Coke strength prediction and coal blending method, equipment and medium based on coupling analysis of coal quality characteristics

Abstract

The present application relates to a kind of coke strength prediction and coal blending method, equipment and medium based on coal quality characteristics coupling analysis, including according to the first process performance index of multiple dimensions in historical coking coal data, historical coking coal ratio, historical coking coal quality and historical coke quality, determine the upper and lower limit value of each coking coal index index requirement;Each coking coal index is input into the target model pre-trained, and the predicted value of each target coking coal index output by the target model is obtained;According to the predicted value of target coking coal index, determine the influence coefficient of each target coking coal index on coke quality influence, and according to whether the predicted value of target coking coal index is within the upper and lower limit value of index requirement and the influence coefficient of target coking coal index on coke quality influence, the predicted value of target coking coal index is corrected, and the corrected target predicted value is obtained;According to the target predicted value of target coking coal index after correction, determine the coking coal ratio result of coking coal.

Description

Technical Field

[0001] This disclosure relates to the field of coking coal technology, and in particular to a method, equipment and medium for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics. Background Technology

[0002] The goal of a coal blending system is to ensure that all indicators of coke are within the required range, such as ash content, sulfur content, calorific value, and cold strength, while reducing blending costs and decreasing the amount of expensive coking coals such as coking coal and fat coal, thereby achieving economic benefits.

[0003] The core of a coal blending system is the blending method. For example, the coking coal blending method proposed in relevant scenarios, which controls the volatile matter content of blended coal to 28-29%, includes different weight percentages of gas-rich coal, coking coal, 1 / 3 coking coal, coking coal, lean coal, and gas coal. The volatile matter content of the gas-rich coal is 39-41%, the maximum fluidity is ≥80000 ddpm, and the solid-soft temperature range is ≥100℃. The volatile matter content of the coking coal is 29-32%, the maximum fluidity is ≥60000 ddpm, and the solid-soft temperature range is ≥100℃. The 1 / 3 coking coal is divided into 1 / 3 coking coal #1, 1 / 3 coking coal #2, and 1 / 3 coking coal #3 according to different volatile matter content and Y value. The coking coal is divided into coking coal #1, coking coal #2, and coking coal #3 according to different G value. The volatile matter content of the lean coal is 15-18%, and the G value is 30≤G≤60. The volatile matter content of the gas coal is ≥37%, and the Y value is ≤13. This method achieves the adjustment of blended coal caking properties by controlling the total ratio of gas coal and lean coal, and effectively controls the volatile matter content of the blended coal at 28-29%. This reduces blending costs and ensures coke quality while maintaining a lower required coke production. The method controls the volatile matter content of blended coal by controlling the proportions of gas coal, fat coal, lean coal, coking coal, gas-fat coal, and 1 / 3 coking coal. However, this method mainly involves indicators such as volatile matter, maximum fluidity, solid-soft temperature range, Y value, and G value, and fails to comprehensively consider factors such as coke strength and coal petrographic properties in small coke oven experiments with individual coal types, making it difficult to better characterize the quality of individual coking coals.

[0004] For example, the coking coal blending method involves obtaining N types of coal to be processed and the Kidney flowability benchmark value; based on the N types of coal to be processed, obtaining the capacity-inertia curve and the maximum vitrinite reflectance of each coal; based on the Kidney flowability benchmark value and the capacity-inertia curve of the coal to be processed, determining the capacity-inertia amount of each coal; based on the N types of coal to be processed and the capacity-inertia amount of each coal, determining the blending ratio of the candidate blended coal; and based on the blending ratio of the candidate blended coal and the maximum vitrinite reflectance of each coal to be processed, determining the average maximum vitrinite reflectance and the maximum vitrinite reflectance of the candidate blended coal. The standard deviation of reflectance; the coal blending ratio of the candidate blended coal whose average maximum reflectance of the vitrinite group and the standard deviation of the maximum reflectance of the vitrinite group meet the preset conditions is determined as the coal blending ratio of the first target blended coal; the above coal blending method can use fewer coal blending indicators to accurately blend coal, but the coal blending method based on Gibbs freeness, inertia curve and maximum reflectance of the vitrinite group fails to comprehensively consider other industrial indicators, petrographic indicators, ash composition indicators, etc. of coking coal, so its robustness is low and it is difficult to avoid the imbalance of other indicators of coking coal.

[0005] For example, the method for optimizing coking coal blending involves acquiring historical data; establishing a data chain sample group based on the ratio relationships between historical data and various coal types; establishing a predictive model for blended coal properties based on historical data and coal blending mechanisms; establishing a predictive model for coke properties using big data algorithms based on historical data and the coking mechanism model; and using a genetic algorithm to filter the data chain sample group through a fitness function and the predictive models for blended coal and coke properties to obtain the target coal blending ratio. Through the method and system disclosed herein, and in conjunction with actual production characteristics, the optimization of coal blending ratios, reduction of coal blending costs, and improvement of coke precision can be achieved. However, the genetic algorithm is a heuristic algorithm that solves approximate solutions. While it is highly applicable to solving complex problems, it is not suitable for situations with numerous constraints, such as hard constraints related to the production process. Summary of the Invention

[0006] The purpose of this invention is to provide a method, equipment, and medium for predicting coke strength and blending coal based on coupled analysis of coal quality characteristics. The aim is to accurately obtain a robust coal blending method, avoid the instability of coke quality caused by fluctuations in a certain index of coking coal, and at the same time have high precision, strong applicability and interpretability. It can also provide solutions that meet all constraints under the conditions of multiple coal blending process requirements and complex operating conditions, and can automatically adapt to the current production mode, so that the blending ratio is more in line with the current production.

[0007] To achieve the above objectives, a first aspect of this disclosure provides a method for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics, the method comprising:

[0008] Based on the primary process performance indicators, historical coking coal blending ratios, historical coking coal quality, and historical coke quality from multiple dimensions of historical coking coal data, the upper and lower limits of the indicator requirements for each coking coal indicator are determined.

[0009] The obtained coking coal indicators of the blended coal are input into the pre-trained target model. After selecting the target coking coal indicators from the coking coal indicator set for the blended coal, the target model outputs the predicted values ​​of each target coking coal indicator.

[0010] Based on the predicted values ​​of the target coking coal indicators, the influence coefficient of each target coking coal indicator on coke quality is determined. Then, based on whether the predicted values ​​of the target coking coal indicators are within the upper and lower limits of the indicator requirements and the influence coefficient of the target coking coal indicators on coke quality, the predicted values ​​of the target coking coal indicators are corrected to obtain the corrected target predicted values.

[0011] The coking coal blending ratio is determined based on the target predicted value after correction of the target coking coal index.

[0012] In one possible implementation, the step of correcting the predicted value of the target coking coal index based on whether the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement and the influence coefficient of the target coking coal index on coke quality, to obtain a corrected target predicted value, includes:

[0013] Based on the proportion of each coking coal in the blended coal and the corresponding unit price of coking coal, the cost per ton of blended coal is determined. Based on the predicted values ​​of each coking coal index, the constraint of coking coal usage, the constraint of the maximum number of coking coal used, and the constraint of coke index requirements, the predicted values ​​are optimized and corrected with the goal of minimizing the cost per ton of blended coal, and the corrected target predicted value is obtained.

[0014] The optimization and correction of the predicted value of the coking coal index requirement constraint includes: when the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement, it is determined that no reward or penalty will be imposed on the coking coal index.

[0015] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality, which is characterized by the influence coefficient of the target coking coal index on the coke quality, and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper and lower limits of the index requirement.

[0016] The total penalty value of the target coking coal index is determined based on the penalty value in the reward and penalty value, and the total reward value of the target coking coal index is determined based on the reward value in the reward and penalty value.

[0017] Based on the total reward value and the total penalty value corresponding to the target coking coal index, the predicted value of the target coking coal index is corrected by adding the total reward value and subtracting the total penalty value to the predicted value, thus obtaining the corrected target predicted value.

[0018] In one possible implementation, when the predicted value corresponding to the target coking coal index is not within the upper or lower bounds of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality (characterized by the influence coefficient of the target coking coal index on the coke quality) and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper or lower bounds of the index requirement, including:

[0019] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement).

[0020] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index).

[0021] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement).

[0022] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index).

[0023] In one possible implementation, determining the upper and lower bounds of the requirement for each coking coal indicator based on multiple dimensions of the first process performance index, historical coking coal blending ratio, historical coking coal quality, and historical coke quality from historical coking coal data includes:

[0024] Based on the first process performance indicators, historical coking coal blending ratio, historical coking coal quality and historical coke quality of multiple dimensions in the historical coking coal data, the target coking coal type in the preset single coking coal type of the historical coking coal in the historical coking coal data is determined.

[0025] After reclassifying the target coking coal type of the historical coking coal, based on the target coking coal type after the reclassification of the historical coking coal, determine whether multiple second process performance indicators of the historical coking coal are risk indicators.

[0026] Based on the target coking coal type and the risk indicators, the historical values ​​of each coking coal indicator in the coking coal are determined, and in response to the manual verification of the historical values ​​of each coking coal indicator, the upper and lower limits of the indicator requirements for each coking coal indicator are determined.

[0027] In one possible implementation, before determining whether multiple second process performance indicators of the coking coal are risk indicators based on the revised target coking coal type after the target coking coal type has been reclassified, the process includes:

[0028] If the target coking coal type is determined to be coking coal from the preset single coking coal types, then the small coke oven experimental heat intensity and coking-fat coal-rock interval of the coking coal are determined. If the small coke oven experimental heat intensity of the coking coal is less than the coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the coking coal threshold range, then the target coking coal type of the coking coal is downgraded to lean coking coal. If, after the target coking coal type of the coking coal is downgraded to lean coking coal, the small coke oven experimental heat intensity of the coking coal is less than the lean coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the lean coking coal threshold range, then the target coking coal type of the coking coal is further downgraded to lean coal.

[0029] If the target coking coal type is determined to be fat coal from a preset single coking coal type, and the volatile matter, small coke oven experimental heat intensity, and coking-fat coal-lithology range of the coking coal are determined, then if the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is less than the first fat coal volatile matter threshold, then the target coking coal type is changed to coking coal. If the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to 1 / 3 coking coal. If the small coke oven experimental heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, the coking-fat coal-lithology range is not within the 1 / 3 coking coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to gas coal.

[0030] If the target coking coal type is determined to be 1 / 3 coking coal from the preset single coking coal type, and the small coke oven test heat intensity and coking-fat coal-rock interval of the coking coal are determined, if the small coke oven test heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, and / or the coking-fat coal-rock interval does not meet the 1 / 3 coking coal threshold range, then the target coking coal type of the coking coal is downgraded to gas coal.

[0031] If the target coking coal type is determined to be lean coal from the preset single coking coal types, and the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal are determined, and if the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal all meet the threshold standards of the corresponding indicators for lean coal, then the target coking coal type of the coking coal is upgraded to lean coking coal.

[0032] If the target coking coal type is determined to be lean coking coal from the preset single coking coal types, and the small coke oven test heat intensity and coking coal-lithology range of the coking coal are determined, if the small coke oven test heat intensity and / or coal-lithology index of the coking coal does not meet the threshold standard of the corresponding index of lean coking coal, then the target coking coal type of the coking coal is downgraded to lean coal.

[0033] In one possible implementation, after reclassifying the target coking coal type, the determination of whether multiple second process performance indicators of the coking coal are risk indicators based on the reclassified target coking coal type includes:

[0034] After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the alkali metal content of the coking coal is higher than a preset first threshold, then the alkali metal content of the coking coal is determined to be a risk item indicator, and / or, if it is determined that the catalytic index of the coking coal is higher than a preset second threshold, then the catalytic index of the coking coal is determined to be a risk item indicator.

[0035] After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the volatile matter of the coking coal is higher than a preset third threshold, then the volatile matter of the coking coal is determined to be a risk indicator.

[0036] After performing a reassessment on the target coking coal type of the coking coal, if it is determined that the standard deviation of the coking coal is higher than a preset fourth threshold, the standard deviation of the coking coal will be determined as a risk indicator.

[0037] In one possible implementation, the first process performance index includes at least one of the following: adhesion index, dry ash-free volatile matter, maximum thickness of the adhesive layer, and maximum expansion.

[0038] The single type of coking coal includes at least one of the following: coking coal, fat coal, 1 / 3 coking coal, lean coking coal, gas coal, lean coal, semi-lean coal, anthracite, non-caking coal, weakly caking coal, medium caking coal, lean coal, and long-flame coal.

[0039] The target coking coal indicators include at least one of the following: post-reaction strength, reactivity index, crushing strength, and abrasion resistance.

[0040] In one possible implementation, determining the influence coefficient of each of the target coking coal indicators on coke quality based on the predicted values ​​of the target coking coal indicators includes:

[0041] Based on the predicted values ​​of the target coking coal index, the weighted summation is used to determine the blending coal index of the coking coal, and based on the blending coal index of the coking coal, the increase or decrease of the dry ash-free volatile matter of the coking coal is determined.

[0042] Determine the target range of increase or decrease in mass within the preset range of increase or decrease in mass of the dry ash-free volatile matter;

[0043] The influence coefficients of each core indicator corresponding to the target increase or decrease in quality are used as the influence coefficients of the target coking coal indicators on coke quality. Each of the target increase or decrease in quality influence ranges is assigned an influence coefficient for each core indicator.

[0044] A second aspect of this disclosure provides an electronic device, comprising:

[0045] A memory on which computer programs are stored;

[0046] A processor for executing the computer program in the memory to implement the steps of the method of any one of the first aspects.

[0047] A third aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in any of the first aspects.

[0048] This invention provides a method, equipment, and medium for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics. Compared with existing technologies, it has the following advantages:

[0049] This invention comprehensively utilizes multi-dimensional indicators of coking coal to conduct quality analysis. Because it considers multiple dimensions, the quality analysis of coking coal is comprehensive and reasonable. Based on the quantitative impact of fluctuations in various indicators of blended coal on coke quality, these quantitative impacts are integrated into coke quality prediction. This allows for reasonable and interpretable coke quality predictions even when the blending ratio is unbalanced relative to historical data. Furthermore, the upper and lower bounds of the distribution of important indicators of blended coal are extracted from historical data. Compared to methods relying solely on manual judgment of these bounds, this invention relies less on human expertise by introducing a blending optimization model with penalty and reward terms. This model exhibits strong robustness, requiring the upper and lower bounds of important blended coal indicators to fall within the range determined by historical data and expert experience, thus preventing fluctuations in a single indicator from causing abnormal coke quality. It also demonstrates strong flexibility, allowing for penalty or reward terms to be applied to coke indicators even when some indicators deviate from their bounds, enabling the system-calculated blending ratio to exceed the limits of manual blending and explore more optimization opportunities. This enables the accurate acquisition of robust coal blending methods, avoiding instability in coke quality caused by fluctuations in certain indicators of coking coal. It also boasts high precision, strong applicability and interpretability, and can provide solutions that satisfy all constraints even under complex coal blending processes and operating conditions. Furthermore, it can automatically adapt to the current production mode, making the blending ratio more suitable for the current production.

[0050] Other features and advantages of this disclosure will be described in detail in the following detailed description section. Attached Figure Description

[0051] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings:

[0052] Figure 1 This is a flowchart illustrating a method for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics, as shown in the embodiments of the specification.

[0053] Figure 2 This is a block diagram of a coke intensity prediction and coal blending device based on coal quality characteristic coupling analysis, as shown in the embodiments of the specification. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] The specific embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this disclosure.

[0056] This disclosure provides a method for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics. Figure 1 This is a flowchart illustrating a method for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics, according to an embodiment. Specifically, the method includes:

[0057] In step S11, based on the first process performance indicators, historical coking coal blending ratio, historical coking coal quality, and historical coke quality from multiple dimensions of historical coking coal data, the upper and lower limits of the indicator requirements for each coking coal indicator are determined.

[0058] The first process performance index is a series of quantitative parameters used in the coking coal field to measure the performance of coking coal under specific processes (such as coking). These parameters may include indicators of different dimensions such as caking properties and coking characteristics, reflecting the properties of coking coal during the coking process. Historical coking coal blending records data on the mixing of different types or properties of coking coal in a certain proportion to produce coke that meets quality requirements. Historical coking coal quality refers to the past quality status of coking coal, encompassing various quality parameters such as ash content, volatile matter, and sulfur content. These parameters affect the coking process and coke quality. Historical coke quality refers to the quality of coke produced in the past, usually measured by a series of quality indicators such as crushing strength and abrasion resistance. The upper and lower limits of the indicator requirements are reasonable ranges determined for each coking coal indicator based on historical data and actual production needs. These ranges define the minimum (lower limit) and maximum (upper limit) values ​​of the indicator; values ​​exceeding these ranges may adversely affect the coking process and coke quality.

[0059] In this embodiment of the disclosure, by analyzing historical coking coal data and using statistical methods or machine learning algorithms, the relationship between each coking coal indicator and historical coking coal blending ratio, historical coking coal quality, and historical coke quality is identified. Based on these relationships, a reasonable range for each coking coal indicator is determined, i.e., the upper and lower limits of the indicator requirements, while ensuring that the coke quality meets the requirements.

[0060] For example, suppose historical data shows that when the ash content is between 8% and 12%, the coke's crush resistance and abrasion resistance are at a good level. If the ash content is below 8%, it may increase coking costs; if it is above 12%, the coke quality will significantly decrease. Therefore, the required upper and lower limits for the ash content index are determined to be 8% to 12%.

[0061] In step S12, the obtained coking coal indicators of the blended coal are input into the pre-trained target model to obtain the predicted values ​​of each target coking coal indicator after the target model selects the target coking coal indicators from the coking coal indicator set for the blended coal.

[0062] Blended coal is a type of coal feedstock made by mixing various types or properties of coking coal in a certain proportion to meet the requirements of the coking process. Coking coal indicators are parameters used to describe the characteristics of different coking coals in the blended coal, such as ash content, volatile matter, and caking index. These indicators affect the coking process and coke quality.

[0063] In this embodiment, the target model is a trained mathematical model that learns the mapping relationship between the coking coal indicators of the blended coal and the predicted values ​​of the target coking coal indicators in the set of coking coal indicators. When new coking coal indicator data for the blended coal are obtained, they are input into the target model, and the model outputs the predicted values ​​of the target coking coal indicators according to its internal learning rules and parameters.

[0064] For example, suppose the target model is a neural network model. After being trained on a large amount of historical data, it learns how to predict the value of the caking index, a target coking coal indicator, when given input data on various coking coal types, volatile matter, proportions, and caking index. When new coking coal indicator data is input (coking coal type is lean coal, volatile matter is 25%, proportion is 13:87, and caking index is 10), the model outputs a predicted value of 75 for the caking index.

[0065] Among them, an AI algorithm module is constructed based on historical data on qualified coke quality and domain knowledge. The function of the AI ​​algorithm module is to provide coke quality predictions, including ash content, sulfur content, CSR, CRI, M40, and M10. The AI ​​algorithm module mainly consists of the following parts.

[0066] The selection of coking coal indicators refers to choosing the set of coking coal indicators that are most strongly correlated with given coke indicators. For coke ash content, the ash content of blended coal is used as the input indicator. For coke sulfur content, the sulfur content and sulfur conversion rate of blended coal are used as input indicators. For coke CSR, CRI, M40, and M10, possible relevant blended coal indicators include volatile matter, G value, Y value, experimental coke strength of single-type coal in small coke ovens, average Rmax, and distribution values ​​of coking coal within the coke-fat coal-lithology range. This can be achieved using methods based on statistical significance tests, such as stepwise regression and p-value screening; methods based on regularization, such as Lasso regression; or machine learning methods, such as random forests. By selecting coking coal indicators, assuming a given number of 5 indicators, the 5 indicators most relevant to coke quality can be selected.

[0067] Coke quality prediction refers to using the blended coal indicators calculated from selected coking coal indicators as input to output coke quality. This invention employs mixed integer programming to solve the blending ratio. Mixed integer programming is difficult to handle nonlinear and nonconvex problems, therefore, linear or quadratic models are needed to calculate coke quality. For simplicity, a linear model is used to calculate coke quality. Based on the selected coking coal indicators, derived features are generated for each coking coal, including square terms, radical terms, and cross terms. Then, a model based on the blending ratio with linear weighting is used, ensuring that the various indicators of the blended coal are linear functions of the blending ratio. Next, the coke indicators are required to be linear functions based on the blended coal indicators. Since a linear combination of linear functions is still a linear function, the coke indicators are linear functions based on the blending ratio. Under this premise, the coefficients are calculated using the least squares method, resulting in the coke quality prediction model.

[0068] In step S13, based on the predicted values ​​of the target coking coal indicators, the influence coefficient of each target coking coal indicator on the quality of coke is determined. Then, based on whether the predicted values ​​of the target coking coal indicators are within the upper and lower limits of the indicator requirements and the influence coefficient of the target coking coal indicators on the quality of coke, the predicted values ​​of the target coking coal indicators are corrected to obtain the corrected target predicted values.

[0069] In this embodiment, the influence of each target coking coal index on coke quality is analyzed using a target model to obtain an influence coefficient. Then, it is checked whether the predicted value of the target coking coal index is within the upper and lower bounds of the index requirements. If the predicted value exceeds the upper and lower bounds, the predicted value is corrected according to the magnitude of the influence coefficient. The larger the influence coefficient, the greater the influence of the index on coke quality, and the larger the possible correction.

[0070] For example: Suppose the target coking coal index is ash content, with a predicted value of 13%, while the required upper and lower limits for ash content are 8% - 12%. Meanwhile, model analysis shows that the influence coefficient of ash content on coke quality is 0.8. Since the predicted value exceeds the upper limit, based on the influence coefficient, the predicted ash content is corrected to 12% (assuming the correction rule is that when it exceeds the upper limit, the predicted value is adjusted to the upper limit proportionally according to the influence coefficient).

[0071] In step S14, the coking coal blending result of the coking coal is determined based on the target predicted value after correction of the target coking coal index.

[0072] The coking coal blending result is the mixing ratio of various coals in the coking coal, determined based on the revised target prediction value. This blending result aims to produce coke that meets quality requirements.

[0073] In this embodiment of the disclosure, based on the corrected target predicted value, combined with the requirements of the coking process and the target coke quality standard, the mixing ratio of various coals in the coking coal is determined by an optimization algorithm (such as linear programming, genetic algorithm, etc.), i.e., the coking coal blending result. This blending result should make the actual value of the target coking coal index as close as possible to the corrected predicted value, thereby producing coke that meets the quality requirements.

[0074] For example, suppose the corrected target prediction values ​​include an ash content of 11%, a volatile matter content of 26%, and a sulfur content of 0.7%. According to coking process requirements, coke needs to be produced with a crushing strength greater than 80% and an abrasion resistance greater than 6%. Through optimization algorithms, the coking coal blend is determined to be: 50% prime coking coal, 30% fat coal, and 20% 1 / 3 coking coal. This blend ensures that the ash, volatile matter, and sulfur contents in actual production are close to the corrected prediction values, while simultaneously meeting the coke quality requirements.

[0075] In this embodiment, a self-iterative module can also be configured to automatically update the upper and lower bounds of various important indicators of blended coal that meet the requirements for qualified coke, as well as the coke quality prediction model, based on the latest online data, including proportions, coking coal indicators, and coke indicators. This module can be updated every half month or month to ensure that the model learns the latest production data.

[0076] The above technical solution comprehensively utilizes multi-dimensional indicators of coking coal to conduct quality analysis. Because it considers multiple dimensions, this invention provides a more comprehensive and reasonable quality analysis of coking coal. Based on the quantitative impact of fluctuations in various indicators of blended coal on coke quality, these quantitative impacts are integrated into coke quality prediction. This allows for reasonable and interpretable coke quality predictions even when the blending ratio is unbalanced relative to historical data. Furthermore, the upper and lower bounds of the distribution of important indicators of blended coal are extracted from historical data. Compared to methods relying solely on manual judgment of these upper and lower bounds, this invention relies less on human expertise by introducing a blending optimization model with penalty and reward terms. This model exhibits strong robustness, requiring the upper and lower bounds of important blended coal indicators to fall within the range determined by historical data and expert experience, thus preventing abnormal coke quality caused by fluctuations in a single indicator. It also possesses strong flexibility, allowing for penalty or reward terms to be applied to coke indicators even when some indicators deviate from their limits, enabling the system-calculated blending ratio to exceed the limits of manual blending and explore more optimization opportunities. This enables the accurate acquisition of robust coal blending methods, avoiding instability in coke quality caused by fluctuations in certain indicators of coking coal. It also boasts high precision, strong applicability and interpretability, and can provide solutions that satisfy all constraints even under complex coal blending processes and operating conditions. Furthermore, it can automatically adapt to the current production mode, making the blending ratio more suitable for the current production.

[0077] In one possible implementation, in step S13, the step of correcting the predicted value of the target coking coal index based on whether the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement and the influence coefficient of the target coking coal index on coke quality, to obtain the corrected target predicted value, includes:

[0078] Based on the proportion of each coking coal in the blended coal and the corresponding unit price of coking coal, the cost per ton of blended coal is determined. Based on the predicted values ​​of each coking coal index, the constraint of coking coal usage, the constraint of the maximum number of coking coal used, and the constraint of coke index requirements, the predicted values ​​are optimized and corrected with the goal of minimizing the cost per ton of blended coal, and the corrected target predicted value is obtained.

[0079] The optimization objective is to minimize the cost per ton of blended coal. , Indicates the first The proportion of coking coal in the formula, Indicates the first The price of coking coal.

[0080] Constraints: (1) Constraints on the amount of coking coal used in a single project , , , Indicates the first Minimum usage of coking coal, the first The proportion of coking coal, the first (2) Other process requirements constraints, including the maximum number of coking coal units used. (3) Blending coal index requirements constraints. The blended coal indexes and upper and lower limits here can include manually given indexes and upper and lower limits, or the indexes and corresponding upper and lower limits given in step B. (4) Coke index requirements and constraints:

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] The optimization and correction of the predicted value of the coking coal index requirement constraint includes: when the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement, it is determined that no reward or penalty will be imposed on the coking coal index.

[0088] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality, which is characterized by the influence coefficient of the target coking coal index on the coke quality, and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper and lower limits of the index requirement.

[0089] In this embodiment of the disclosure, when the predicted value of the target coking coal index does not meet the required range (upper and lower bounds), the reward or penalty for production efficiency needs to be quantified based on the direction of the index's influence on coke quality (positive or negative correlation) and the degree of deviation of the predicted value from the upper and lower bounds. The specific steps are as follows:

[0090] Influence coefficient direction: Determine the influence coefficient (positive or negative value) of a certain coking coal index (such as ash content, volatile matter) on coke quality (such as strength, ash content) through historical data or experiments.

[0091] Deviation quantification: Calculate the difference between the predicted value and the upper and lower bounds, and take the smaller one as the benchmark for punishment or reward to avoid excessive rewards or punishments.

[0092] Product calculation: Multiply the influence coefficient by the degree of deviation to obtain the reward or penalty value.

[0093] For example: Suppose that the influence coefficient of a certain coking coal index (such as the Y value) on coke strength is 0.5 (positive correlation), the required range of the index is [16%, 20%], and the predicted value is 14%. Deviation: 16% - 14% = 2% (take the smaller one, since the predicted value exceeds the upper limit). Reward / penalty calculation: 0.5 × 2% = 1% (penalty value).

[0094] The total penalty value of the target coking coal index is determined based on the penalty value in the reward and penalty value, and the total reward value of the target coking coal index is determined based on the reward value in the reward and penalty value.

[0095] In this embodiment of the disclosure, the penalty value is accumulated by summing the penalty values ​​of all target coking coal indicators to obtain the total penalty value, which reflects the comprehensive losses caused by indicator deviations in production. The reward value is accumulated by summing all reward values ​​to obtain the total reward value, which reflects the comprehensive benefits brought about by indicator optimization in production.

[0096] For example: Suppose a batch of coking coal contains three indicators: Y value: predicted value 14% (penalty 1%). Volatile matter: predicted value 25% (indicator requirement [28%, 32%]), influence coefficient -0.3, deviation 28% - 25% = 3%, reward 0.3 × 3% = 0.9%. G value: predicted value 65% (indicator requirement [68%, 75%]), influence coefficient 0.2, deviation 68% - 65% = 3%, penalty -0.2 × 3% = -0.6%.

[0097] Total penalty value = 1% + 0.6%. Total reward value = 0.9%.

[0098] Based on the total reward value and the total penalty value corresponding to the target coking coal index, the predicted value of the target coking coal index is corrected by adding the total reward value and subtracting the total penalty value to the predicted value, thus obtaining the corrected target predicted value.

[0099] In this embodiment, the predicted value is adjusted by adding the total reward value to the predicted value and subtracting the total penalty value to reflect the gains or losses that may occur due to deviations from the indicators in actual production. The corrected predicted value is closer to the actual production results and is used to optimize production decisions.

[0100] For example: Suppose the original forecast value for a certain coking coal indicator (such as CSR) is 64%, the total reward value is 1.9%, and the total penalty value is -0.6%. The corrected forecast value = 64% + 0.9% - 1.6% = 63.3%.

[0101] Through the above steps, the predicted values ​​of coking coal indicators can be dynamically adjusted, and the reward and punishment mechanism in production can be quantified to help enterprises optimize raw material ratios, improve coke quality, and reduce production costs. For example, if the Y value is too low (e.g., 14%), the system corrects the predicted value with a penalty value (-1%), prompting a need to increase the Y value. If the volatile matter content is optimized (e.g., 25%), the system encourages maintaining or further optimizing this indicator with a reward value (+0.9%).

[0102] In one possible implementation, when the predicted value corresponding to the target coking coal index is not within the upper or lower bounds of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality (characterized by the influence coefficient of the target coking coal index on the coke quality) and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper or lower bounds of the index requirement, including:

[0103] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement).

[0104] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index).

[0105] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement).

[0106] If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index).

[0107] Taking CSR as an example, a predictive model for CSR was obtained based on the AI ​​algorithm module and using historical data on qualified coke quality. ,in, This represents the predicted value for CSR. R represents the proportion of each coking coal, and X represents the index of each coking coal. Based on R and X, the index values ​​of important blended coals can be calculated. If the index value is within the upper and lower bounds, it is considered to meet normal production requirements and no modification is made. If the index is outside the upper and lower bounds, based on the quantitative analysis of the impact of various blended coal indexes on coke quality, certain penalty or reward items are added to the predicted CSR value. The method is as follows:

[0108] Traverse execution: All blended coal indicators in the quantitative analysis function of the impact of various blended coal indicators on coke quality. If the indicator is within the upper and lower bounds, the penalty value is 0 and the reward value is 0.

[0109] If the influence coefficient of the indicator is positive and the value of the indicator is greater than the upper limit, then the reward value is: Reward value = Influence coefficient (Value of the indicator - Upper limit).

[0110] If the influence coefficient of the indicator is positive and the value of the indicator is less than the lower bound, then the following penalty is imposed: Penalty value = influence coefficient (lower bound - value of the indicator).

[0111] If the influence coefficient of this indicator is negative and the value of this indicator is greater than the upper limit, then the following penalty will be imposed: Penalty value = influence coefficient (value of this indicator - upper limit).

[0112] If the influence coefficient of the indicator is negative and the value of the indicator is less than the lower bound, then the reward value is: influence coefficient (lower bound - value of the indicator).

[0113] Add up the penalty values ​​of all important metrics to get the total penalty value. Add up the reward values ​​of all important metrics to get the total reward value. Therefore, the predicted value for the modified CSR is: .

[0114] visible, If the value is a linear function of the proportions of various coking coals, and both the reward and penalty values ​​are piecewise linear continuous functions of the proportions, then the predicted CSR value is also a piecewise linear continuous function of the proportions.

[0115] Similarly, we obtain the modified prediction values ​​for CRI, M40, and M10.

[0116] Modifying coke temperature and heat intensity prediction using penalty and reward values ​​has the following advantages: The prediction performance of AI models is greatly affected by the distribution of input data. Since historical data largely conforms to normal production data distribution, the prediction performance of AI models is guaranteed when the input data conforms to the distribution of normal production data. However, if the input data does not conform to the distribution of normal production data, the prediction performance of AI models lacks feasibility due to the lack of similar historical data. For example, when the usage of low-quality coal types such as lean coal and gas coal is relatively high compared to historical levels, AI models trained on normal production data may not achieve good prediction results. In this case, introducing a penalty term that incorporates domain knowledge and expert experience forces the model's prediction value to be lower, thus better aligning with human cognition. Similarly, when the usage of high-quality coal types such as coking coal and fat coal is relatively high compared to historical levels, introducing a reward value forces the model's prediction value to be higher.

[0117] Besides addressing the interference of data distribution on the model, the penalty and reward values ​​assigned to domain knowledge and expert experience also possess high interpretability. For a given ratio and coking coal index, it is possible to analyze how many reward items and how many penalty items there are, and provide explanations by combining domain knowledge.

[0118] Now, the optimization objective is a linear function, and all constraints are linear or piecewise linear, which meets the solution conditions for integer mixed programming. An optimizer can be used to obtain the exact solution that minimizes coal blending costs while satisfying all constraints.

[0119] In one possible implementation, determining the upper and lower bounds of the requirement for each coking coal indicator based on multiple dimensions of the first process performance index, historical coking coal blending ratio, historical coking coal quality, and historical coke quality from historical coking coal data includes:

[0120] Based on the first process performance indicators, historical coking coal blending ratio, historical coking coal quality and historical coke quality of multiple dimensions in the historical coking coal data, the target coking coal type in the preset single coking coal type of the historical coking coal in the historical coking coal data is determined.

[0121] In this embodiment of the disclosure, the quality analysis of coking coal is carried out by comprehensively considering various indicators, including industrial indicators (ash content, sulfur content, volatile matter, G value, Y value), coal-lithology range (average value of Rmax, standard deviation, distribution value of coal-lithology range), experimental coke strength of small coke ovens (CSR of single coal, etc.), ash composition (content of alkali metals such as iron oxide, calcium oxide, and magnesium oxide, catalytic index), active-inert ratio, and fluidity (solid-softening temperature range, Gibbs freeness).

[0122] The analysis is mainly conducted using the following methods: Based on national standards, combined with G value, volatile matter, Y value, and maximum expansion value b, single coking coal types are classified into coking coal, fat coal, 1 / 3 coking coal, lean coking coal, gas coal, lean coal, semi-lean coal, anthracite, non-caking coal, weakly caking coal, medium caking coal, lean coal, and long-flame coal.

[0123] In this embodiment, an expert experience module is constructed based on historical data and human experience. The historical data includes three parts: historical blending ratios, coking coal quality, and coke quality. After aligning these three parts, historical data indicating qualified coke quality is extracted. For each historical data point, the coking coal quality analysis function in step A is used to identify, modify, and determine the risk items of the coking coal corresponding to that historical data. The quantitative analysis function of the impact of various blending coal indicators on coke quality in step A is used to parse the values ​​of important blending coal indicators for that historical data point. For example, the values ​​of volatile matter, G value, single-coke small-scale coke oven thermal intensity, catalytic index, average Rmax, proportion of fat coal, proportion of lean coal, and proportion of gas coal are parsed. Thus, for each historical data point, there are corresponding values ​​of important blending coal indicators. Considering all historical data, the upper and lower bound distributions of important blending coal indicators that meet the requirements for qualified coke production are obtained. Human experience is then used to adjust the upper and lower bound distributions obtained from historical data or to add new indicators. This results in a range distribution of important blending coal indicators that meets the requirements for qualified coke production.

[0124] After reclassifying the target coking coal type of the historical coking coal, based on the target coking coal type after the reclassification of the historical coking coal, determine whether multiple second process performance indicators of the historical coking coal are risk indicators.

[0125] Based on the target coking coal type and the risk indicators, the historical values ​​of each coking coal indicator in the coking coal are determined, and in response to the manual verification of the historical values ​​of each coking coal indicator, the upper and lower limits of the indicator requirements for each coking coal indicator are determined.

[0126] In one possible implementation, before determining whether multiple second process performance indicators of the coking coal are risk indicators based on the revised target coking coal type after the target coking coal type has been reclassified, the process includes:

[0127] If the target coking coal type is determined to be coking coal from the preset single coking coal types, then the small coke oven experimental heat intensity and coking-fat coal-rock interval of the coking coal are determined. If the small coke oven experimental heat intensity of the coking coal is less than the coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the coking coal threshold range, then the target coking coal type of the coking coal is downgraded to lean coking coal. If, after the target coking coal type of the coking coal is downgraded to lean coking coal, the small coke oven experimental heat intensity of the coking coal is less than the lean coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the lean coking coal threshold range, then the target coking coal type of the coking coal is further downgraded to lean coal.

[0128] In this embodiment of the disclosure, considering the average thermal intensity and Rmax of the small coke oven test, as well as coal and petrographic indicators such as the coking coal-fat coal-rock range, if the thermal intensity of the small coke oven test is less than the defined threshold or the coal and petrographic range does not meet the defined threshold range, then the coking coal is downgraded to lean coking coal. If the threshold range required for the thermal intensity of the small coke oven test or the coal and petrographic range for lean coking coal is not met, then the coking coal is further downgraded to lean coal.

[0129] If the target coking coal type is determined to be fat coal from a preset single coking coal type, and the volatile matter, small coke oven experimental heat intensity, and coking-fat coal-lithology range of the coking coal are determined, then if the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is less than the first fat coal volatile matter threshold, then the target coking coal type is changed to coking coal. If the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to 1 / 3 coking coal. If the small coke oven experimental heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, the coking-fat coal-lithology range is not within the 1 / 3 coking coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to gas coal.

[0130] In this embodiment of the disclosure, for the classified coking coal, considering its volatile matter, small coke oven experimental heat intensity and average Rmax, and the coking coal-co ...

[0131] If the target coking coal type is determined to be 1 / 3 coking coal from the preset single coking coal type, and the small coke oven test heat intensity and coking-fat coal-rock interval of the coking coal are determined, if the small coke oven test heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, and / or the coking-fat coal-rock interval does not meet the 1 / 3 coking coal threshold range, then the target coking coal type of the coking coal is downgraded to gas coal.

[0132] In this embodiment of the disclosure, for the 1 / 3 coking coal, the coal and rock indicators such as the average value of its small coke oven experimental heat intensity and Rmax, and the coking-fat coal-rock interval are comprehensively considered. If its small coke oven experimental heat intensity or coal and rock indicators do not meet the set threshold standards, then the 1 / 3 coking coal is downgraded to gas coal.

[0133] If the target coking coal type is determined to be lean coal from the preset single coking coal types, and the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal are determined, and if the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal all meet the threshold standards of the corresponding indicators for lean coal, then the target coking coal type of the coking coal is upgraded to lean coking coal.

[0134] In this embodiment of the disclosure, if the coal and petrographic indicators such as the average value of the experimental heat intensity and Rmax of the small coke oven, the coking coal-fat coal-rock interval, and the standard deviation of the lean coal all meet the set threshold standards, then the lean coal is upgraded to lean coking coal.

[0135] If the target coking coal type is determined to be lean coking coal from the preset single coking coal types, and the small coke oven test heat intensity and coking coal-lithology range of the coking coal are determined, if the small coke oven test heat intensity and / or coal-lithology index of the coking coal does not meet the threshold standard of the corresponding index of lean coking coal, then the target coking coal type of the coking coal is downgraded to lean coal.

[0136] In this embodiment of the disclosure, if the experimental thermal intensity or coal petrographic index of the small coke oven does not meet the set threshold standard for the classified lean coking coal, then the lean coking coal is downgraded to lean coal.

[0137] In one possible implementation, after reclassifying the target coking coal type, the determination of whether multiple second process performance indicators of the coking coal are risk indicators based on the reclassified target coking coal type includes:

[0138] After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the alkali metal content of the coking coal is higher than a preset first threshold, then the alkali metal content of the coking coal is determined to be a risk item indicator, and / or, if it is determined that the catalytic index of the coking coal is higher than a preset second threshold, then the catalytic index of the coking coal is determined to be a risk item indicator.

[0139] In this embodiment of the disclosure, the ash content of various coking coals is considered. According to field knowledge, excessive alkali metal content or an excessively high catalytic index can reduce coke quality. For each coking coal, if its alkali metal content or catalytic index exceeds a pre-set threshold, then that coking coal is marked as a risk item.

[0140] After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the volatile matter of the coking coal is higher than a preset third threshold, then the volatile matter of the coking coal is determined to be a risk indicator.

[0141] In this embodiment of the disclosure, the volatile matter content of various coking coals is considered. According to field knowledge, coking coal with excessively high volatile matter content can cause quality instability. For each coking coal, if its volatile matter content exceeds a pre-set threshold, then that coking coal is marked as a risk item.

[0142] After performing a reassessment on the target coking coal type of the coking coal, if it is determined that the standard deviation of the coking coal is higher than a preset fourth threshold, the standard deviation of the coking coal will be determined as a risk indicator.

[0143] In this embodiment of the disclosure, the standard deviation of each coking coal is considered. Standard deviation is an indicator of the degree of blending; the more severe the blending, the larger the standard deviation. According to field knowledge, coking coal with excessively high blending levels can cause quality instability. Considering each coking coal, if its standard deviation exceeds a pre-set threshold, then that coking coal is marked as a risk item.

[0144] The above technical solution realizes the functions of determining and re-determining the type of coking coal, identifying coking coal with risk items in various dimensions, and realizing the quality analysis of coking coal under multi-dimensional coupling effect.

[0145] In one possible implementation, the first process performance index includes at least one of the following: adhesion index, dry ash-free volatile matter, maximum thickness of the adhesive layer, and maximum expansion.

[0146] The single type of coking coal includes at least one of the following: coking coal, fat coal, 1 / 3 coking coal, lean coking coal, gas coal, lean coal, semi-lean coal, anthracite, non-caking coal, weakly caking coal, medium caking coal, lean coal, and long-flame coal.

[0147] The target coking coal indicators include at least one of the following: post-reaction strength, reactivity index, crushing strength, and abrasion resistance.

[0148] In one possible implementation, determining the influence coefficient of each of the target coking coal indicators on coke quality based on the predicted values ​​of the target coking coal indicators includes:

[0149] Based on the predicted values ​​of the target coking coal index, the weighted summation is used to determine the blending coal index of the coking coal, and based on the blending coal index of the coking coal, the increase or decrease of the dry ash-free volatile matter of the coking coal is determined.

[0150] Determine the target range of increase or decrease in mass within the preset range of increase or decrease in mass of the dry ash-free volatile matter;

[0151] The influence coefficients of each core indicator corresponding to the target increase or decrease in quality are used as the influence coefficients of the target coking coal indicators on coke quality. Each of the target increase or decrease in quality influence ranges is assigned an influence coefficient for each core indicator.

[0152] In this embodiment, the various index values ​​of the blended coal are calculated using a weighted sum of the corresponding index values ​​of the blended coking coal. Based on domain knowledge, the influence of each index value of the blended coal on coke is analyzed to determine whether it is positive or negative, and the corresponding influence coefficient is determined. The influence coefficient can be defined as: if the index increases by one unit, how many units will the coke quality increase or decrease? For example, assuming the influence coefficient of volatile matter on CSR is -0.5, it means that if volatile matter increases by 1, then CSR will decrease by 0.5.

[0153] Based on domain knowledge, a set of important blending coal indicators and their influence coefficients on various coke indicators are defined. For example, the indicator set includes volatile matter, experimental thermal intensity of a single type of coal in a small coke oven, catalytic index, average Rmax, proportion of coking coal, lean coal, and gas coal. Taking volatile matter as an example, its influence coefficient is: -0.5 for CSR, 0.3 for CRI, 0.02 for M10, and -0.3 for M40. This means that if the Vdaf of the blending coal increases by 1, then correspondingly, CSR decreases by 0.5, CRI increases by 0.3, M10 increases by 0.02, and M40 decreases by 0.3.

[0154] This disclosure has been verified. Taking a coking plant as an example, six months of production data from the plant were collected, and a blending ratio was provided based on the latest coking coal resources. Comparative testing showed that the cost of the blended ratio was 1100 yuan / ton, compared to 1109 yuan / ton for manual blending, representing a cost reduction of 9 yuan / ton. The accuracy of coke prediction reached ≤2% for CSR, ≤1.5% for CRI, ≤0.05% for M10, ≤2% for M40, ≤0.3% for coke ash content, and ≤0.03% for coke sulfur content, meeting the accuracy requirements for industrial production forecasting.

[0155] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in the foregoing embodiments.

[0156] This disclosure also provides an electronic device, including:

[0157] A memory on which computer programs are stored;

[0158] A processor for executing the computer program in the memory to implement the steps of any of the methods described in the foregoing embodiments.

[0159] Figure 2 The crystal growth thickness real-time online measurement device 100 shown includes a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, for example, via a bus 1002. Optionally, the crystal growth thickness real-time online measurement device 100 may further include a communication component 1004, which can be used for data interaction between the device 100 and other devices, such as data transmission and / or data reception. It should be noted that in actual operation, the communication component 1004 is not limited to one, and the structure of this crystal growth thickness real-time online measurement device 100 does not constitute a limitation on the embodiments of this application.

[0160] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0161] Bus 1002 may include a pathway for transmitting information between the aforementioned components. Bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 1002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 2 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0162] The memory 1003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of carrying or storing program code and capable of being read by a computer, without limitation herein.

[0163] The memory 1003 is used to store program code for executing embodiments of this disclosure, and its execution is controlled by the processor 1001. The processor 1001 is used to execute the program code stored in the memory 1003 to implement the steps shown in the aforementioned embodiments of the coke intensity prediction and coal blending method based on coal quality characteristic coupling analysis.

[0164] This disclosure also provides a computer-readable storage medium storing program code. When the program code is executed by a processor, it can implement the steps and corresponding content of the aforementioned embodiment of the coke intensity prediction and coal blending method based on coal quality characteristic coupling analysis.

[0165] The preferred embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings. However, the present disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present disclosure, various changes, modifications, substitutions and variations can be made to these embodiments, and all such changes, modifications, substitutions and variations fall within the protection scope of the present disclosure.

[0166] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction, and such combinations should also be considered as part of this disclosure. To avoid unnecessary repetition, this disclosure will not further describe the various possible combinations. The technical scope of this application is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for predicting coke intensity and blending coal based on coupled analysis of coal quality characteristics, characterized in that, The term includes: Based on the primary process performance indicators, historical coking coal blending ratios, historical coking coal quality, and historical coke quality from multiple dimensions of historical coking coal data, the upper and lower limits of the indicator requirements for each coking coal indicator are determined. The obtained coking coal indicators of the blended coal are input into the pre-trained target model. After selecting the target coking coal indicators from the coking coal indicator set for the blended coal, the target model outputs the predicted values ​​of each target coking coal indicator. Based on the predicted values ​​of the target coking coal indicators, the influence coefficient of each target coking coal indicator on coke quality is determined. Then, based on whether the predicted values ​​of the target coking coal indicators are within the upper and lower limits of the indicator requirements and the influence coefficient of the target coking coal indicators on coke quality, the predicted values ​​of the target coking coal indicators are corrected to obtain the corrected target predicted values. The coking coal blending ratio is determined based on the target predicted value after correction of the target coking coal index.

2. The method according to claim 1, characterized in that, The step of correcting the predicted value of the target coking coal index based on whether the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement and the influence coefficient of the target coking coal index on coke quality, to obtain the corrected target predicted value, includes: Based on the proportion of each coking coal in the blended coal and the corresponding unit price of coking coal, the cost per ton of blended coal is determined. Based on the predicted values ​​of each coking coal index, the constraint of coking coal usage, the constraint of the maximum number of coking coal used, and the constraint of coke index requirements, the predicted values ​​are optimized and corrected with the goal of minimizing the cost per ton of blended coal, and the corrected target predicted value is obtained. The optimization and correction of the predicted value of the coking coal index requirement constraint includes: when the predicted value corresponding to the target coking coal index is within the upper and lower limits of the index requirement, it is determined that no reward or penalty will be imposed on the coking coal index. If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality, which is characterized by the influence coefficient of the target coking coal index on the coke quality, and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper and lower limits of the index requirement. The total penalty value of the target coking coal index is determined based on the penalty value in the reward and penalty value, and the total reward value of the target coking coal index is determined based on the reward value in the reward and penalty value. Based on the total reward value and the total penalty value corresponding to the target coking coal index, the predicted value of the target coking coal index is corrected by adding the total reward value and subtracting the total penalty value to the predicted value, thus obtaining the corrected target predicted value.

3. The method according to claim 2, characterized in that, When the predicted value corresponding to the target coking coal index is not within the upper or lower bounds of the index requirement, the reward or penalty value for the coking coal index is determined by multiplying the positive or negative relationship of the coking coal index's influence on the coke quality (characterized by the influence coefficient of the target coking coal index on the coke quality) and the smaller of the difference between the predicted value corresponding to the target coking coal index and the upper or lower bounds of the index requirement, including: If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement). If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is positive, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index). If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is greater than the upper limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: penalty value = influence coefficient × (predicted value of target coking coal index - upper limit of the upper and lower limits of the index requirement). If the predicted value corresponding to the target coking coal index is not within the upper and lower limits of the index requirement, and if the influence coefficient of the target coking coal index on the coke quality is negative, and the predicted value of the target coking coal index is less than the lower limit of the upper and lower limits of the index requirement, then the reward value for the coking coal index is determined as: reward value = influence coefficient × (lower limit of the upper and lower limits of the index requirement - predicted value of the target coking coal index).

4. The method according to claim 1, characterized in that, The method involves determining the upper and lower limits of the requirement for each coking coal indicator based on multiple dimensions of historical coking coal data, including first-process performance indicators, historical coking coal blending ratios, historical coking coal quality, and historical coke quality. Based on the first process performance indicators, historical coking coal blending ratio, historical coking coal quality and historical coke quality of multiple dimensions in the historical coking coal data, the target coking coal type in the preset single coking coal type of the historical coking coal in the historical coking coal data is determined. After reclassifying the target coking coal type of the historical coking coal, based on the target coking coal type after the reclassification of the historical coking coal, determine whether multiple second process performance indicators of the historical coking coal are risk indicators. Based on the target coking coal type and the risk indicators, the historical values ​​of each coking coal indicator in the coking coal are determined, and in response to the manual verification of the historical values ​​of each coking coal indicator, the upper and lower limits of the indicator requirements for each coking coal indicator are determined.

5. The method according to claim 4, characterized in that, Before determining whether multiple second process performance indicators of the coking coal are risk indicators based on the revised target coking coal type after the reclassification of the coking coal, the process includes: If the target coking coal type is determined to be coking coal from the preset single coking coal types, then the small coke oven experimental heat intensity and coking-fat coal-rock interval of the coking coal are determined. If the small coke oven experimental heat intensity of the coking coal is less than the coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the coking coal threshold range, then the target coking coal type of the coking coal is downgraded to lean coking coal. If, after the target coking coal type of the coking coal is downgraded to lean coking coal, the small coke oven experimental heat intensity of the coking coal is less than the lean coking coal heat intensity threshold, and / or the coking-fat coal-rock interval of the coking coal is not within the lean coking coal threshold range, then the target coking coal type of the coking coal is further downgraded to lean coal. If the target coking coal type is determined to be fat coal from a preset single coking coal type, and the volatile matter, small coke oven experimental heat intensity, and coking-fat coal-lithology range of the coking coal are determined, then if the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is less than the first fat coal volatile matter threshold, then the target coking coal type is changed to coking coal. If the small coke oven experimental heat intensity of the coking coal meets the fat coal heat intensity threshold, the coking-fat coal-lithology range is within the fat coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to 1 / 3 coking coal. If the small coke oven experimental heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, the coking-fat coal-lithology range is not within the 1 / 3 coking coal threshold range, and the volatile matter is greater than the second fat coal volatile matter threshold, then the target coking coal type is changed to gas coal. If the target coking coal type is determined to be 1 / 3 coking coal from the preset single coking coal type, and the small coke oven test heat intensity and coking-fat coal-rock interval of the coking coal are determined, if the small coke oven test heat intensity of the coking coal does not meet the 1 / 3 coking coal heat intensity threshold, and / or the coking-fat coal-rock interval does not meet the 1 / 3 coking coal threshold range, then the target coking coal type of the coking coal is downgraded to gas coal. If the target coking coal type is determined to be lean coal from the preset single coking coal types, and the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal are determined, and if the small coke oven experimental heat intensity, Rmax average value, coking-fat coal-lithology range, and standard deviation of the coking coal all meet the threshold standards of the corresponding indicators for lean coal, then the target coking coal type of the coking coal is upgraded to lean coking coal. If the target coking coal type is determined to be lean coking coal from the preset single coking coal types, and the small coke oven test heat intensity and coking coal-lithology range of the coking coal are determined, if the small coke oven test heat intensity and / or coal-lithology index of the coking coal does not meet the threshold standard of the corresponding index of lean coking coal, then the target coking coal type of the coking coal is downgraded to lean coal.

6. The method according to claim 5, characterized in that, After reclassifying the target coking coal type, the process involves determining whether multiple second process performance indicators of the coking coal are risk indicators based on the reclassified target coking coal type. These indicators include: After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the alkali metal content of the coking coal is higher than a preset first threshold, then the alkali metal content of the coking coal is determined to be a risk item indicator, and / or, if it is determined that the catalytic index of the coking coal is higher than a preset second threshold, then the catalytic index of the coking coal is determined to be a risk item indicator. After performing a re-judgment on the target coking coal type of the coking coal, if it is determined that the volatile matter of the coking coal is higher than a preset third threshold, then the volatile matter of the coking coal is determined to be a risk indicator. After performing a reassessment on the target coking coal type of the coking coal, if it is determined that the standard deviation of the coking coal is higher than a preset fourth threshold, the standard deviation of the coking coal will be determined as a risk indicator.

7. The method according to claim 5, characterized in that, The first process performance index includes at least one of the following: adhesion index, dry ash-free volatile matter, maximum thickness of adhesive layer, and maximum expansion. The single type of coking coal includes at least one of the following: coking coal, fat coal, 1 / 3 coking coal, lean coking coal, gas coal, lean coal, semi-lean coal, anthracite, non-caking coal, weakly caking coal, medium caking coal, lean coal, and long-flame coal. The target coking coal indicators include at least one of the following: post-reaction strength, reactivity index, crushing strength, and abrasion resistance.

8. The method according to any one of claims 1-7, characterized in that, The step of determining the influence coefficient of each target coking coal index on coke quality based on the predicted values ​​of the target coking coal index includes: Based on the predicted values ​​of the target coking coal index, the weighted summation is used to determine the blending coal index of the coking coal, and based on the blending coal index of the coking coal, the increase or decrease of the dry ash-free volatile matter of the coking coal is determined. Determine the target range of increase or decrease in mass within the preset range of increase or decrease in mass of the dry ash-free volatile matter; The influence coefficients of each core indicator corresponding to the target increase or decrease in quality are used as the influence coefficients of the target coking coal indicators on coke quality. Each of the target increase or decrease in quality influence ranges is assigned an influence coefficient for each core indicator.

9. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-8.

Images