output rate and the content of high-quality concentrate.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In the first aspect, the present invention provides a method for optimizing the dynamic allocation of ore resources, including:

[0008] Taking the maximization of concentrate grade and the maximization of concentrate output as the optimization objectives, a multi-objective optimization model for raw ore allocation is constructed;

[0009] Regarding the ore configuration plan as the decision variable, using the point set registration method, and based on the ore configuration under the historical working conditions, the ore configuration under the next working condition is predicted as an individual of the initial population;

[0010] Using the dual-task idea, combining the multi-objective optimization model of each type of raw ore and the optimal configuration information of the historical working conditions, the initial population is processed until convergence, and the ore configuration plan corresponding to the optimal value of the objective under global optimization is obtained.

[0011] In a second aspect, the present invention provides an ore resource dynamic allocation optimization system, including:

[0012] A model construction module, which is configured to: take the maximization of concentrate grade and the maximization of concentrate output as the optimization objectives, and construct a multi-objective optimization model for raw ore allocation;

[0013] A prediction module, which is configured to: regard the ore configuration plan as the decision variable, use the point set registration method, and based on the ore configuration under the historical working conditions, predict the ore configuration under the next working condition as an individual of the initial population;

[0014] An optimization module, which is configured to: use the dual-task idea, combine the multi-objective optimization model of each type of raw ore and the optimal configuration information of the historical working conditions, process the initial population until convergence, and obtain the ore configuration plan corresponding to the optimal value of the objective under global optimization.

[0015] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in the first aspect is completed.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by the processor, the method described in the first aspect is completed.

[0017] The above one or more technical solutions have the following beneficial effects:

[0018] In the present invention, a point set registration method is used to predict the ore configuration in the next working condition environment as an individual of the initial population based on the ore configuration in the historical working condition environment. After obtaining an excellent initial population through prediction, an initial population is optimized and solved based on a multi-objective optimization model to obtain an ore configuration plan corresponding to the optimal value of the global optimization objective. The present invention integrates the distribution experience in historical working conditions and conducts dynamic optimization in combination with the optimization objective to achieve reasonable ore distribution in complex working condition scenarios, and overall improve the ore output rate and the content of high-quality concentrate.

[0019] Advantages of additional aspects of the present invention will be partly given in the following description, partly will become apparent from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0021] Figure 1 It is a block diagram of a method for optimizing the dynamic allocation of ore resources in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0023] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0024] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0025] Embodiment 1

[0026] This embodiment discloses a method for optimizing the dynamic allocation of ore resources, including:

[0027] Construct a dynamic multi-objective optimization model for raw ore allocation with the optimization objectives of maximizing concentrate grade and maximizing concentrate output;

[0028] Taking the raw ore configuration plan as a decision variable, using the point set registration method, and predicting the ore configuration in the next working condition environment as an individual of the initial population based on the ore configuration in the historical working condition environment;

[0029] Using the dual-task idea, combining the multi-objective optimization model of each type of raw ore and the optimal configuration information of historical working conditions, the initial population is processed until convergence, and the ore configuration plan corresponding to the optimal value of the objective under global optimization is obtained.

[0030] The specific implementation is mainly divided into three parts: constructing the problem, the multi-agent trajectory tracking strategy based on registration, and the dual-task optimization strategy of experience fusion. The relevant technical details of these three parts are described separately below.

[0031] 1. Disclosure of constructing the problem:

[0032] The optimization objectives of this model are as follows:

[0033]

[0034] Among them, represents the content of each type of raw ore, represents the output of concentrate, represents the grade of concentrate.

[0035] The first optimization objective is to maximize the output of concentrate:

[0036]

[0037] Among them, u i represents the lump ore ratio of raw ore i; k 1,i represents the beneficiation ratio of the fine ore screened from raw ore i; k 2,i represents the beneficiation ratio of the lump ore screened from raw ore i.

[0038] The second optimization objective is to maximize the grade of concentrate:

[0039]

[0040] Among them, β 1,i refers to the grade of strong magnetic concentrate of raw ore i; β 2,i refers to the grade of weak magnetic concentrate of raw ore i.

[0041] The constraint conditions of this model are as follows:

[0042] Constraint on the output of concentrate:

[0043]

[0044] Among them, [Q L ,Q H represents the target interval of the grade of concentrate.

[0045] Constraint on the grade of concentrate:

[0046]

[0047] Among them, [β L , β H represents the target interval of concentrate grade.

[0048] Equipment production capacity constraint:

[0049]

[0050] Among them, N k,j represents the number of operating units of the jth type of equipment when the production series number is k; T k represents the production operation time of series number k within a certain time period; q j,H (t) represents the upper limit of the average hourly processing capacity of the jth type of equipment; η b represents the roasting yield.

[0051] Tailings grade constraint:

[0052]

[0053]

[0054] Among them, α 1,i represents the grade of the fine ore obtained after the separation of the original ore i through the screening process; α 2,i represents the grade of the lump ore obtained after the separation of the original ore i through the screening process; η w represents the waste rock rate; β w represents the waste rock grade; θ 0,H represents the upper limit of the total tailings grade, θ 1,H represents the upper limit of the strong magnetic tailings grade, θ 2,H represents the upper limit of the weak magnetic tailings grade.

[0055] Available original ore resource constraint:

[0056] Q i,min ≤ x i ≤ Q i,max , i = 1, 2,..., I (12)

[0057] Among them, [Q i,min , Q i,max represents the available supply interval of the original ore i during the planning period.

[0058] To predict potential configuration solutions under new working conditions, a multi-agent trajectory tracking strategy based on registration is designed, including: using the point set registration method to convert the individual trajectory prediction problem into a point set registration problem; using the first-order linear prediction method to predict individuals in combination with the registration results; adding a duplicate removal operation to reduce misleading trajectories; using a random strategy to enhance diversity and avoid falling into local optima; discovering that the tasks of different populations are different, and making targeted predictions based on the respective tasks of the two populations to improve the search accuracy of the optimization solution.

[0059] The algorithm flow of the multi-agent trajectory tracking strategy based on registration is as follows:

[0060] Since different types of raw ores need to be processed simultaneously, the contents of different types of raw ores are used as decision vectors, and one dimension of the vector represents the content of one type of ore; a set of decision vectors, that is, a raw ore configuration solution, is regarded as an individual; each individual can calculate the optimization objective and constraint conditions according to the proposed ore dressing model; during the evolution process, the set of individuals is represented as a population.

[0061] It should be noted that a raw ore configuration solution includes the processing amounts of all types of ores. For example, if there are three types of ores A, B, and C, there are many choices for the processing amounts of the three ores, and each choice represents a configuration solution. For example: Configuration 1: A(10), B(15), C(20); Configuration 2: A(20), B(13), C(10)…. Each configuration solution represents an individual, and the set of several configuration solutions (individuals) constitutes a population.

[0062] Under the background of dynamic changes in working conditions, the distribution and position of the optimal configuration set in the vector space are time-series correlated. Therefore, according to the movement trend of the optimal configuration set in the vector space under historical working conditions, its position under new working conditions can be predicted. Based on this idea, this algorithm pairs the configuration sets at adjacent historical moments to determine the movement direction of each configuration after the environmental change, and takes the movement direction of each configuration as its movement trend under new working conditions, so as to realize the prediction of the optimal configuration set.

[0063] Since it is necessary to overcome complex constraints such as output, grade, and equipment production capacity brought by dynamic working conditions, the analysis of specific optimization tasks is carried out for each type of population, and sample groups are respectively formed according to the populations under different tasks:

[0064] Step 1-1: For each sample group, extract the data Set_t at time t, that is, the set of each configuration solution at time t as the source domain A, and the data Set_{t - 1} at time t - 1, that is, the set of each configuration solution at time t - 1 as the target domain B;

[0065] Step 1-2: Take each source domain configuration as the center point of the Gaussian mixture model, and use the Gaussian mixture model to obtain:

[0066]

[0067] Determine the selection probability of each target domain configuration, where p(l) and p(a|l) respectively represent the prior probability and probability density function of the l-th configuration at time t, w represents the proportion of outliers, 0 ≤ w ≤ 1, U represents the number of configuration schemes in the set Sett, and L represents the number of configuration schemes in the set Sett-1;

[0068] Step 1-3: Convert the formula in Step 1-2 into an optimization to minimize the negative log-likelihood function:

[0069]

[0070] where θ is used to estimate the centroid position of the Gaussian mixture model, σ is the offset parameter, p(a u |l) represents the probability of being paired to the a u configuration at time t-1 in the case of the l-th configuration at time t.

[0071] Step 1-4: Optimize the likelihood function in Step 1-3 through the expectation-maximization algorithm to determine the optimal configuration under the working condition at time t-1 and the configuration pairing relationship of the optimal configuration under the working condition at time t:

[0072]

[0073] where p old (l|a u ) represents the probability that the a u configuration on the point set Sett-1 at time t-1 and the l-th configuration on the point set Sett belong to a one-to-one matching relationship. T(b l ,θ) represents the configuration paired with the configuration b l at time t under the θ parameter, and D represents the number of ore types.

[0074] The configuration sets at two times are regarded as two point sets, and there is a certain similarity between the two point sets. Based on the similarity, the two point sets can be registered to achieve a one-to-one matching relationship for each individual between the point sets.

[0075] Step 1-5: Based on the configuration pairing relationship, mark the orientation of the corresponding configuration as the moving direction of the source domain configuration;

[0076] Step 1-6: Take the vector set V t of the moving direction of each determined configuration as its future moving trend, and predict the potential optimal configuration set Sett+1 in the next environment through the following formula:

[0077] Set t+1 = Set t + V t

[0078] Both Set and V are spatial vectors with the same dimension. Each row represents the position and direction of a configuration respectively.

[0079] Steps 1 - 7: Randomly initialize a set of configurations Rand t+1 with a size of N / 10;

[0080] Steps 1 - 8: Randomly replace N / 10 individuals in Sett+1 with those from Rand t+1.

[0081] It should be noted that the above is the configuration prediction process for a certain population. Due to the differences in tasks among different populations, targeted predictions are made based on the respective tasks of the two populations, which can effectively improve the accuracy of finding the optimization scheme. Based on this discovery, the configurations of the two populations are archived, recording the historical feasible optimal configuration individuals CPSs, the historical unconstrained optimal configuration individuals UPSs, and the historical feasible optimal configuration individual objectives CPFs. Respectively, take the two most recent working condition environment configuration individual sets CPSt and CPSt-1 in CPSs as the source domain and the target domain; take the two most recent working condition environment configuration individual sets UPSt and UPSt-1 in UPSs as the source domain and the target domain; take the two most recent working condition environment configuration individual objective sets CPSFt and CPSFt-1 in CPFs as the source domain and the target domain, and predict three populations, which are respectively used as the main population, the auxiliary population, and the judgment population, so as to achieve targeted predictions for the initial populations of different tasks.

[0082] To predict potential configuration schemes under new working conditions, for this purpose, a dual-task optimization strategy with experience fusion is designed. This strategy sets two different optimization tasks, uses two populations to optimize these two tasks respectively, and realizes knowledge transfer between the two populations, thereby improving the optimization efficiency of the two populations. The algorithm process is as follows:

[0083] For the main population task: Under the condition of meeting various working condition constraints, find the ore configuration that can achieve the optimal concentrate content and ore output.

[0084] Optimize and solve through the differential evolution algorithm, specifically:

[0085] Steps 2 - 1: Generate the main population through the multi-individual trajectory tracking strategy algorithm, and calculate the concentrate content, ore output, and constraint violation degree (the sum of each constraint value exceeding the corresponding constraint range) of each configuration in the initial configuration set through the optimization objective of the ore dressing model;

[0086] Steps 2 - 2: For each configuration x in the main populationi Generate mutant configuration v i :

[0087] v i = x r 1 + F · (x r 2 - x r 3)

[0088] where x r 1, x r 2, x r 3 are three different configuration individuals randomly selected from the main population, and F is the scaling factor;

[0089] Step 2 - 3: Generate the trial configuration through crossover operation:

[0090]

[0091] where CR is the crossover probability; rand j (0, 1) is a random number; j rand is a randomly selected dimension;

[0092] Step 2 - 4: Compare the trial configuration u ij and the current configuration x ij and select the better configuration according to the fitness value as the offspring configuration;

[0093] Step 2 - 5: Sort the offspring configuration sets of the main population and the auxiliary population according to the following criteria: For two configurations, if both satisfy the constraints, select the one with a better objective; if only one satisfies the constraints, select the one that satisfies the constraints; if both do not satisfy the constraints, select the configuration with a smaller constraint violation degree.

[0094] Step 2 - 6: Determine the best N / 2 configurations in the sorting and let them survive as the potential excellent configurations of the main population under the new working conditions to enter the next generation;

[0095] Step 2 - 7: Judge whether the working condition environment has changed. If it has changed, output the best N / 2 configurations for the constructor to select by themselves and enter Step 2 - 1; otherwise, enter Step 2 - 2.

[0096] For the auxiliary population task: Without considering or partially considering the working condition constraints, quickly find the potential optimal concentrate content and ore configuration of the ore output.

[0097] Optimize and solve through the differential evolution algorithm, specifically:

[0098] Step 3-1: Generate an auxiliary population through multiple individual trajectory tracking strategy algorithms, and calculate the concentrate content, ore output, and constraint violation degree (the sum of each constraint value exceeding the corresponding constraint range) of each configuration in the initial configuration set through the optimization objective of the ore dressing model;

[0099] Step 3-2: For each configuration x in the auxiliary population i Generate a mutant configuration v i :

[0100] v i = x r 1 + F·(x r 2 - x r 3)

[0101] where x r 1, x r 2, x r 3 are three different configuration individuals randomly selected from the auxiliary population, and F is the scaling factor;

[0102] Step 3-3: Generate a trial configuration through crossover operation:

[0103]

[0104] where CR is the crossover probability; rand j (0, 1) is a random number; j rand is a randomly selected dimension;

[0105] Step 3-4: Compare the trial configuration u ij and the current configuration x ij and select the better configuration as the offspring configuration according to the fitness value;

[0106] Step 3-5: Calculate the constraint violation degree of each configuration in the main population and the auxiliary population, and record the maximum value Gmax;

[0107] Step 3-6: Set ε = Gmax;

[0108] Step 3-7: Update ε through . Among them, represents the proportion of the evaluation resources already used in the current environment;

[0109] Step 3-8: Record the configurations with constraint violation degrees less than ε as feasible configurations that meet the working condition constraints, and perform non-dominated and crowding degree sorting on the population to obtain the ε relaxation ranking Ranking1 of each configuration;

[0110] Step 3-9: Determine the Pareto front SCPF of the similar environment according to the similar environment judgment rule;

[0111] Step 3-10: Calculate the distances from each vector of SCPF to all configuration individuals in sequence, and sort the distances.

[0112] Step 3-11: Configurations with the same distance sorting serial number are considered to be in the same sequence, and ranking within the sequence is performed through crowding degree to obtain the EDT-1 ranking Ranking2 of each configuration individual.

[0113] Step 3-12: Based on the rankings of Ranking1 and Ranking2, group according to the conclusion of whether to retain the configuration individuals.

[0114] Step 3-13: Through Determine the weights of the two strategies when dealing with controversial configurations.

[0115] Step 3-14: Determine the number of configurations screened by the two strategies according to the weights of different strategies.

[0116] Step 3-15: Determine the surviving N / 2 configurations based on the grouping situation and the number of screened configurations, and use them as potential excellent configurations of the auxiliary population under the new working conditions.

[0117] Step 3-16: Judge whether the working condition environment has changed. If it has changed, enter Step 3-1; otherwise, enter Step 3-7.

[0118] To verify the effectiveness of this embodiment, this embodiment is used to optimize the raw ore allocation problem with 20 different working condition environments, and the relevant data are shown in Table 1, Table 2, and Table 3. Among them, the parameters of the present invention are: the population size N is set to 100, and the maximum number of evaluations is 20*τ t *D.

[0119] Table 1 Settings of relevant parameters for the raw ore allocation model constraints

[0120]

[0121]

[0122] Table 2 Settings of relevant parameters for the raw ore in the raw ore allocation model

[0123] i <![CDATA[α i > <![CDATA[α 1,i > <![CDATA[α 2,i > <![CDATA[β 1,i > <![CDATA[β 2,i > <![CDATA[u i > <![CDATA[k 1,i > <![CDATA[k 2,i > <![CDATA[r i > <![CDATA[Q i,min > <![CDATA[Q i,max > 1 33.3 32.16 34.0 47.5 56.1 62.0 2.3 2.0 67 60000 80000 2 34.1 32.34 35.6 47.5 56.1 54.0 2.0 1.92 67 180000 210000 3 32.5 31.8 33.2 47.0 54.0 50.0 2.3 2.2 67 80000 100000 4 33.1 31.99 33.7 44.5 52.0 65.0 2.8 2.4 67 50000 70000 5 55.0 52.56 57.0 60.0 61.0 55.0 1.45 1.4 220 5000 10000 6 52.0 52.0 0 60.0 0 0 1.3 0 220 30000 40000

[0124] Table 3 Settings of relevant parameters for the working condition environment in the raw ore allocation model

[0125] Index t＝1 t＝2 t＝3 t＝4 t＝5 t＝6 t＝7 t＝8 t＝9 t＝10 <![CDATA[q 1,H > 25 25 25 25 25 25 25 25 25 25 <![CDATA[q 2,H > 95 85 80 90 85 80 80 75 75 90 <![CDATA[q 3,H > 70 70 70 80 80 75 70 75 70 95 Index t＝11 t＝12 t＝13 t＝14 t＝15 t＝16 t＝17 t＝18 t＝19 t＝20 <![CDATA[q 1,H > 25 25 25 25 25 25 25 25 25 25 <![CDATA[q 2,H > 90 90 80 85 85 80 80 95 85 85 <![CDATA[q 3,H > 90 80 80 75 70 85 80 80 75 70

[0126] The optimization method of the present invention is compared with several existing methods, namely DCNSGAII-A, DCNSGAII-B, DC-MOEA, dCMOEA, mEDCMOA, TDCEA, HATC, and SKTEA. All algorithms are run 20 times to obtain statistical results. Among them, DCNSGAII-A and DCNSGAII-B are the algorithms disclosed in the literature "R. Azzouz, S. Bechikh, and L. Ben Said, 'Multi-objective optimization with dynamic constraints and objectives: new challenges for evolutionary algorithms,' in Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation, 2015, pp. 615–622."; DC-MOEA is the algorithm disclosed in the literature "R. Azzouz, S. Bechikh, L. B. Said, and W. Trabelsi, 'Handling time-varying constraints and objectives in dynamic evolutionary multi-objective optimization,' Swarm and Evolutionary Computation, vol. 39, pp. 222–248, 2018."; dCMOEA is the algorithm disclosed in the literature "Q. Chen, J. Ding, S. Yang, and T. Chai, 'A novel evolutionary algorithm for dynamic constrained multiobjective optimization problems,' IEEE Transactions on Evolutionary Computation, vol. 24, no. 4, pp. 792–806, 2019."; mEDCMOA is the algorithm disclosed in the literature "Q. Chen, J. Ding, G. G. Yen, S. Yang, and T.The algorithms disclosed in Chai, "Multi-population evolution based dynamic constrained multiobjective optimization under diverse changing environments," IEEE Transactions on Evolutionary Computation, 2023, doi: 10.1109 / TEVC.2023.3241762; TDCEA is the algorithm disclosed in the literature "G. Chen, Y. Guo, Y. Wang, J. Liang, D. Gong, and S. Yang, "Evolutionary dynamic constrained multiobjective optimization: Test suite and algorithm," IEEE Transactions on Evolutionary Computation, 2023, doi: 10.1109 / TEVC.2023.3313689. HATC is the algorithm disclosed in the literature "D. Zhang, K. Yu, J. Liang, K. Qiao, B. Qu, K. Chen, C. Yue, and L. Wang, "History-assisted two-state auxiliary task collaboration approach for dynamic constrained multiobjective optimization," IEEE Transactions on Evolutionary Computation, doi: 10.1109 / TEVC.2024.3425756; SKTEA is the algorithm disclosed in the literature "G. Chen, Y. Guo, M. Jiang, S. Yang, X. Zhao, and D. Gong, "A subspace-knowledge transfer based dynamic constrained multiobjective evolutionary algorithm," IEEE Transactions on Emerging Topics in Computational Intelligence, 2023, doi: 10.1109 / TETCI.2023.3336918.

[0127] The average maximum hypervolume (MHV) value and the average proportion of feasible solutions (MFSR) obtained by the algorithm are shown in Table 3. Among them, the larger the MHV value and MFSR, the better the algorithm performance. Analyzing the data in Table 3, it can be seen that compared with 8 comparison methods, the method of the present invention achieves optimal or sub-optimal results in most configurations and cases of the two indicators of MHV value and MFSR value. This is attributed to the proposed multi-agent trajectory tracking method based on registration and the two-level environment selection method based on dynamic fusion. In addition, generating multiple populations using the newly proposed trajectory tracking method also improves the optimization performance. In actual production operations, users can fully combine these solutions according to their own situations to plan a satisfactory energy scheduling and allocation plan.

[0128] Table 3 Performance comparison statistics of the MP-DFR of the present invention and comparison algorithms according to MFSR and MHV indicators

[0129]

[0130] "NaN" means that no feasible solution can be found in the environment. The bold in the gray background represents the best value, and the bold represents the sub-optimal value

[0131] Example Two

[0132] The purpose of this embodiment is to provide a dynamic allocation optimization system for ore resources, including:

[0133] A model construction module, which is configured to: construct a multi-objective optimization model for raw ore allocation with the maximization of concentrate grade and the maximization of concentrate output as the optimization objectives;

[0134] A prediction module, which is configured to: use the raw ore configuration plan as a decision variable, and use the point set registration method to predict the ore configuration in the next working condition environment as an individual of the initial population based on the ore configuration in the historical working condition environment;

[0135] An optimization module, which is configured to: use the dual-task idea, combine the multi-objective optimization model of each type of raw ore and the optimal configuration information of the historical working conditions to process the initial population until convergence, and obtain the ore configuration plan corresponding to the optimal value of the objective under global optimization.

[0136] In more embodiments, there is also provided:

[0137] An electronic device includes a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in Embodiment 1 is completed. For the sake of brevity, it will not be elaborated here.

[0138] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0139] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0140] A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the method described in Embodiment 1 is completed.

[0141] The method in Embodiment 1 can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0142] A computer program product includes a computer program. When the computer program is executed by the processor, the method described in Embodiment 1 is implemented.

[0143] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to execute the process / method as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules may be combined or divided as needed. The machine-executable instructions for program modules may be executed locally or within a distributed device. In a distributed device, program modules may be located in local and remote storage media.

[0144] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program codes are executed by the computer or other programmable data processing devices, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program codes can be executed entirely on the computer, partially on the computer, as an independent software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0145] In the context of the present invention, the computer program code or related data can be carried by any suitable carrier so that the devices, apparatuses, or processors can perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals can include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0146] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0147] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, this is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A method for optimizing the dynamic allocation of ore resources, characterized in that: include: Taking maximizing concentrate grade and maximizing concentrate output as optimization goals, a multi-objective optimization model for raw ore allocation is constructed; The original ore configuration scheme is used as a decision variable, and the point set registration method is used to predict the ore configuration under the next working condition environment as the initial population individual based on the ore configuration under the historical working condition environment. Using the dual-task idea, combined with the multi-objective optimization model of each original ore and the optimal configuration information of historical working conditions, the initial population is processed until convergence, and the ore configuration scheme corresponding to the optimal value of the target under global optimization is obtained.

2. A method for optimizing the dynamic allocation of mineral resources according to claim 1, characterized in that: The constraints include concentrate output constraints, concentrate grade constraints, equipment production capacity constraints, tailings grade constraints and available raw ore resource constraints.

3. A method for optimizing dynamic allocation of ore resources as claimed in claim 1, characterized in that: The original ore configuration scheme is used as the decision variable, and the point set registration method is used to predict the ore configuration under the next working condition as the initial population individual based on the ore configuration under the historical working condition environment. Specifically, The current ore configuration scheme set and the previous ore configuration scheme set are used as the source domain and the target domain respectively; Each ore configuration scheme in the source domain is taken as the center point of the Gaussian mixture model, and the Gaussian mixture model is used to determine the selection probability of each ore configuration in the target domain; Determine the configuration correspondence between the source domain and the target domain according to the selection probability, and mark the orientation of the corresponding ore configuration scheme as the moving direction of the ore configuration scheme in the source domain; According to the moving direction of each ore configuration scheme in the determined source domain, the ore configuration under the next working condition environment as the initial population individual is predicted.

4. A method for optimizing the dynamic allocation of mineral resources as claimed in claim 3, characterized in that: Each ore configuration scheme in the source domain is taken as the center point of the Gaussian mixture model, and the Gaussian mixture model is used to determine the selection probability of each ore configuration in the target domain, specifically: The calculation of the probability of selection of each ore configuration determined by the Gaussian mixture model is converted into minimizing the negative log-likelihood function; The negative log-likelihood function is minimized by the expectation maximization algorithm to determine the satisfied configuration matching relationship.

5. A method for optimizing dynamic allocation of ore resources as claimed in claim 1, characterized in that: Using the dual-task idea, combined with the multi-objective optimization model of each ore and the optimal configuration information of historical working conditions, the initial population is processed until convergence, and the ore configuration scheme corresponding to the optimal value of the target under global optimization is obtained, specifically: According to different tasks, the initial population is divided into the main population and the auxiliary population; For the main population, find the ore configuration that can achieve the optimal concentrate content and ore output under the conditions of meeting various working condition constraints; For auxiliary populations, the ore configuration with the potential optimal concentrate content and ore output is sought without or partially considering the operating condition constraints; The differential evolution algorithm is used to optimize the main population and the auxiliary population, and the ore configuration scheme corresponding to the optimal value of the target under global optimization is obtained.

6. A method for optimizing dynamic allocation of mineral resources as claimed in claim 5, characterized in that: The main population is optimized and solved by differential evolution algorithm, specifically: Generate variant ore configurations for each ore configuration of the initial population, and generate test ore configurations through crossover operations; Compare the test ore configuration with the current ore configuration, and select the ore configuration as the offspring ore configuration according to the fitness value; Sort the offspring ore configuration sets of the main population and auxiliary population to determine the potential excellent configurations that survive and enter the next generation as the main population under the new working conditions; Determine whether the working environment has changed. If so, output the ore configuration for selection.

7. A method for optimizing dynamic allocation of ore resources as claimed in claim 5, characterized in that: The auxiliary population is optimized and solved by differential evolution algorithm, specifically: The auxiliary population is subjected to mutation and crossover operations to obtain the offspring configuration set of the auxiliary population; Calculate the constraint violation of each configuration in the main population and auxiliary population; The configurations with constraint violation less than ε are recorded as feasible configurations that meet the working condition constraints, and the population is sorted by non-domination and congestion to obtain the ε relaxation ranking of each configuration Ranking1; Determine the Pareto frontier of similar environments based on the similar environment judgment rule; Calculate the distance from each vector of the Pareto front to all configuration individuals in turn, and sort the distances; Configurations with the same distance sorting number are considered to be in the same sequence, and are ranked within the sequence by crowding to obtain the EDT-1 ranking Ranking2 of each configuration individual; Based on Ranking1 and Ranking2, group the results according to whether to retain the individual configurations; Based on the grouping situation and the number of screened configurations, the potential excellent configuration of the auxiliary population under the new working condition is determined.

8. A dynamic allocation optimization system for ore resources, characterized in that: include: A model building module, which is configured to: construct a multi-objective optimization model for raw ore allocation with maximizing concentrate grade and maximizing concentrate output as optimization objectives; The prediction module is configured to: take the original ore configuration scheme as a decision variable, use a point set registration method, and predict the ore configuration under the next working condition environment as an individual of the initial population based on the ore configuration under the historical working condition environment; The optimization module is configured as follows: using the dual-task concept, combining the multi-objective optimization model of each original ore and the optimal configuration information of historical working conditions, the initial population is processed until convergence, and the ore configuration scheme corresponding to the optimal value of the target under global optimization is obtained.

9. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 7 is completed.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 7.

Images