Comprehensive evaluation method, system and device based on single coke quality and mixed coke use and readable storage medium

Abstract

The invention provides a comprehensive evaluation method, system and device based on single coke quality and mixed coke use and a readable storage medium, and the method comprises the steps: constructing a multi-layer progressive evaluation system from single coke quality evaluation to mixed coke performance prediction, and then to economical efficiency dynamic optimization; performing quantitative evaluation on the quality of the single coke through the basic evaluation layer to obtain a quality index; predicting a synergistic effect of various coke proportions through the mixing performance layer, and determining a mixing performance index; performing cost analysis on the proportioning scheme by combining an economic evaluation layer with price fluctuation to obtain a cost performance index; and a plurality of matching schemes are generated through the dynamic optimization layer, and an optimal solution meeting balance of quality, performance and cost is output.

Description

Technical Field

[0001] This invention relates to blast furnace ironmaking technology, and more particularly to a comprehensive evaluation method, system, apparatus, and readable storage medium based on the quality of single coke and the use of mixed coke. Background Technology

[0002] With the development of large-scale pulverized coal injection technology, the decline in iron ore quality, the increasing depletion of high-quality coking coal resources, and the increasingly severe environmental situation, the contradiction between higher requirements for coke quality and the gradual deterioration of coke quality has become increasingly prominent. Therefore, it is urgent to develop an evaluation system that can evaluate coke quality.

[0003] Previous patents disclosed evaluation systems based on the quality of single coke and evaluation systems based on the performance of mixed coke, the details of which are as follows:

[0004] The evaluation system based on single coke quality is as follows: First, 11 core quality indicators were selected through systematic analysis, including crush resistance (M40), abrasion resistance (M10), reactivity (CRI), post-reaction strength (CSR), sulfur content (S), ash content (A), volatile matter (V), high-temperature strength (M5 high-temperature, M10 high-temperature), equal melting rate (CRR25), and equal melting strength (CSR25). Grey relational analysis was used to quantify the influence weight of each indicator on the blast furnace fuel ratio. Second, data cleaning was performed to eliminate interference from production anomalies and confounding factors. The coke ratio, coke slurry ratio, and coal ratio were integrated to construct a comprehensive fuel ratio parameter, and the grade of the blast furnace feed and the strength of the sinter drum were introduced as correction variables. Then, based on dimensionless processing and correlation calculation, a differentiated scoring mechanism for positively and negatively correlated indicators was established, ultimately forming a weighted comprehensive evaluation model. To further extend the construction of a cost-effectiveness evaluation system, a cost-effectiveness index, a dynamic price prediction model, and a price difference analysis method were proposed. By quantifying the deviation between unit price quality benefits and market value, a multi-dimensional scientific basis for coke procurement decisions was provided.

[0005] An evaluation system based on the performance of blended coke is developed as follows: First, by analyzing the dissolution mechanism and skeletal support requirements of blast furnace coke, five core evaluation dimensions are established, including gradient dissolution strength matching degree (CSRGm), mixing reactivity index (CRIc), air permeability resistance coefficient (CPR), dynamic dissolution balance index (DSB), and particle size gradient coordination coefficient (GSCI). Parameter thresholds are validated using industrial practice data from companies such as Nippon Steel and China Baowu Steel Group. Second, cross-proportioning experiments are designed for the particle size distribution (coarse / medium / fine) and reactivity gradation (high / medium / low) of blended coke, constructing a multi-dimensional database including performance indicators and costs. Then, the economic efficiency index (CEQR) is introduced, and the contribution weights of indicators such as strength matching degree, reactivity, and air permeability are quantified through piecewise functions to establish a dynamic balance model between performance and cost. Finally, a dynamic cost-effectiveness prediction algorithm based on real-time price fluctuations is developed, combined with coke market price difference analysis and unit quality benefit assessment, to provide full-chain decision support for blast furnace coke procurement and proportion optimization.

[0006] The advantages of a single coke quality-based evaluation system are: ① Comprehensive indicators: It covers 11 core quality indicators (such as M40, CSR, sulfur content, etc.), evaluating coke quality from multiple dimensions from physical strength to chemical properties, with strong systematicity; ② Scientific quantification: It adopts grey relational analysis, data cleaning and variable correction to reduce interference from production anomalies, with objective weight allocation and high model credibility; ③ Cost-effectiveness extension: It combines dynamic price forecasting and quality benefit analysis to provide direct economic reference for procurement, with clear applicable scenarios (such as single coke procurement); ④ Simple operation: It relies on fixed quality parameters, and the data acquisition and calculation process is standardized, making it easy to implement quickly.

[0007] However, this method does not consider the synergistic effects of mixed coke in the blast furnace (such as particle size complementarity and reactivity gradient), which may underestimate its actual performance. It relies on external parameters such as the grade of the coke entering the furnace for correction; if the data fluctuates or deviates significantly, the model's stability may be affected. Dynamic price prediction based solely on a single type of coke is insufficient to address cost optimization needs in mixed-coal blends.

[0008] The advantages of the evaluation system based on the performance of blended coke are: ① Practical application orientation: Focusing on the blast furnace dissolution mechanism and skeleton support requirements, it designs indicators such as gradient dissolution intensity (CSRGm) and air permeability resistance (CPR), which are more in line with actual production; ② Dynamic balance capability: Through cross-proportioning experiments and multi-dimensional databases, it quantifies the impact of particle size and reactivity classification of blended coke on performance, supporting flexible proportioning optimization; ③ Economic integration: It introduces the CEQR index and real-time price fluctuation algorithm to achieve a dynamic trade-off between performance and cost, adapting to complex market environments; ④ Industrial verification: Combining practical data from Baowu Steel, Nippon Steel and other companies, the parameter thresholds are more universal and reliable.

[0009] This method also has some drawbacks, such as its inability to guide enterprises' single coke procurement, quality acceptance, and other needs, as well as the high cost of data acquisition and model maintenance in constructing a cross-experimental database of particle size / reactivity.

[0010] Given that both methods have their advantages and disadvantages, the key to evaluating the usability of coke lies in combining these two methods to construct a three-level evaluation system of "single coke quality evaluation - mixed coke performance prediction - economic decision-making for different proportions". Summary of the Invention

[0011] This invention provides a comprehensive evaluation method based on the quality of single coke and the use of blended coke, mainly including:

[0012] A multi-layered, progressive evaluation system is constructed, progressing from single coke quality assessment to blended coke performance prediction, and then to dynamic economic optimization. The basic evaluation layer quantifies the quality of single coke to obtain a quality index; the blending performance layer predicts the synergistic effects of various coke blend ratios to determine blending performance indicators; the economic evaluation layer combines price fluctuations to conduct cost analysis on blending schemes, obtaining a cost-effectiveness index; and the dynamic optimization layer generates multiple blending schemes, outputting the optimal solution that balances quality, performance, and cost.

[0013] Furthermore, the quantitative evaluation of single coke quality through the basic evaluation layer to obtain a quality index includes: classifying multiple quality indicators of coke into positive and negative indicators; standardizing the multiple quality indicators and calculating standardized values ​​using a dimensionless formula; calculating the weights of each indicator using a combination of grey relational analysis and entropy weighting to obtain a combined weight value; calculating an initial quality index based on the combined weight value and the standardized value; and introducing a fuel ratio correction coefficient to dynamically correct the initial quality index to obtain the final quality index, wherein the fuel ratio correction coefficient is calculated based on the grade of the sinter and the drum strength of the sinter.

[0014] Furthermore, the prediction of the synergistic effect of multiple coke ratios through the mixing performance layer to determine the mixing performance index includes: constructing multiple core parameters to characterize the performance characteristics of the mixed coke; constructing a mixing effect model based on the core parameters and calculating the mixing effect index; designing mixing experiments for different coke particle size distributions and reactivity classifications to obtain mixing scheme data; and predicting the permeability, strength balance, and solubility loss characteristics under different ratios based on the mixing scheme data and the mixing effect model to determine the mixing performance index.

[0015] Furthermore, the step of conducting cost analysis on the blending scheme by combining price fluctuations through an economic evaluation layer to obtain a cost-effectiveness index includes: acquiring real-time price data and logistics cost data of coke; introducing a price fluctuation term based on the real-time price data to calculate the full-cycle cost of the blending scheme; calculating the cost-effectiveness index based on the full-cycle cost and the quality index of the blended coke, wherein the cost-effectiveness index considers the impact of time decay factors on technological iterations; and ranking multiple blending schemes using the cost-effectiveness index to determine the scheme that balances cost and performance.

[0016] Furthermore, the step of generating multiple proportioning schemes through a dynamic optimization layer and outputting the optimal solution that satisfies the balance between quality, performance, and cost includes: constructing a multi-objective optimization model, setting an objective function and constraints, wherein the constraints include the total proportion, a lower limit on quality, and a permeability limit; acquiring the current operating parameters of the blast furnace and real-time price data as input; using an optimization algorithm to generate multiple solution sets, determining the optimal solution set that satisfies the objective function; displaying the optimal solution set through a visual interface, selecting the final proportioning scheme based on the performance and cost balance range; outputting the final proportioning scheme and recording implementation feedback data.

[0017] Furthermore, the construction of a multi-layered progressive evaluation system, from single coke quality assessment to mixed coke performance prediction, and then to economic dynamic optimization, includes: dividing the evaluation system into a basic evaluation layer, a mixed performance layer, an economic assessment layer, and a dynamic optimization layer; processing the raw data of single coke through the basic evaluation layer to generate quality assessment results; performing performance simulations on multiple coke combinations through the mixed performance layer to generate synergistic effect prediction results; conducting cost-benefit analysis on the blending scheme through the economic assessment layer to generate a dynamic cost model; and integrating the results of the aforementioned layers through the dynamic optimization layer to generate a blending recommendation that meets multi-objective constraints.

[0018] Furthermore, the process of generating multiple proportioning schemes through a dynamic optimization layer and outputting the optimal solution that satisfies the balance between quality, performance, and cost includes: setting the objective function of a multi-objective optimization model, which includes maximizing the mixed effect index, minimizing the cost-effectiveness index, and minimizing the quality index fluctuation; inputting blast furnace operating parameters and market price data according to the objective function and constraints; calculating and generating a solution set through an optimization algorithm, and filtering the solution set that meets the constraints; generating a visualization result based on the solution set and dividing it into different priority regions; selecting a proportioning scheme according to the priority regions, outputting the final instruction, and monitoring its implementation.

[0019] Furthermore, the present invention also provides a comprehensive evaluation system based on the quality of single coke and the use of blended coke, comprising:

[0020] The progressive system construction module, oh, that is, the construction of a multi-layer progressive evaluation system, from single coke quality assessment to mixed coke performance prediction, and then to economic dynamic optimization;

[0021] The basic evaluation module is used to quantitatively evaluate the quality of a single coke through the basic evaluation layer and obtain a quality index.

[0022] The index prediction module is used to predict the synergistic effect of multiple coke ratios through the mixing performance layer and determine the mixing performance index.

[0023] The cost analysis module is used to perform cost analysis on the allocation scheme by combining the economic evaluation layer with price fluctuations, and obtain the cost-effectiveness index.

[0024] The solution production module is used to generate multiple formulation schemes through a dynamic optimization layer, and output the optimal scheme that satisfies the balance between quality, performance and cost.

[0025] Furthermore, the present invention also provides a comprehensive evaluation device based on the quality of a single coke and the use of mixed cokes, the electronic device comprising:

[0026] processor;

[0027] A memory storing computer-readable instructions, which, when loaded and executed by the processor, implement any of the comprehensive evaluation methods based on the quality of a single coke and the use of mixed coke.

[0028] Furthermore, the present invention also provides a computer-readable storage medium storing program code that can be called by a processor to execute the comprehensive evaluation method based on single coke quality and the use of blended coke as described in any one of the claims.

[0029] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0030] This invention discloses a method for optimizing the coke blend ratio in blast furnaces. By acquiring basic coke data, the method classifies and standardizes the indicators, constructs a comprehensive weight model using grey relational analysis and entropy weighting, and calculates the quality index of a single coke. Combining particle size distribution and reactivity data, a three-dimensional blending space is constructed, and the mixing effect index is calculated. Taking cost factors into account, a full-cycle cost model is established, and the cost-effectiveness index is calculated. A non-dominated sorting genetic algorithm is used to generate the optimal solution set, and the final blending scheme is determined through a three-dimensional visualization interface. This invention inputs the blending scheme into an automatic batching system for implementation, monitors blast furnace operating parameters in real time, and triggers an adaptive correction mechanism based on deviations, achieving dynamic optimization of the blast furnace coke blend ratio and improving smelting efficiency and economic benefits. Attached Figure Description

[0031] Figure 1 This is a flowchart of a comprehensive evaluation method based on the quality of a single coke and the use of mixed coke according to the present invention;

[0032] Figure 2 This is a schematic diagram illustrating the research approach of the present invention for a comprehensive evaluation method based on the quality of single coke and the use of mixed coke;

[0033] Figure 3 Pareto front plots are generated for the optimization algorithm of this invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0035] like Figure 1 This embodiment of a comprehensive evaluation method based on the quality of single coke and the use of blended coke may specifically include:

[0036] Step S101: Obtain basic data on coke for blast furnace ironmaking. The basic data includes various physicochemical indicators of coke, furnace grade data, and sinter drum strength data. The basic data is then classified and processed, with the indicators divided into positive and negative indicators.

[0037] Through a data acquisition system, various physicochemical indicators of coke were extracted from blast furnace ironmaking databases, including 11 core quality indicators such as crush resistance, abrasion resistance, reactivity, post-reaction strength, sulfur content, ash content, and volatile matter, resulting in a comprehensive coke basic dataset. Based on the collected coke basic dataset, the furnace grade data and sinter drum strength data were integrated as correction variables to ensure the completeness and consistency of the comprehensive dataset. Data cleaning techniques were used to preprocess the comprehensive dataset, eliminating production outliers and confounding factors to obtain a cleaned, noise-free dataset. Dimensionless processing methods were used to standardize the indicators in the cleaned dataset, determining the numerical range of the indicators under a unified dimension. Based on the standardized numerical ranges, the indicators were classified into positive and negative indicators, resulting in a categorized set of indicators. Grey relational analysis was used to calculate the influence weights of the categorized indicator set, determining the specific degree of influence of each indicator on the blast furnace fuel ratio. A differentiated scoring mechanism was used to assign different scoring rules to the positive and negative indicators, obtaining the weighted scoring results for each indicator. Based on the weighted scoring results, a comprehensive evaluation model is constructed. Combining fuel ratio parameters such as coke ratio, coke-to-coal ratio, and coal ratio, a comprehensive evaluation score for coke quality is determined. Through a dynamic data update module, real-time collected data on furnace feed grade fluctuations and sinter drum intensity changes are integrated into the comprehensive evaluation model to obtain adjusted coke quality evaluation results.

[0038] Specifically, the data acquisition system extracts physicochemical indicators of coke from the blast furnace ironmaking database, such as the crushing strength M40 value of 88.5%, abrasion resistance M10 of 6.2%, reactivity CRI of 24.8%, and post-reaction strength CSR of 65.3%, forming a complete dataset containing 11 indicators. It integrates the furnace grade data, such as TFe content of 55.8%, and the sinter drum strength data, such as 78.5%, and uses a data consistency verification algorithm to ensure that the comprehensive dataset has no missing values. The data is cleaned using box plots, removing records with outliers of M40 below 75% or above 95%, retaining valid data with a standard deviation within ±2σ. The Min-Max standardization method is used to transform the ash content A from the original range of 10-15% to the 0-1 range, eliminating dimensional differences. Indicator types are categorized according to blast furnace process requirements; for example, CSR is set as a positive indicator with a threshold of ≥60%, and CRI is set as a negative indicator with a threshold of ≤30%. The grey relational analysis algorithm was used to calculate the weights of each indicator, with a resolution coefficient ρ = 0.5, yielding quantitative results such as a correlation degree of 0.72 for M40 and 0.68 for sulfur content (S). A piecewise scoring function was constructed, awarding 2 points for each 1% increase in the positive indicator CSR and 1.5 points for each 1% decrease in the negative indicator CRI. A linear weighted model was established based on the weights and scoring results, with the formula: Total Score = Σ(Weight × Score). A comprehensive evaluation value was output by combining fuel ratio parameters such as coke ratio (480 kg / t) and coal ratio (160 kg / t). The dynamic update module obtained the furnace feed grade fluctuation data ±0.5% through a real-time API interface and adjusted the model parameters using a moving average algorithm to ensure the timeliness of the evaluation results.

[0039] Step S102: Based on the basic data after classification, the various indicators are standardized using a dimensionless processing method to obtain a standardized indicator dataset, which is used for subsequent weight analysis.

[0040] Step S103: The standardized indicator dataset is weighted using grey relational analysis and entropy weighting. The combined weight values ​​of each indicator are determined by combining the grey relational analysis and information entropy results to form a comprehensive weight model.

[0041] By standardizing the coke quality indicator dataset, the influence of different indicator dimensions is eliminated, resulting in a unified standardized indicator dataset. Based on the standardized dataset, grey relational analysis is used to calculate the correlation between each indicator and the blast furnace fuel ratio, determining the initial influence weight values ​​for each indicator. Information entropy is calculated on the standardized indicator dataset to analyze the data dispersion of each indicator, obtaining the indicator weight distribution based on the entropy method. A combined weighting method is used to weight and fuse the two weights based on the initial weight values ​​obtained from grey relational analysis and the weight distribution obtained from the entropy method, determining the comprehensive weight value for each indicator. By combining the comprehensive weight value with the standardized indicator dataset, a weighted comprehensive evaluation model is constructed to obtain a quantitative score for coke quality. Based on the quantitative score, the grade of the blast furnace feed and the sinter drum strength are introduced as correction variables to dynamically adjust the score, determining the corrected comprehensive evaluation score. If the corrected comprehensive evaluation score shows data fluctuations or significant deviations, a data cleaning algorithm is used to eliminate interference from production anomalies and confounding factors, obtaining a stable evaluation result. Based on the stable evaluation result and real-time coke market price data, the unit price quality benefit is calculated to determine the cost-effectiveness index. By linking the cost-effectiveness index with the dynamic price forecasting model, multi-dimensional coke procurement decision-making basis is generated, and the final decision support parameters are obtained.

[0042] Specifically, the coke quality index dataset was standardized, and the Z-score method was used to normalize 11 core indicators, including M40, CSR, and sulfur content, eliminating dimensional differences and ensuring all indicators fell within the range of [-1, 1]. Based on the standardized dataset, a grey relational analysis algorithm was used, with the blast furnace fuel ratio as the reference sequence, to calculate the correlation degree of each indicator. For example, the correlation coefficient of CRI was 0.78, and CSR was 0.85, thus determining the preliminary weight distribution. The information entropy was calculated using the entropy weight method on the standardized data. If the entropy value of an indicator, such as ash content (A), was 0.92, its entropy weight was 0.08, reflecting a low degree of data dispersion. The grey relational weight (e.g., CRI weight 0.15) and the entropy weight (e.g., CRI weight 0.12) were linearly weighted and combined, with the grey relational weight accounting for 60% and the entropy weight accounting for 40%, resulting in a comprehensive CRI weight of 0.138. A weighted scoring model is constructed by multiplying the comprehensive weights by the standardized data matrix. For example, a coke sample might have an M40 score of 85 and a CSR score of 92. The furnace charge grade (e.g., 58%) and sinter drum strength (e.g., 72%) are introduced as correction factors. The scores are adjusted using multiple linear regression. If the corrected score fluctuation exceeds 5%, outliers are removed using the 3σ principle. Combined with real-time market prices (e.g., coke price of 2100 yuan / ton), the unit quality benefit (e.g., a cost increase of 50 yuan for every 1-point increase in score) is calculated to generate a cost-effectiveness index. Historical price trends are analyzed using a dynamic price forecasting model (e.g., the ARIMA model predicts next month's price fluctuation of ±3%), and this analysis, linked to the cost-effectiveness index, outputs the optimal procurement plan.

[0043] Step S104: Based on the comprehensive weighting model, the furnace grade data, and the sinter drum strength data, calculate the fuel ratio correction coefficient to obtain the dynamically corrected single coke quality index, which reflects the basic properties of coke.

[0044] Step S105: For the single coke quality index, a three-dimensional proportioning space is constructed by combining the particle size distribution and reactivity data of multiple cokes, and the performance parameters of the mixed coke are calculated using a mixed synergistic effect model to obtain the mixed effect index.

[0045] By combining grey relational analysis and entropy weighting, the weights of each core quality indicator in the single coke quality index are calculated to obtain the comprehensive score data for single coke quality. Based on the comprehensive score data for single coke quality, and combined with particle size distribution information and reactivity classification data of various cokes stored in the database, a three-dimensional proportioning space model is constructed, including coarse, medium, and fine particle sizes and high, medium, and low reactivity characteristics. Using the three-dimensional proportioning space model, characteristic parameters of various cokes under different particle size distributions and reactivity combinations are extracted to determine the initial set of proportioning schemes for blended coke. Through a mixing synergy effect model, the gradient dissolution intensity matching degree, mixing reactivity index, and air permeability resistance coefficient of each scheme in the initial proportioning scheme set are calculated to obtain the performance parameter matrix of blended coke. Based on the performance parameter matrix, the dynamic dissolution balance index and particle size gradient coordination coefficient are introduced to calculate the mixing effect index of each proportioning scheme to judge the overall performance of blended coke. If the mixing effect index is lower than a preset threshold, the proportion of each coke in the proportioning scheme is adjusted, and the performance parameter matrix is ​​recalculated to obtain an updated mixing effect index. If the mixing effect index reaches a preset threshold, a dynamic balance model between performance and cost is constructed by combining the economic index and real-time price fluctuation data to determine the optimal blending scheme. Using the optimal blending scheme, the fuel ratio correction coefficient and the charge grade correction variable are integrated to dynamically correct the performance parameters of the blended coke, yielding the final performance prediction result. Based on the final performance prediction result, an intelligent blending recommendation scheme incorporating quality, performance, and cost synergistic optimization is generated to determine the coke blending combination suitable for blast furnace production decisions.

[0046] Specifically, grey relational analysis was used to calculate the correlation between 11 core quality indicators, including crush resistance (M40), abrasion resistance (M10), and reactivity (CRI), and the blast furnace fuel ratio. The weight of each indicator was determined using the entropy weight method; for example, M40 had a weight of 0.15, and CRI had a weight of 0.18. A weighted sum was then used to obtain a single coke quality index. Based on this index, particle size distribution data (40%-60% coarse, 20%-30% medium, and 10%-20% fine) and reactivity classification (CRI ≤ 25% for low reactivity, and 25%-30% for medium reactivity) from the database were used to construct a three-dimensional proportioning space model. The coordinate axes were defined as particle size distribution, reactivity classification, and the single quality index. Spatial clustering algorithms were used to extract feature parameters for different combinations. For example, when the proportion of coarse, low-reactivity coke was 50%, 10 preliminary proportioning schemes were generated. The CSRGm (target value ≥ 65%), CRIc (target value ≤ 28%), and CPR (target value ≤ 1.5) of each scheme are calculated using a mixed synergistic effect model, and the performance parameter matrix is ​​output. The DSB index (threshold ≥ 0.8) and GSCI coefficient (threshold ≥ 0.7) are introduced to calculate the mixed effect index. If the index of a certain scheme is 0.75, which is lower than the threshold, the proportion of coarse coke is adjusted to 55% and recalculated. Once the index meets the target, the objective function min(cost) + max(performance) is established by combining the CEQR economic index (e.g., cost ≤ 2000 yuan / ton) and real-time price data to solve for the optimal blend ratio. The fuel ratio correction coefficient (e.g., coke ratio correction value ± 5 kg / t) and the furnace charge grade (e.g., TFe ≥ 58%) are integrated for dynamic correction, and the recommended scheme is finally output. For example, blend ratio A (coarse 50%, medium 30%, fine 20%) meets both performance and cost requirements.

[0047] Step S106: Using the mixed effect index, combined with real-time market price data and logistics cost data, a full-cycle cost model is constructed to calculate the total cost and cost-effectiveness index of different allocation schemes. The cost-effectiveness index is used for economic evaluation.

[0048] A mixing effect index is constructed by mixing performance layers. Core parameters such as gradient dissolution intensity matching degree, mixing reactivity index, air permeability resistance coefficient, dynamic dissolution balance index, and particle size gradient coordination coefficient are used to calculate the comprehensive value of the mixing effect index. Based on the comprehensive value of the mixing effect index and real-time market price data, unit price information of different coke suppliers is obtained to determine the market cost benchmark value of raw materials in each blending scheme. Using the market cost benchmark value, logistics cost data is integrated to calculate the additional costs in transportation, warehousing, and other aspects of different blending schemes, obtaining preliminary full-cycle cost data for each scheme. Based on the preliminary full-cycle cost data, a price fluctuation term formula is introduced for dynamic adjustment. Price fluctuation simulation within a time period is used to determine the cost correction value of each blending scheme at different time points. Using the cost correction value and the performance evaluation results of the mixing effect index, a cost-effectiveness index calculation model is constructed to obtain the initial value of the cost-effectiveness index for each blending scheme. Based on the initial value of the cost-effectiveness index, a time decay factor is introduced to correct for the impact of technological iteration, calculating the final value of the cost-effectiveness index adapted to the current operating conditions. If the final cost-effectiveness index value is lower than the preset threshold, a multi-objective optimization algorithm is invoked through the dynamic optimization layer to adjust the parameters of the proportioning scheme and recalculate the optimized cost-effectiveness index value. Based on the optimized cost-effectiveness index value and combined with the full-cycle cost model, the balance between quality, performance, and cost of different proportioning schemes is analyzed to determine the final economic evaluation result. Using the economic evaluation results, a priority ranking of the proportioning schemes is generated, and the data is output through a three-dimensional visualization interface to obtain a final list of schemes for decision-making reference.

[0049] Specifically, the mixing performance layer uses parameters such as gradient dissolution intensity matching degree CSRGm (threshold ≥ 1.8), mixing reactivity index CRIc (range 0.9-1.2), and air permeability resistance coefficient CPR (upper limit 0.82). The mixing effect index is calculated using the formula HSE = CSRGm × CRIc × CPR × DSB × GSCI × 10^5. If CSRGm = 2.1, CRIc = 1.05, and CPR = 0.75, then HSE = 2.1 × 1.05 × 0.75 × 1.1 × 0.95 × 10^5 ≈ 1.72 × 10^5. Real-time market price data is accessed from the coke supplier database. If supplier A's coke with sulfur content S = 0.6% has a unit price of 2300 yuan / ton, and supplier B's coke with S = 0.8% has a unit price of 2100 yuan / ton, then the market cost benchmark value is weighted and summed according to the mixing ratio weight. The logistics cost model includes additional transportation costs (e.g., railway freight of 0.12 yuan / ton·km) and warehousing costs (0.5 yuan / ton per day). If the blending scheme requires transportation of 500 kilometers and storage for 10 days, the additional cost is 500 × 0.12 + 10 × 0.5 = 65 yuan / ton. The price fluctuation term uses Pi(t) = 2300 × [1 + 0.2sin(0.017 × 30)] to simulate price fluctuations over a 30-day period, resulting in a corrected cost of 2300 × 1.096 ≈ 2520 yuan / ton. The initial value of the cost-effectiveness index (QSI) is calculated using QSI = HSE / (cost × Kadj). If HSE = 1.72 × 10^5, cost = 2520 yuan, and Kadj = 1.1 (62% grade entering the furnace), then QSI ≈ 62.3. The time decay factor λ = 0.98^t (t is the number of technology iterations in years). If t = 2 years, then the corrected QSI = 62.3 × 0.98^2 ≈ 59.8. When QSI < 60, the NSGA-Ⅲ algorithm adjusts the matching weights and re-optimizes with CPR ≤ 0.82 as a constraint, generating the Pareto front solution set. The 3D visualization interface maps the 3D coordinates of HSE, CEQR, and QSI. Schemes with HSE > 2.5 × 10^5 and CEQR < 3 in the red interval are automatically marked as high priority.

[0050] Step S107: Based on the cost-effectiveness index and the mixed effect index, set the constraints of the multi-objective optimization model, use the non-dominated sorting genetic algorithm to generate the optimal solution set, and obtain the Pareto front solution set of various matching schemes.

[0051] Eleven core quality indicators from the single coke quality evaluation system, including crush resistance strength M40, abrasion resistance strength M10, and reactivity CRI, were acquired. The weight of each indicator on the blast furnace fuel ratio was calculated using grey relational analysis to obtain a weighted comprehensive evaluation score. The entropy weight method was used to assign weights to five core dimensions of the mixed coke performance evaluation system, including gradient dissolution strength matching degree CSRGm and mixed reactivity index CRIc, to determine the contribution weight of each dimension. The evaluation data for both single and mixed cokes were cleaned to eliminate interference from production anomalies and confounding factors, resulting in a standardized comprehensive dataset of quality and performance. Based on the standardized dataset, a cost-effectiveness index was constructed. A piecewise function was used to quantify the unit price quality benefit and market value deviation to determine the cost-effectiveness scores for both single and mixed cokes. Particle size distribution and reactivity grading data of mixed cokes within the blast furnace were acquired. The mixing effect index was calculated using a cross-mixing experimental database to obtain a quantitative value for the synergistic effect of mixed cokes. If both the cost-effectiveness index and the mixed-effect index meet the preset thresholds, constraints are set through a multi-objective optimization model, including quality score, performance weight, and cost limit, to determine the mathematical expression of the optimization problem. A non-dominated sorting genetic algorithm is used to solve the multi-objective optimization model, generating Pareto front solution sets for various blending schemes through iterative calculations, and obtaining the optimal solution set. Based on the Pareto front solution set and combined with real-time coke market price fluctuation data, a dynamic price prediction algorithm is used to adjust the blending scheme and determine the economically optimal recommended blending ratio. The actual performance of the blending scheme is verified using blast furnace production practice data, and the influence weight of the fuel ratio is recalculated using grey relational analysis to obtain the evaluation results of the optimized blending scheme.

[0052] Specifically, data on 11 indicators, including crush resistance M40 (85%), abrasion resistance M10 (6.5%), and reactivity CRI (25%), were collected from the coke quality testing system. The correlation between each indicator and the blast furnace fuel ratio was calculated using grey relational analysis (e.g., M40 weight 0.18, CRI weight 0.12), generating a weighted comprehensive score (out of 100). Based on five dimensions of mixed coke data, including gradient dissolution strength CSRGm (≥65%) and permeability resistance CPR (≤120Pa), the entropy value of each dimension was calculated using the entropy weight method (e.g., CSRGm entropy value 0.21), and weights were determined after normalization (e.g., CPR weight 0.15). Outliers (e.g., sulfur content S > 0.7%) were removed from the original data, and Z-score standardization was applied to output a dataset with a mean of 0 and a variance of 1. Based on standardized data, a cost-effectiveness index is constructed (e.g., a price index of 1.2 corresponds to a quality score of 80). Market value deviation is quantified using a piecewise function (coefficient of 1.5 when the price range is ≤800 yuan, and coefficient of 1.2 when it is >800 yuan). Particle size distribution (40% coarse, 35% medium, 25% fine) and reactivity classification (20% high, 50% medium, 30% low) of mixed coke are extracted from the blast furnace database. Combined with cross-experimental data (e.g., a CSRGm increase of 8% for the coarse + high reactivity combination), a mixed effect index (e.g., a synergistic effect value of 1.15) is calculated. If the cost-effectiveness index is ≥1.0 and the mixed effect index is ≥1.1, multi-objective constraints are set (e.g., quality score ≥75, cost ≤900 yuan / ton), and an objective function (min cost, max performance) is established. The NSGA-II algorithm is used with a population size of 100 and 200 iterations. Pareto solutions are generated with a crossover probability of 0.9 and a mutation probability of 0.1 (e.g., Scheme A: cost 850 yuan / ton, performance score 88). Based on real-time price data (e.g., coke A price fluctuation ±50 yuan / ton), the solution set weights are dynamically adjusted (e.g., cost weight adjusted from 0.4 to 0.5), outputting the optimal blend ratio (coke A: 60%, coke B: 30%, coke C: 10%). Using blast furnace production data (e.g., actual fuel ratio reduced by 2 kg / t), the grey relational degree is recalculated (e.g., M40 weight corrected to 0.20), updating the blend ratio evaluation results.

[0053] Step S108: For the Pareto front solution set, the performance and cost distribution of different schemes are displayed through a three-dimensional visualization interface to determine the final proportioning scheme, which is used to guide the blast furnace coke proportioning decision.

[0054] The data from the dynamic optimization layer is processed using a multi-objective optimization algorithm to generate a Pareto front solution set containing multi-dimensional indicators such as quality, performance, and cost. Based on the generated Pareto front solution set, the quality index, performance parameters, and cost values ​​corresponding to each solution are extracted to form a structured multidimensional dataset. Using 3D visualization technology, the extracted multidimensional dataset is mapped onto a 3D coordinate system, constructing a graphical interface displaying the distribution of quality, performance, and cost. Cluster analysis is performed on the data points in the 3D visualization interface to identify the distribution patterns of quality, performance, and cost among different blending schemes. Based on the cluster analysis results, combined with the constraints of blast furnace production on fuel ratio and economic index, candidate blending schemes that meet preset thresholds are selected. If the number of candidate blending schemes is greater than a predetermined value, the quality, performance, and cost indicators of the candidate schemes are weighted using the entropy weight method to obtain a comprehensive score ranking; if the number is less than or equal to the predetermined value, the process proceeds directly to the next processing stage. Based on the comprehensive score ranking or the directly obtained candidate schemes, a dynamic price prediction model is invoked to calculate the cost deviation value of each scheme under the current market price fluctuations. By comparing and analyzing cost deviations with performance indicators, the final blending scheme with the minimum cost deviation is determined, ensuring that the quality and performance meet the blast furnace requirements. Data from the final blending scheme is acquired to generate a decision report containing specific blending ratios, quality indicators, and cost forecasts, which is then used for the automated adjustment of the blast furnace coke blending ratio.

[0055] Specifically, the data of the dynamic optimization layer is processed using a multi-objective optimization algorithm (such as NSGA-II). The optimization objectives are set as crush resistance (M40≥85%), post-reaction strength (CSR≥65%), and cost (≤3000 RMB / ton), generating a set of 20 Pareto optimal solutions. The quality index (e.g., weighted score 80-95 points), performance parameters (e.g., CRIc≤28%, CPR≤1.5kPa), and cost (2800-3200 RMB / ton) for each solution in the solution set are extracted to construct a structured database. Using OpenGL 3D rendering technology, the data is mapped to a coordinate system, with the X-axis representing the quality index, the Y-axis representing CPR performance, and the Z-axis representing cost, generating an interactive heatmap. K-means clustering (k=3) was used to group the 3D data points, identifying high-cost-performance clusters (quality > 90 points, cost < 2900 yuan), balanced clusters (quality 85-90 points, cost 2900-3100 yuan), and performance-priority clusters (CPR < 1.2 kPa). Combining blast furnace fuel ratio constraints (≤ 500 kg / t) and the economic index (CEQR ≥ 0.7), 12 candidate schemes were selected. When there were more than 5 candidate schemes, the entropy weight method was used to calculate weights (quality 40%, performance 35%, cost 25%) and rank the schemes; if there were fewer than 5, the weighting process was skipped. A dynamic price forecasting model (ARIMA time series analysis) was used, inputting current coking coal market price fluctuation data (±5%), to calculate the cost deviation value for each scheme (e.g., scheme A deviation +2.1%). The deviation value was compared with the performance threshold (CRIc ≤ 30%), and the scheme with the smallest deviation (-0.8%) that met the performance requirements was selected. Output the final scheme data (coke A: 60%, coke B: 30%, coke C: 10%), generate a JSON format report, including the proportion details, M40 (88%), CSR (68%), and predicted cost (2950 yuan / ton).

[0056] Step S109: According to the final proportioning scheme, input the data into the automatic batching system for implementation, monitor the blast furnace operating parameters in real time, and obtain the permeability index and fuel ratio data. The monitoring data is used for subsequent feedback correction.

[0057] Based on the final batching scheme, data is entered and parameters are initialized through an automatic batching system to obtain batching execution instructions. The automatic batching system executes these instructions, collecting real-time data on the permeability index and fuel ratio during blast furnace operation to obtain a real-time monitoring dataset. Based on this dataset, a data cleaning algorithm is used to remove outliers and noise interference, determining standardized blast furnace operating parameters. These standardized parameters, combined with preset permeability index and fuel ratio threshold ranges, are used to determine if the operating status meets expected standards. If the operating status exceeds the preset threshold range, a dynamic optimization algorithm generates adjustment suggestions, resulting in corrected batching parameters. Based on these corrected parameters, the automatic batching system's batching instructions are updated, obtaining a new batching execution scheme. Using this new scheme, the changing trends of the blast furnace operating parameters are monitored in real-time, yielding updated permeability index and fuel ratio data. Based on these updated permeability index and fuel ratio data, combined with performance indicators and cost data from a historical database, it is determined whether the correction effect achieves the optimization objective. If the correction effect does not achieve the optimization objective, a multi-objective optimization algorithm is used to regenerate the Pareto front solution set, determining a new batching scheme for subsequent adjustments.

[0058] Specifically, after receiving the final batching scheme, the automatic batching system stores coke batching parameters (e.g., M40 ≥ 85%, CSR ≥ 65%, sulfur content ≤ 0.7%) in an SQL database. It then initializes the batching equipment parameters via a PLC controller, generating batching instructions that include particle size distribution (coarse / medium / fine particle ratio 30% / 50% / 20%) and reactivity classification (high / medium / low reactivity coke ratio 15% / 60% / 25%). The system collects permeability index (target value 4.5-5.5 kPa) and fuel ratio (target value 480-520 kg / t) in real time through the blast furnace sensor network, forming a monitoring dataset with timestamps. The data is cleaned using the 3σ criterion, removing outliers exceeding the mean ± 3 standard deviations (e.g., permeability index < 3.0 or > 7.0 kPa), resulting in a standardized parameter matrix. The permeability index (4.8 kPa) and fuel ratio (495 kg / t) after cleaning are input into the threshold judgment module. If the permeability index is detected to be > 5.5 kPa and the fuel ratio is < 480 kg / t, the NSGA-II algorithm optimization module is triggered. With CSRGm ≥ 72%, CPR ≤ 2.8, and CEQR ≥ 1.2 as constraints, the system outputs proportion correction parameters (e.g., increasing coarse coke to 35%). The system updates the batching instructions according to the correction parameters, re-monitors and obtains the permeability index (5.2 kPa) and fuel ratio (490 kg / t), and combines them with the CRIc (target value 23-28) and DSB (target value 1.0-1.5) indicators in the historical database. If the CEQR is still lower than 1.2, the Pareto front solution set generation module is called to recalculate the optimal proportion using gray relational weights (M40 weight 0.18, CSR weight 0.22) and entropy weight method (sulfur weight 0.15).

[0059] Step S1010: By comparing and analyzing the monitoring data with the predicted values, if the deviation exceeds a preset threshold, an adaptive correction mechanism is triggered to recalculate the mixing ratio scheme and update the final mixing ratio scheme to adapt to changes in working conditions.

[0060] By collecting real-time monitoring data during blast furnace operation, including information such as coke quality indicators, fuel ratio parameters, and feed grade, the actual operating status data under the current conditions is obtained. Based on the real-time monitoring data and the established mixed coke performance prediction model, the predicted performance value under the current blending scheme is calculated, and a benchmark for comparison between the predicted value and the actual monitoring data is determined. By comparing the real-time monitoring data and the predicted performance value, the deviation between the two is calculated, and the specific value and distribution of the deviation are obtained. If the deviation value exceeds a preset threshold, an adaptive correction mechanism is triggered to extract the core influencing factors under the current operating conditions and determine the range of key parameters that need to be adjusted. Based on the extracted core influencing factors, a multi-objective optimization algorithm is invoked to regenerate the Pareto front solution set that includes a balance between quality, performance, and cost, resulting in a set of candidate blending schemes. By simulating and verifying the candidate blending schemes, and combining dynamic economic index and real-time price fluctuation data, the priority ranking of the optimal blending scheme is determined. If the simulation verification results show that the prediction deviation of the optimal blending scheme is still within the threshold range, the parameter thresholds and model weights in the database are updated to obtain the corrected model parameters. Based on the revised model parameters, the final blending scheme is recalculated. Combining indicators such as gradient dissolution intensity matching degree and permeability resistance coefficient, a blending result adapted to the current operating conditions is obtained. By feeding the data of the final blending scheme back to the blast furnace control system, the coke procurement and blending parameters are dynamically adjusted to determine the initial operating status for the next monitoring cycle.

[0061] Specifically, 11 quality indicators, including coke crushing strength M40 (82±2), reactivity CRI (26±1), and ash content A (12.5±0.3%), were collected through a blast furnace sensor network. These were combined with comprehensive fuel ratio parameters (coke ratio 320 kg / t, coal ratio 150 kg / t) to construct a real-time dataset. The dynamic dissolution balance index DSB (target value ≥1.8) and particle size gradient coordination coefficient GSCI (target value 0.9-1.1) from the mixed coke performance prediction model were used to calculate the deviation between the predicted and measured values ​​of CSRGm under the current ratio (78±2). When the deviation between the measured (29) and predicted (26) CRI values ​​exceeded the threshold (±2), a correction mechanism was triggered to extract key parameters, prioritizing the adjustment of highly correlated CSR (weight 0.35) and M40 high temperature (weight 0.28). The NSGA-II multi-objective optimization algorithm was invoked, with constraints of fuel ratio ≤315 kg / t and CEQR economic index ≥0.85, generating a Pareto solution set containing 5 blending schemes (CSR gradient 65-75%, coarse / medium / fine particle size ratio 4:3:3). Combined with real-time coke prices (a premium of 50 yuan / ton for coke with sulfur ≤0.7%), the entropy weight method was used to calculate the comprehensive score of each scheme, and the selected blending scheme (average CSR 70%, CRI 27.5) was chosen. The CRI threshold in the database was updated to 28±1.5, the gradient dissolution intensity matching degree CSRGm (76.5) was recalculated, and the blending instruction was output to the DCS system to adjust the coke bin feeding ratio (A coke proportion decreased from 40% to 35%).

[0062] Example 1:

[0063] like Figure 2 As shown, the method of this invention is specifically as follows:

[0064] 1. Basic Evaluation Layer (D1: Single Coke Quality Index)

[0065] 1) Standardization of indicators

[0066] 1> Indicator Classification:

[0067] Positive indicator (the higher the value, the better): M 40 CSR, etc.

[0068] Negative indicators (the smaller the value, the better): CRI, S, A, etc.

[0069] 2> Dimensionless formula:

[0070]

[0071] (Where i represents the coke sample, and j represents the index number)

[0072] 2) Weight Calculation (Grey-Entropy Weight Combination Method)

[0073] 1> Grey relational degree calculation:

[0074] Determine the reference sequence: Construct an ideal coke series X0 based on the industry best value.

[0075] Calculate the correlation coefficient:

[0076]

[0077] (ρ is set to 0.5)

[0078] Grey relational degree:

[0079] 2> Entropy weight method calculation:

[0080] Calculate information entropy: E j =―k∑(p ij lnp ij ),

[0081] in k = 1 / ln(n)

[0082] Entropy weight:

[0083] Combined weight: ω j =0.6α j +0.4β j

[0084] 3) Dynamic correction module

[0085] 1> Fuel ratio correction factor: K adj =1.2-0.05(TIO-62)+0.1(SST-78)

[0086] TIO: Furnace Grade (%)

[0087] SST: Sinter drum strength (%)

[0088] 2> Final Quality Index:

[0089] 2. Hybrid Performance Layer (D2: Hybrid Synergistic Effect Index)

[0090] 1) Core parameter construction

[0091]

[0092] 2) Mixed-effects model

[0093] HSE=CSRGm×CRIc×CPR×DSB×GSCI×10 5

[0094] 3) Hybrid verification experimental design

[0095] factor Horizontal setting Particle size distribution Coarse particles >25mm, medium particles 15-25mm, fine particles <15mm Reactivity classification High CRI (>30), Medium CRI (25-30), Low CRI (<25)

[0096] 3. Economic Evaluation Layer (D3: Full Life Cycle Cost Model)

[0097] 1) Dynamic cost calculation

[0098]

[0099] Price fluctuation term: P i (t)=P base ×[1+0.2sin(0.017t)] (t is the time period)

[0100] 2) Cost-effectiveness index

[0101]

[0102] QSI min The lowest QSI value in blended coke

[0103] • Time decay factor: reflects the impact of technological iteration

[0104] 4. Dynamic Optimization Layer (D4: Intelligent Proportioning Optimization)

[0105] 1) Multi-objective optimization model

[0106]

[0107] Restrictions:

[0108] • Total proportion: sum(W) i ) = 1,

[0109] • Quality lower limit: QSI min ≥0.4

[0110] • Breathability limitation: CPR≤0.8

[0111] 2) Solution Algorithm Flow

[0112] 1> Input real-time price data

[0113] 2> Load the current operating parameters of the blast furnace (blast temperature, oxygen enrichment rate, etc.)

[0114] 3> NSGA-Ⅲ algorithm generates Pareto front solution set

[0115] 4. Decision-makers can select their preferred solutions through a 3D visualization interface.

[0116] For example: Red zone: HSE > 2.5 and CEQR < 3

[0117] Yellow zone: Performance-cost balance zone

[0118] 5> Output the final formula

[0119] This architecture achieves closed-loop decision-making from raw material quality assessment to production plan optimization through a four-layer progressive analysis. Each module can run independently and supports data callback correction.

[0120] 1. Basic Evaluation Layer (D1) – Single Coke Quality Assessment

[0121] Objective: Calculate the QSI quality index for each type of coke.

[0122] Required data: Coke test report (including 11 indicators such as M40 and CRI), sinter drum strength, and furnace feed grade.

[0123] 1) Data standardization

[0124] • Categorize the 11 indicators by type (positive / negative)

[0125] Data standardization

[0126] 2) Weight Calculation

[0127]

[0128]

[0129] 3) Dynamic correction

[0130] Sinter drum strength (SST) = 80%, furnace feed grade (TIO) = 61.5%.

[0131] K adj =1.2 - 0.05*(61.5 - 62) + 0.1*(80 - 78) = 1.2 + 0.025 + 0.2 = 1.425

[0132] 4) Calculate QSI

[0133]

[0134]

[0135] QSI = 1.425 × (0.415 × 0.0694 + 0.615 × 0.0762 + ...) = 1.425 × 0.425 = 0.6056

[0136] The QSI values ​​of other cokes were calculated using the same method, and finally a coke quality ranking table (arranged in descending order) was generated.

[0137] coke name QSI Ranking A 0.6819 1 B 0.6560 2 C 0.6292 3 D 0.6131 4 E 0.6056 5 F 0.5821 6 G 0.5488 7 H 0.4547 8 I 0.4252 9

[0138] 2. Mixed Performance Layer (D2) – Prediction of Mixed Coke Effect

[0139] Objective: To predict the HSE mixed effect index for different proportions.

[0140] Required data: QSI value, particle size distribution, and CRI / CSR data of candidate coke.

[0141] 1) Data Input

[0142] Taking three types of coke as examples:

[0143]

[0144]

[0145] 2) Design proportions

[0146] According to particle size (coarse particles > 25 mm, medium particles 15-25 mm, fine particles < 15 mm), coke I and E belong to coarse particle size, while coke A belongs to medium particle size.

[0147] Based on reactivity (CRI≥35%, 25-35%, <25%), coke I and E are classified as high reactivity, while coke A is classified as low reactivity.

[0148] Based on the principles of particle size complementarity and reactivity balance, the following formulation scheme is designed:

[0149] plan Overall ratio (I:E:A) Cost (RMB) Option 1 4:4:2 1880 Option 2 3:3:4 1910 Option 3 2:2:6 1940 Option 4 1:1:8 1970

[0150] 3) Core parameter calculation

[0151] plan CSRGm CRIc CPR DSB GSCI 1 1.17 37.49 0.88 0.13 0.34 2 1.22 34.39 0.56 0.33 0.35 3 1.29 31.29 0.33 0.60 0.37 4 1.36 28.19 0.06 0.87 0.38

[0152] 4) Calculation of Mixed Effects Model (HSE)

[0153] plan Overall ratio (I:E:A) HSE 1 4:4:2 1.7061 2 3:3:4 2.7137 3 2:2:6 2.9571 4 1:1:8 0.7605

[0154] 3. Economic Evaluation Layer (D3) – Cost-Benefit Analysis

[0155] Objective: Calculate the total cost of ownership (TCO) and the CEQR cost-effectiveness index of the formulation scheme.

[0156] Required data: Real-time coke prices, logistics costs, and historical blast furnace efficiency data.

[0157] 1) Total Cost of Ownership (TCO) Calculation

[0158] The real-time price and logistics costs of coke are as follows:

[0159] coke Unit price (yuan / t) Logistics costs (RMB / ton) I 1800 200 E 1900 180 A 2000 150

[0160] Calculate the total cost of ownership (TCO)

[0161] Formula Overall ratio (I:E:A) TCO 1 4:4:2 2062 2 3:3:4 2084 3 2:2:6 2106 4 1:1:8 2128

[0162] 2) Calculate the Cost-Effectiveness Ratio (CEQR)

[0163] Find the minimum QSI value: A(0.6819), E(0.6056), I(0.4547) → QSI min =0.4547

[0164] Let t = 2024.2

[0165] Calculate CEQR:

[0166]

[0167]

[0168] 4. Dynamic Optimization Layer (D4) – Intelligent Proportioning Decision

[0169] Objective: To generate the Pareto optimal solution set and select the best solution.

[0170] Required tools: Optimization software pre-installed with the NSGA-Ⅲ algorithm (such as MATLAB / Python).

[0171] 1) Parameter Input

[0172] Enter the following in the software interface:

[0173]

[0174] 2) Run the optimization algorithm

[0175] Click the "Start Optimization" button to observe the iteration process and generate a Pareto front plot, such as... Figure 3 As shown.

[0176] • Red Zone Option: HSE > 2.5 and CEQR < 3 and σ(QSI) < 0.1 → Preferred Selection

[0177] Yellow Zone Scheme: HSE>1 and CEQR<5 and σ(QSI)<0.3 → Manual review required.

[0178] • Gray zone scheme: HSE≤1 or CEQR≥5 or σ(QSI)≥0.3 → Automatic exclusion

[0179] 3) Scheme selection and verification

[0180] Choose the option in the red area:

[0181] Key Indicators numerical values HSE 2.96 CEQR 3.42 σ(QSI) 0.094

[0182] 4) Output the final solution

[0183] Print the mixing instruction sheet:

[0184] [Final Formula]

[0185] Types of coke Proportion Coke I 15％ Coke E 25％ Coke A 60％

[0186] 5) Implementation and Feedback

[0187] 1> Input the mixing ratio into the automatic batching system

[0188] 2> Monitor key parameters for the next 8 hours:

[0189] • Breathability index change curve

[0190] • CO2 content in furnace top gas

[0191] • Static pressure gradient of the furnace body

[0192] 3. Record the actual fuel ratio every 4 hours and compare it with the predicted value:

[0193] time Predicted fuel ratio Actual fuel ratio deviation 8:00 505 508 +0.6％ 12:00 505 502 -0.6％

[0194] 4> If two consecutive deviations are greater than ±2%:

[0195] • Click the system's "Adaptive Correction" button

[0196] • Rerun the D4 layer optimization.

[0197] The evaluation method and system of this invention include: ① a basic evaluation layer (D1), which calculates the quality index of a single coke using the grey-entropy weight combination method and introduces a fuel ratio correction coefficient to achieve dynamic correction; ② a mixing performance layer (D2), which constructs a three-dimensional proportioning space of particle size gradient, reactivity gradient, and intensity matching degree, and establishes a prediction model for the synergistic effect of mixing; ③ an economic evaluation layer (D3), which combines dynamic price fluctuations and blast furnace efficiency losses to construct a full-cycle cost model and a cost-effectiveness index; and ④ a dynamic optimization layer (D4), which generates a Pareto front solution set through a multi-objective optimization algorithm to achieve intelligent proportioning recommendation. This invention solves the problem of the disconnect between the traditional single evaluation system and the performance prediction of mixed coke, and achieves synergistic optimization of quality, performance, and cost, which can significantly improve the scientific nature of blast furnace coke procurement and proportioning decisions.

[0198] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. A comprehensive evaluation method based on the quality of single coke and the use of blended coke, characterized in that, include: Construct a multi-layered, progressive evaluation system, from single coke quality assessment to mixed coke performance prediction, and then to dynamic economic optimization; The quality index is obtained by quantitatively evaluating the quality of a single coke through a basic evaluation layer. The synergistic effect of various coke ratios is predicted by using a mixed performance layer, and the mixed performance index is determined. By combining the economic evaluation layer with price fluctuations, a cost-effectiveness index is obtained by conducting a cost analysis of the allocation scheme. Multiple proportioning schemes are generated through a dynamic optimization layer, and the optimal solution that satisfies the balance between quality, performance and cost is output.

2. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The quantitative evaluation of single coke quality through a basic evaluation layer to obtain a quality index includes: The various quality indicators of coke are classified into positive and negative indicators. The aforementioned multiple quality indicators are standardized, and standardized values ​​are calculated using dimensionless formulas. The weights of each indicator are calculated by combining grey relational analysis and entropy weighting to obtain the combined weight value. Calculate the initial quality index based on the combined weight values ​​and standardized values; A fuel ratio correction factor is introduced to dynamically correct the initial quality index to obtain the final quality index, wherein the fuel ratio correction factor is calculated based on the furnace grade and the sinter drum strength.

3. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The method of predicting the synergistic effect of multiple coke ratios through a mixed performance layer and determining the mixed performance index includes: Several core parameters were constructed to characterize the performance features of the mixed coke; A mixed-effect model is constructed based on the core parameters, and the mixed-effect index is calculated. Experiments were designed to measure the particle size distribution and reactivity of different cokes, and data on the proportioning schemes were obtained. Based on the formulation data and the mixing effect model, the air permeability, strength balance and solubility characteristics under different formulation ratios are predicted, and the mixing performance index is determined.

4. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The cost-effectiveness index is obtained by combining the economic evaluation layer with price fluctuations to perform cost analysis on the allocation scheme, including: Obtain real-time price and logistics cost data for coke; Based on the real-time price data, a price fluctuation term is introduced to calculate the full-cycle cost of the allocation scheme; The cost-effectiveness index is calculated based on the full-cycle cost and the quality index of the mixed coke, wherein the cost-effectiveness index takes into account the impact of the time decay factor on technology iteration. The cost-performance ratio index is used to rank various ratio schemes to determine the scheme that balances cost and performance.

5. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The process of generating multiple proportioning schemes through a dynamic optimization layer and outputting the optimal solution that balances quality, performance, and cost includes: A multi-objective optimization model is constructed, and the objective function and constraints are set, wherein the constraints include the total proportion, the lower limit of quality, and the air permeability limit. The current operating parameters of the blast furnace and real-time price data are obtained as input. Multiple solution sets are generated using optimization algorithms, and the optimal solution set that satisfies the objective function is determined. The optimal solution set is displayed through a visual interface, and the final ratio scheme is selected based on the performance and cost balance zone. Output the final formulation and record the implementation feedback data.

6. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The aforementioned multi-layered, progressive evaluation system, from single coke quality assessment to mixed coke performance prediction, and then to dynamic economic optimization, includes: The evaluation system is divided into a basic evaluation layer, a mixed performance layer, an economic evaluation layer, and a dynamic optimization layer. The raw data of a single coke is processed through the basic evaluation layer to generate quality assessment results; The performance of various coke combinations is simulated using the hybrid performance layer to generate synergistic effect prediction results. The cost-benefit analysis of the proportioning scheme is performed through the economic evaluation layer to generate a dynamic cost model; The dynamic optimization layer integrates the results of the aforementioned layers to generate a recommended ratio that satisfies multiple objective constraints.

7. The comprehensive evaluation method based on the quality of single coke and the use of blended coke as described in claim 1, characterized in that, The process of generating multiple proportioning schemes through a dynamic optimization layer and outputting the optimal solution that balances quality, performance, and cost includes: Define the objective function of the multi-objective optimization model, which includes maximizing the mixed effect index, minimizing the cost-effectiveness index, and minimizing the quality index fluctuation; Based on the objective function and constraints, input the blast furnace operating parameters and market price data; A solution set is generated by optimizing the algorithm, and then the solution set that meets the aforementioned constraints is selected. Based on the solution set, a visualization result is generated, and regions with different priority are divided. Select the appropriate ratio scheme based on the priority region, output the final instruction, and monitor its implementation.

8. A comprehensive evaluation system based on the quality of single coke and the use of blended coke, characterized in that, include: The progressive system construction module, oh, that is, the construction of a multi-layer progressive evaluation system, from single coke quality assessment to mixed coke performance prediction, and then to economic dynamic optimization; The basic evaluation module is used to quantitatively evaluate the quality of a single coke through the basic evaluation layer and obtain a quality index. The index prediction module is used to predict the synergistic effect of multiple coke ratios through the mixing performance layer and determine the mixing performance index. The cost analysis module is used to perform cost analysis on the allocation scheme by combining the economic evaluation layer with price fluctuations, and obtain the cost-effectiveness index. The solution production module is used to generate multiple formulation schemes through a dynamic optimization layer, and output the optimal scheme that satisfies the balance between quality, performance and cost.

9. A comprehensive evaluation device based on the quality of single coke and the use of blended coke, characterized in that, The electronic device includes: processor; A memory storing computer-readable instructions, which, when loaded and executed by the processor, implement the comprehensive evaluation method based on the quality of a single coke and the use of blended coke as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code that can be invoked by a processor to execute the comprehensive evaluation method based on the quality of a single coke and the use of blended coke as described in any one of claims 1 to 7.

Images