Coking coal blending optimization decision-making method and system based on dream bionic mechanism

Abstract

The invention relates to the technical field of coking coal blending, in particular to a coking coal blending optimization decision-making method and system based on a dream bionic mechanism, and the method comprises the steps: obtaining quality index data of coking coal blending, and preprocessing the data; constructing a coke quality prediction model based on a BP neural network, and predicting a coke quality index as a constraint condition of coal blending ratio optimization; and constructing a coal blending optimization model by adopting a dream optimization algorithm, and outputting an optimal coal blending ratio scheme and a corresponding fitness value. According to the method, the problem of model training difficulty caused by insufficient original data volume is solved, meanwhile, reliable quality feedback is provided for coal blending optimization, most importantly, a dream optimization algorithm is adopted, and through staged design and three optimization strategies, the optimization efficiency of a coal blending scheme is improved, global search and local refinement are balanced, and the optimization efficiency of the coal blending scheme is improved. Local optimum is effectively avoided, and multi-target coal blending optimization decision under complex constraint conditions is realized.

Description

Technical Field

[0001] This invention relates to the field of coking coal blending technology, and in particular to a coking coal blending optimization decision-making method and system based on a dream-inspired bionic mechanism. Background Technology

[0002] With the decreasing availability of high-quality coking coal resources, blending various types of coal for coking has become an inevitable trend. Finding suitable coal blending schemes to obtain coke that meets quality standards is an urgent need for the coking industry.

[0003] Certain research achievements have been made in the field of coking coal blending. Addressing the problems of subjective bias, low accuracy, poor inheritability, and low efficiency inherent in manual coal blending, the industry has attempted to use linear programming algorithms for solutions. For example, by establishing a linear programming model to solve the blending ratio problem, a preliminary solution has been found for coal blending optimization problems with relatively simple constraints, improving quality stability and decision-making efficiency. However, the constraints of actual coal blending are very complex, requiring compliance with multiple dimensions such as quality, economy, and environmental protection, and the relationships between indicators are usually non-linear. Therefore, existing technologies also establish blending optimization models based on genetic algorithms, which can effectively capture the non-linear relationships between indicators and reduce coal blending costs by about 9%. However, this algorithm has insufficient convergence performance and cannot track changes in operating conditions in a timely manner. Existing technologies also include intelligent optimization models based on simulated annealing algorithms, which can satisfy multiple constraints and effectively save costs. However, these models are highly dependent on parameters, initial solutions, and neighborhood design, and are prone to getting trapped in local optima. The above results cannot meet the modern industrial demands for timely and precise coal blending. Existing technologies also employ non-dominated sorting genetic algorithms based on expert rules to reduce solutions that do not conform to common sense in the process, thereby reducing the search space. Additionally, improved particle swarm optimization algorithms incorporate adaptive constraint handling mechanisms to guide the optimization process towards convergence; however, they are prone to infeasible solutions when dealing with high-dimensional constraints and have weak anti-interference capabilities.

[0004] Therefore, to address the above issues, it is necessary to design an efficient coal blending ratio optimization framework based on accurate prediction of coke quality. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of the aforementioned background technology, namely the key engineering pain points in the coking industry such as large lag in obtaining coke quality results, low optimization efficiency, and poor constraint satisfaction. This invention proposes a solution using a BP neural network prediction model and a dream optimization algorithm to achieve efficient and accurate coal blending under complex constraints, aiming to form a closed-loop solution for the entire process.

[0006] To achieve the above objectives, this invention provides a coking coal blending optimization decision-making method based on a dream-inspired bionic mechanism, comprising the following steps:

[0007] S1, Obtain quality index data of coking coal blending, preprocess the data, and expand the quality index data to have industrial rationality;

[0008] S2. Construct a coke quality prediction model based on BP neural network to predict coke quality indicators as constraints for coal blending ratio optimization.

[0009] S3 uses the dream optimization algorithm to construct a coal blending optimization model. The model sets the fitness function by multi-objective weighted minimization and superimposing constraint violation penalties.

[0010] The dream optimization algorithm is divided into an initialization phase, an exploration phase, and a development phase.

[0011] During the initialization phase, each candidate coal blend ratio is determined by random generation combined with the upper and lower bounds of the search space;

[0012] During the exploration phase, a memory strategy, a forgetting and replenishment strategy, and a dream-sharing strategy were implemented. The memory strategy enabled each candidate scheme to learn from and approach the best scheme in its group. The forgetting and replenishment strategy performed a self-organizing update on the randomly selected coal blending ratio. A scheme was randomly selected, and the proportions of several types of coal were forgotten. Based on the core adjustment factor and referring to the proportions of the corresponding coal types in the current best scheme within the group, random perturbations were added, and the proportions of these coal types were redistributed. The dream-sharing strategy and the forgetting and replenishment strategy were executed in parallel, and each coal blending scheme could randomly obtain the proportions of discarded coal types from each other.

[0013] During the development phase, grouping is eliminated, and all coal blending schemes are updated locally based on the global optimal solution. The memory strategy and forgetting and replenishment strategy are retained to perform fine search and fine-tune the coal blending ratio so that the scheme converges to the global optimal solution.

[0014] Output the optimal coal blending ratio scheme and the corresponding fitness value.

[0015] Furthermore, S1 employs generative adversarial networks to construct a data generation model, expanding the quality indicator data to have industrial rationality; GAN includes a generator and a discriminator.

[0016] Furthermore, the blended coal is set to be composed of... It is a mixture of single types of coal. For the first The blending ratio of a single type of coal, The quality indicators for a single type of coal are based on the total water content. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness The quality indicators for blended coal adopt the total water content method. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness The quality index of coke is expressed as ash content. volatile matter , sulfur content Reactivity index Post-reaction strength shatter resistance abrasion resistance shatter resistance .

[0017] Furthermore, in S2, different coke quality indicators are used as inputs with blended coal quality indicators that are not entirely consistent.

[0018] Furthermore, the fitness function set in S3 is:

[0019]

[0020] in, For the weighted terms of the two objectives, To constrain violations and penalties.

[0021] Furthermore, in S3, the dual objectives of cost and sulfur content are standardized and weighted, and the cost objective function is:

[0022]

[0023] in, For the first The unit procurement cost of coal;

[0024] The sulfur content was obtained through the coke quality prediction model in S2, denoted as... ;

[0025] Standardize and weight the cost and sulfur content:

[0026]

[0027]

[0028]

[0029] in, , The upper and lower limits for cost normalization should cover the cost range for all coal type combinations. This represents the normalized upper limit of sulfur content. , For weights.

[0030] Furthermore, the constraints set in S3 include proportion conservation constraints, engineering practice constraints, and coke quality index constraints.

[0031] Furthermore, in S3, boundary overflow issues are handled by randomly resetting the out-of-bounds dimension for the proportion conservation constraint and engineering practice constraint.

[0032] This invention also provides a coking coal blending optimization decision system based on a dream-inspired bionic mechanism, which adopts the coking coal blending optimization decision system method based on a dream-inspired bionic mechanism as described above, including a data management module, a quality prediction module, and a coal blending optimization module.

[0033] The data management module is used to input the index data of coking coal blending, modify and clean the data, expand the data through GAN, and verify the expanded data.

[0034] The quality prediction module has a built-in coke quality prediction model, which predicts the coke quality corresponding to a new single coal blending ratio based on the selected coke quality index.

[0035] The coal blending optimization module has a built-in coal blending optimization model based on the dream optimization algorithm, which is used to generate a large number of coal blending ratio schemes, and at the same time evaluate and screen out the optimal coal blending ratio scheme.

[0036] The above-described solution of the present invention has the following beneficial effects:

[0037] The present invention provides a method and system for optimizing coking coal blending based on a dream-inspired bionic mechanism. It employs generative adversarial networks (GANs) to augment the quality index data of coking coal blending, compensating for the difficulty in model training caused by insufficient original data. A coke quality prediction model based on a backpropagation neural network (BPNN) is constructed to provide reliable quality feedback for coal blending optimization. Most importantly, inspired by dream-inspired behavior, a dream optimization algorithm is adopted. Through phased design and three major optimization strategies, it improves the efficiency of coal blending scheme optimization, balances global search and local refinement, effectively avoids getting trapped in local optima, and achieves multi-objective coal blending optimization decision-making under complex constraints.

[0038] In addition, the system provided by this invention integrates four core functions: data management, automatic data expansion, coke quality prediction, and multi-objective coal blending optimization. Combined with the method, it can achieve efficient and accurate coal blending.

[0039] Other beneficial effects of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0040] Figure 1 This is a flowchart of the steps of the present invention;

[0041] Figure 2 This is a comparison chart of the results obtained using the method of this invention with other methods;

[0042] Figure 3 This is a system block diagram of the present invention. Detailed Implementation

[0043] The following specific examples illustrate the implementation of this disclosure. Those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0044] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0045] It should also be noted that the illustrations provided in the following embodiments are merely schematic representations of the basic concept of this disclosure. The illustrations only show components relevant to this disclosure and are not drawn according to the actual number, shape, and size of components in implementation. In actual implementation, the type, quantity, and proportion of each component can be arbitrarily changed, and the component layout may be more complex. Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0046] like Figure 1As shown, embodiments of the present invention provide a method for optimizing coking coal blending based on a dream-inspired bionic mechanism. First, the data is preprocessed. Based on this, a coke quality prediction model based on a BP neural network is constructed. Then, combined with expert experience, a coal blending optimization model is built using the Dream Optimization Algorithm (DOA). These models are deeply coupled to achieve a reasonable optimization decision for coking coal blending. Specifically, this step includes the following sub-steps:

[0047] S1: Obtain quality index data of coking coal blending and preprocess the data.

[0048] In this embodiment, the quality index data of coking coal blending may include ash content, sulfur content, volatile matter, reactivity, post-reaction strength, crushing strength, and abrasion resistance. During data preprocessing, a generative adversarial network (GAN) is used to construct a data generation model, expanding the quality index data to have industrial applicability.

[0049] A GAN consists of a generator and a discriminator. The generator receives random noise as input and synthesizes data through a neural network. The discriminator receives real and synthesized data, judges them, outputs a confidence score, and learns to distinguish between real and fake data. This information is then fed back to the generator, allowing it to indirectly learn the distribution of real data and optimize its parameters to improve the realism of the synthesized data. The two work together iteratively, and eventually the generator can produce synthetic data that is indistinguishable from real data, while the discriminator struggles to distinguish between real and fake data, achieving a Nash equilibrium. Specifically, the network objective function can be expressed as:

[0050]

[0051] in, It is the actual data distribution. It is the distribution of input noise. It's fake data generated by the generator. It is the discriminator that judges the data The probability of judging based on real data.

[0052] S2, Construct a coke quality prediction model based on BP neural network.

[0053] It should be noted that coke quality indicators are required as constraints in the optimization of coking coal blending; therefore, accurate prediction of coke quality is essential. In this embodiment, the blended coal is considered to be composed of... It is a mixture of single types of coal. For the first The blending ratio of a single type of coal, The quality indicators for a single type of coal are based on the total water content. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness The quality indicators for blended coal are based on total water content. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness The quality index of coke is measured by ash content. volatile matter , sulfur content Reactivity index Post-reaction strength shatter resistance abrasion resistance shatter resistance For the adhesion index The synergistic or antagonistic effects between different coal types make the mixture... The value is usually not equal to the weighted average; it needs to be multiplied by a correction factor. Other indicators can be directly weighted, as follows:

[0054]

[0055] Based on expert experience, certain constraints are proposed for the quality indicators of blended coal:

[0056]

[0057] in, , , , , , , , These represent the maximum and minimum values ​​for moisture, ash, volatile matter, and sulfur content of the blended coal, respectively.

[0058] When constructing a prediction model based on a BP neural network, coke quality indicators are predicted by blending coal quality indicators. It should be noted that the correlation between each coke quality indicator and different blending coal quality indicators varies. To reduce the interference of irrelevant indicators on the neural network learning, blending coal quality indicators that are not entirely consistent are used as inputs for different coke quality indicators.

[0059] Specifically, the standardized feature vector of a single blended coal sample is: The characteristic values ​​of all blended coals after standardization are Number of input neurons conform to Then the output of the input layer is:

[0060]

[0061] The predicted coke quality, with practical physical meaning, is obtained through linear transformation, activation function, regularization, and denormalization. Then the first The output of each model is:

[0062]

[0063] in, Here are the weight matrices for hidden layers 1-3. These are the bias vectors for hidden layers 1 through 3. For the first Each model outputs layer weights. Indicates the first The bias vector of the output layer of the model. and It's the Dropout mask. To randomly turn off the proportion of neurons, To train the average of coke quality indicators, The standard deviation is denoted as .

[0064] During model training, all parameters (i.e., weight matrix and bias vector) in the BP neural network are continuously adjusted to make the model's predicted output as close as possible to the actual coke quality indicators. The training process includes forward propagation → loss calculation → backpropagation → parameter update, until the model's loss converges to a small value or the predetermined number of training epochs is reached. Finally, the trained coke quality prediction model is used to predict the coke quality corresponding to a new single-coal blending ratio, serving as a constraint for coking coal blending.

[0065] S3 uses the dream optimization algorithm to construct a coal blending optimization model and calculate the coal blending ratio.

[0066] In this embodiment, the dream optimization algorithm is used to construct a coal blending optimization model. This algorithm features fast convergence speed, strong optimization ability, and excellent adaptability to coal blending industrial scenarios. The core assumptions of the dream optimization algorithm include:

[0067] 1) The quality of dreams can be quantified by a fitness function that combines the cost of coal blending ratio, sulfur content, and constraint satisfaction.

[0068] 2) The onset of dreams is strongly correlated with existing memories, and new coal blending ratio candidate solutions need to be generated based on historically optimal ratios to avoid random invalid solutions;

[0069] 3) Humans partially forget their memories and supplement them with self-organized information. Therefore, some dimensions of the coal blending ratio (such as the proportion of certain types of coal) are randomly "forgotten" and new proportions are generated through self-organization.

[0070] 4) Individual memory abilities vary and are random, corresponding to the randomness of the "number of forgetting dimensions" of different "coal blending scheme candidate solutions", which enhances population diversity.

[0071] Based on this, the model is designed to achieve a weighted minimization of both objectives (cost and sulfur content), with additional penalties for constraint violations. This results in a fitness evaluation system where "the smaller the value, the better the solution." The mathematical expression for the objective function (fitness function) is as follows:

[0072]

[0073] in, For the weighted terms of the two objectives, To constrain violations and penalties.

[0074] For the dual objectives of total coal blending cost and coke sulfur content, the two differ greatly in magnitude, with the cost being on the order of magnitude of... The sulfur content is on the order of magnitude of Therefore, a weighted average is applied after standardization. Specifically, the cost objective function is:

[0075]

[0076] in, For the first The unit procurement cost of coal.

[0077] The sulfur content of coke needs to be obtained through the coke quality prediction model in S2, denoted as... .

[0078] Therefore, the cost and sulfur content are standardized and weighted as follows:

[0079]

[0080]

[0081]

[0082] in, , The upper and lower limits for cost normalization should cover the cost range for all coal type combinations. This represents the normalized upper limit of sulfur content. , For weights.

[0083] Meanwhile, coke quality must meet certain constraints. If the solution violates these constraints, the fitness value must be significantly increased to ensure that the optimization algorithm prioritizes feasible solutions. These constraints include the proportion conservation constraint (hard constraint) and coke quality index constraints (soft constraint). The proportion conservation constraint can be expressed as follows:

[0084]

[0085] In addition, considering engineering practice constraints, engineering practice shows that the proportion of a single type of coal does not exceed the maximum value. Not lower than the minimum value ,Right now:

[0086]

[0087] The constraints on coke quality indicators can be expressed using the following formulas:

[0088]

[0089] in, , , , , These are the maximum values ​​of the corresponding coke indicators. , , These are the minimum values ​​of the corresponding coke indicators. Therefore, the penalty term calculation formula can be set as follows:

[0090]

[0091]

[0092] in, The penalty coefficient is... For the first The degree of violation of the coke quality indicators, For the first Predicted values ​​of several coke quality indicators. , For the first The constraint threshold for a certain coke quality index. By setting a penalty term, if the cost of a certain coal blending scheme is very low, but the quality of the produced coke is unqualified, the score (fitness function) of that scheme will exceed the set value, and thus it will be eliminated in the optimization process.

[0093] In this embodiment, the dream optimization algorithm is divided into three parts: initialization stage, exploration stage, and development stage. Each stage uses mathematical formulas to transform "dream characteristics" into optimization logic.

[0094] During the initialization phase, each candidate coal blend ratio is determined by random generation combined with the upper and lower bounds of the search space.

[0095]

[0096] in, refer to 3D random vector, The number of types of coal. , For the upper and lower bounds of the search space, This represents the number of coal blending schemes.

[0097] Therefore, a large number of coal blending ratio schemes are randomly generated during the initialization phase to form an initial population. Each scheme is a possible coal blending ratio. The initial population matrix is ​​constructed using the multiple initial solutions obtained during the initialization phase.

[0098]

[0099] in, Indicates the first Among the candidate solutions, the th... The proportion of coal planted.

[0100] During the exploration phase, three core strategies were established: memory, forgetting recovery, and dream sharing. All initial coal blending schemes were first divided into 5 groups based on differences in memory ability. Based on three core strategies, each group performs the following operations:

[0101] When using memory strategies:

[0102]

[0103] in, For the first During the nth iteration The position of the coal blending ratio For the first During the nth iteration The optimal coal blending ratio for the group. Through a memory strategy, each candidate solution learns from and approaches the best solution in its group, ensuring that good characteristics are preserved and propagated.

[0104] When employing the forgetting-and-replenishment strategy, the randomly selected coal blending ratios are self-organized and updated. Specifically, a scheme is randomly selected, the blending ratios of several coal types are forgotten, and based on the core adjustment factor, the proportions of the corresponding coal types in the current optimal scheme within the group are referenced, and after adding random perturbations, the blending ratios of these coal types are redistributed. The specific formula is as follows:

[0105]

[0106]

[0107] in, Indicates from the range arrive A randomly selected integer. Indicates the number of iterations. Indicates the maximum number of iterations. This indicates the maximum number of iterations during the exploration phase. For the first During the nth iteration The first group of optimal proportions The proportion of coal planting Random selection A dimension of forgetting, The term serves as the core adjustment factor, gradually decreasing from 1 to 0. Therefore, it allows for thinking outside the box during iterative optimization, generating entirely new solutions.

[0108] The dream-sharing strategy and the forgetting-replenishment strategy are executed in parallel. Each coal blending scheme can randomly obtain the proportion of discarded coal types from each other, enhancing the diversity of the blending and avoiding getting trapped in local optima. The formula is as follows:

[0109]

[0110]

[0111] in, The index is a randomly selected individual. The dream-sharing strategy significantly increases population diversity and prevents premature convergence of all schemes, making it a key strategy for escaping local optima.

[0112] During the development phase, grouping is eliminated, and all coal blending schemes are updated locally based on the global optimal solution, retaining both the memory strategy and the forgetting supplementation strategy. The specific formulas are as follows:

[0113]

[0114]

[0115]

[0116] in, To achieve the globally optimal coal blending ratio, The development phase involves forgetting the dimension. Therefore, after identifying a promising region during the exploration phase, the development phase conducts a fine-tuning search within this region, adjusting the coal blending ratio to make the solution increasingly accurate and converge to the global optimum.

[0117] Furthermore, based on the hard constraints in the coal blending ratio optimization given above, such as the single coal type ratio limit and the total 100%, the boundary overflow problem is handled by randomly resetting the out-of-bounds dimension. The specific formula is as follows:

[0118]

[0119] Finally, after the number of iterations reached The iteration terminates when the optimal coal blending ratio scheme and its corresponding fitness value are output. Engineers can further evaluate the optimal coal blending ratio scheme and fitness value based on the actual situation to confirm the rationality of the coal blending ratio scheme, etc.

[0120] The following specific examples further illustrate the effectiveness of this invention. For the quality index data of coking coal blending, the GAN training parameters are set as follows: training batch size 64, iteration count 1000, using the Adam optimizer, and generator learning rate... Discriminator learning rate attenuation rate Data is generated with a constraint of ±5% of the original mean, and rounded to two decimal places. A random seed of 42 is used to ensure reproducibility. Samples exceeding the range by 0.1 times the feature span are discarded.

[0121] The constraints are set as follows: total water (8.0, 12.0), ash (8.5, 10.5), volatile matter (29.0, 33.0), sulfur (0.7, 1.0), fixed carbon (60.0, 65.0), caking index (30.0, 50.0), and fineness (85.0, 92.0).

[0122] The quality prediction model structure is as follows: input layer ( The layer consists of three layers: a first layer (128 neurons), a second layer (64 neurons), a third layer (32 neurons), and an output layer (1 neuron). The activation function is ReLU, and the dropout rate is 0.2.

[0123] The number of coke quality indicators is 8, and the number of neurons selected for each coke indicator is: ash content ( ), sulfur content ( ), volatile matter ( ), reactivity ( ), post-reaction strength ( ), shatter resistance ( (using GAN extension), wear resistance ( (using GAN expansion), shatter resistance ( (using GAN for augmentation).

[0124] Quality constraints for blended coal: , , , , , , , .

[0125] The loss function is mean squared error (MSE), the optimizer is Adam, the initial learning rate is 0.001, and a learning rate scheduler is used (the learning rate is multiplied by 0.5 when the validation loss stalls). The training batch size is 32, the maximum number of training epochs is 1000, and early stopping is set to 20.

[0126] Coke quality index constraints: , , , , , , , .

[0127] When using the dream optimization algorithm, the population size is 100, and the maximum number of iterations is... Explore the step size decay coefficient. Local development step size Memory retention coefficient The penalty coefficient for violating the constraint is 10000. The cost weight is 0.6, and the sulfur content weight is 0.4. , Probability of dream sharing , .

[0128] Figure 2 The results of various optimization algorithms were shown. The cost of optimization using the Dream Optimization Algorithm was reduced by about 25 yuan, or 2.7%, compared to the original cost, and the sulfur content was also reduced. Compared with other optimization algorithms, the Dream Optimization Algorithm can control the cost to the lowest level while reducing sulfur content, and performs excellently in the comprehensive problem of reducing cost and reducing sulfur content.

[0129] Based on the same inventive concept, this embodiment also provides a coking coal blending optimization decision-making system based on a dream-inspired bionic mechanism, such as... Figure 3 As shown, the system includes a data management module, a quality prediction module, and a coal blending optimization module. The data management module is used to input coking coal blending index data, modify and clean the data, and expand the data using GANs, as well as verify the expanded data. The quality prediction module has a built-in coke quality prediction model that predicts the coke quality corresponding to a new single-coal blending ratio based on selected coke quality indicators. The coal blending optimization module has a built-in coal blending optimization model based on the dream optimization algorithm, which can generate a large number of coal blending ratio schemes and simultaneously evaluate and select the optimal coal blending ratio scheme.

[0130] In addition, the system can include corresponding basic configurations, such as generating coal blending reports, setting cost control rules, and setting quality compliance rules. It can also be equipped with a machine network platform base, which can collect production indicator data and perform real-time monitoring.

[0131] Based on the same inventive concept, this embodiment also provides an apparatus, including: a memory for storing a computer program; and a processor for executing the computer program to implement the relevant steps of the coking coal blending optimization decision-making method based on dream bionic mechanism as described above.

[0132] The processor may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor can be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor may also include a main processor and coprocessors. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessors are low-power processors used to process data in the standby state. In some embodiments, the processor may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor may also include an Artificial Intelligence (AI) processor, which handles computational operations related to machine learning.

[0133] The memory may include one or more computer-readable storage media, which may be non-transitory. The memory may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory is used to store at least the following computer program, which, after being loaded and executed by the processor, is capable of implementing the aforementioned software method steps. In addition, the resources stored in the memory may also include operating systems and data, and the storage method may be temporary or permanent storage. The operating system may include Windows, Unix, Linux, etc.

[0134] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the relevant steps of the coking coal blending optimization decision-making method based on the dream-inspired bionic mechanism described above.

[0135] The system, apparatus, computer-readable storage medium, etc. provided in this embodiment have the same inventive concept and beneficial effects as the aforementioned method, and will not be repeated here.

[0136] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0137] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for optimizing coking coal blending based on a dream-inspired bionic mechanism, characterized in that, Includes the following steps: S1, Obtain quality index data of coking coal blending, preprocess the data, and expand the quality index data to have industrial rationality; S2. Construct a coke quality prediction model based on BP neural network to predict coke quality indicators as constraints for coal blending ratio optimization. S3 uses the dream optimization algorithm to construct a coal blending optimization model. The model sets the fitness function by multi-objective weighted minimization and superimposing constraint violation penalties. The dream optimization algorithm is divided into an initialization phase, an exploration phase, and a development phase. During the initialization phase, each candidate coal blend ratio is determined by random generation combined with the upper and lower bounds of the search space; During the exploration phase, a memory strategy, a forgetting and replenishment strategy, and a dream-sharing strategy were implemented. The memory strategy enabled each candidate scheme to learn from and approach the best scheme in its group. The forgetting and replenishment strategy performed a self-organizing update on the randomly selected coal blending ratio. A scheme was randomly selected, and the proportions of several types of coal were forgotten. Based on the core adjustment factor and referring to the proportions of the corresponding coal types in the current best scheme within the group, random perturbations were added, and the proportions of these coal types were redistributed. The dream-sharing strategy and the forgetting and replenishment strategy were executed in parallel, and each coal blending scheme could randomly obtain the proportions of discarded coal types from each other. During the development phase, grouping is eliminated, and all coal blending schemes are updated locally based on the global optimal solution. The memory strategy and forgetting and replenishment strategy are retained to perform fine search and fine-tune the coal blending ratio so that the scheme converges to the global optimal solution. Output the optimal coal blending ratio scheme and the corresponding fitness value.

2. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 1, characterized in that, S1 uses generative adversarial networks to build a data generation model and expands the quality indicator data with industrial rationality; GAN includes a generator and a discriminator.

3. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 1, characterized in that, Set the blended coal from It is a mixture of single types of coal. For the first The blending ratio of a single type of coal, The quality indicators for a single type of coal are based on the total water content. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness ; The quality indicators of blended coal adopt the total water content. Ash content volatile matter , sulfur content Fixed carbon Adhesion index Fineness The quality index of coke is expressed as ash content. volatile matter , sulfur content Reactivity index Post-reaction strength shatter resistance abrasion resistance shatter resistance .

4. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 3, characterized in that, In S2, different coke quality indicators are used as inputs with blended coal quality indicators that are not entirely consistent.

5. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 1, characterized in that, The fitness function set in S3 is: in, For the weighted terms of the two objectives, To constrain violations and penalties.

6. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 5, characterized in that, In S3, the dual objectives of cost and sulfur content are standardized and weighted, and the cost objective function is: in, For the first The unit procurement cost of coal; The sulfur content was obtained through the coke quality prediction model in S2, denoted as... ; Standardize and weight the cost and sulfur content: in, , The upper and lower limits for cost normalization should cover the cost range for all coal type combinations. This represents the normalized upper limit of sulfur content. , For weights.

7. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 1, characterized in that, The constraints set in S3 include proportion conservation constraints, engineering practice constraints, and coke quality index constraints.

8. The coking coal blending optimization decision-making method based on dream bionic mechanism according to claim 7, characterized in that, In S3, boundary overflow issues are handled by randomly resetting the out-of-bounds dimension for the proportion conservation constraint and engineering practice constraint.

9. A coking coal blending optimization decision-making system based on a dream-inspired bionic mechanism, employing the coking coal blending optimization decision-making system method based on a dream-inspired bionic mechanism as described in any one of claims 1-8, characterized in that, It includes a data management module, a quality prediction module, and a coal blending optimization module; The data management module is used to input the index data of coking coal blending, modify and clean the data, expand the data through GAN, and verify the expanded data. The quality prediction module has a built-in coke quality prediction model, which predicts the coke quality corresponding to a new single coal blending ratio based on the selected coke quality index. The coal blending optimization module has a built-in coal blending optimization model based on the dream optimization algorithm, which is used to generate a large number of coal blending ratio schemes, and at the same time evaluate and screen out the optimal coal blending ratio scheme.

Images