Coking coal blending expert system

Abstract

The invention relates to the technical field of coking coal blending, and discloses a coking coal blending expert system, which comprises an incoming coal data management module, a silo management module, a coal preparation production management module, a coal quality evaluation system module, an optimized coal blending and quality prediction module, a coal coke resource library module and a coal coke corresponding module. The coal as received data management module is used for counting and managing the quality index, the coal feeding condition and the stock of coal as received, and realizing the whole-course traceability of the quality, the weight and the stock condition of the coal as received; the silo management module is used for monitoring the material level, the weight and the consumption condition of the silo in real time and tracking the inventory of coal types in the silo. By providing data management of coal as received, silo management and coal preparation production management, recording key information of each link and constructing a coal quality evaluation system, coal types are deeply evaluated from multiple dimensions of quality, price and the like, detailed and reliable data support can be provided for purchasing departments, purchasing decision making is facilitated, and the purchasing process is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of coking coal blending, in particular to a coking coal blending expert system. Background Art

[0002] The coking coal blending expert system is an intelligent system that integrates computer technology and artificial intelligence algorithms. It is mainly used for optimizing the coal blending process of coking enterprises. It can generate optimized coal blending plans based on raw coal resources, coke quality requirements and production process conditions, predict coke quality based on coal blending plans and coal quality data, and manage and maintain large amounts of data. By analyzing and discovering the laws of coal quality changes, it provides strong support for coking enterprises to improve coke quality and reduce production costs.

[0003] However, the coal quality evaluation system in the current system is imperfect and can only conduct simple evaluations of coal types from a single dimension. It cannot provide comprehensive and reliable data support for the procurement department, making procurement decisions unscientific and difficult to optimize the procurement process. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a coking coal blending expert system to solve the problems of lack of scientificity in procurement decisions and difficulty in optimizing procurement processes.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a coking coal blending expert system, including an incoming coal data management module, a silo management module, a coal preparation production management module, a coal quality evaluation system module, an optimized coal blending and quality prediction module, a coal coke resource library module, and a coal coke corresponding module.

[0006] The incoming coal data management module is used to collect statistics and manage the quality indicators, incoming coal conditions, and inventory of incoming coal, so as to achieve full traceability of the quality, weight, and inventory of incoming coal. The silo management module is used to monitor the material level, weight and consumption of the silo in real time, track the inventory of coal types in the silo and the expected number of days of use, and achieve adaptive use of coal resources by issuing early warnings on the expected number of days of use; The coal preparation production management module is used to manage the on-site coal supply plan, coal storage conditions in the coal yard, and the loading and unloading of coal in the silo. It generates a stacking guide for incoming coal based on the inventory of each type of coal and the coal procurement plan. It also monitors the silo loading and unloading coal production process in real time, and realizes the statistics and management of the incoming coal situation and the coal resource consumption in the coal yard. The coal quality evaluation system module is used to classify coal types by solid standards and application, build a coal source index evaluation system, derive quality index scores, price index scores, and cost-effectiveness scores for each type of coal, and generate a coal type evaluation report for reference by procurement and production departments. The optimized coal blending and quality prediction module is used to build a self-learning coke quality prediction model based on coking coal industrial analysis, petrographic analysis, and small coke oven test data, taking into account various constraints and combining the coal blending coking mechanism. It then uses a big data optimization algorithm to optimize and calculate multiple sets of coal blending plans. After the plans are submitted, verified, and reviewed, they are automatically sent to the silo basic automation PLC control system to execute coal preparation production; The coal and coke resource library module is used to obtain the quality of the blended coal, the quality of the coke, and the test data of the small coke oven, and to issue early warnings for quality indicators that exceed the standard values, thereby providing timely and accurate data support for coal blending; The coal-coke corresponding module is used to automatically track the measured quality indicators of blended coal and coke, incorporate them into the coal blending model sample library, compare the actual coke value with the predicted value, analyze the deviation between the two, and find out the possible coal types that may cause the coke to not meet expectations; The quality indicators of the incoming coal include: coking coal quality indicators and coking coal coal rock, wherein the coking coal quality indicators include industrial indicators, bonding indicators, fluidity indicators, and ash composition indicators. The industrial indicators include moisture, ash, sulfur, and volatile matter of the coking coal. The bonding indicators include G value, Y value, X value, CRI, CSR, and Oya expansion. The fluidity indicators include Kiel fluidity. The ash composition indicators include SiO2, Al2O3, Fe2O3, TiO2, and CaO. Coking coal rock includes coal rock reflectivity and optical structure; The coal arrival status includes real-time monitoring of the coal type, weight, shipping unit, shipping time, and number of departure sections of the coal arrival channels, and statistical management of the gross weight and net weight within a selected time range according to the coal arrival channel or coal type. The coal arrival channels include rail, truck, and ship transportation; The inventory of coal entering the factory is calculated using the following formula: Today's inventory = yesterday's inventory + today's inventory - today's consumption.

[0007] The inventory and expected usage days of the coal types in the tracking silo are calculated using the following formula: Silo coal inventory = (volume of coal in the silo at the end of coal feeding - volume of coal in the silo at the start of coal feeding) * bulk density of coal - silo coal consumption; Estimated number of days of use = silo coal inventory / silo coal consumption; The adaptive use of coal resources by issuing an early warning on the expected usage days includes: manually setting an early warning value for the minimum usage days, and issuing an early warning prompt for the type of coal in the silo when the expected usage days are lower than the early warning value for the minimum usage days.

[0008] The management of on-site coal supply plan, coal storage situation in coal yard, and coal loading / unloading situation in silo includes: The on-site coal plan obtains the coal resource's weighing order number, expected arrival time, weight, coal type, quality index, and freight information, and matches the weight with the quality index based on the coal type's weighing order number; Coal storage situation in the coal yard: Obtain the coal outbound / inbound quantity, inventory, storage period, and storage time. Set different storage periods for different types of coal according to the coalification characteristics of different types of coal to prevent individual batches of coal from being left unused for a long time and causing oxidation and deterioration. The silo loading / unloading coal blending status obtains the silo loading operation status including the loading batch number, coal type information, coal yard number, loading speed, loading time, and loading weighing value. The silo unloading operation status includes the coal type, coal yard number, blending ratio, unloading speed, unloading time, unloading weighing value, target coke oven number, and target coal silo.

[0009] The cost-effectiveness score of coal types is divided into quality index score and price index score, wherein the quality index score includes basic index score, coking index score and coal rock index score. The weight, basic score, benchmark quality index, change unit and plus / minus score of each item in the quality index score and price index score are set, and the cost-effectiveness score of the coal type is finally obtained through model calculation. The basic index score includes individual indicators of ash content, sulfur content and volatile matter, the coking index score includes individual indicators of reactivity and post-reaction strength, and the coal rock index score includes individual indicators of standard deviation and vitrinite reflectance, wherein: Quality index score = basic index score * basic index weight + coking index score * coking index weight + coal rock index score * coal rock index weight Price index score = price index score * price index weight Single indicator score = single basic score + (benchmark quality indicator - actual quality indicator) * (single item plus or minus score / change unit) Among them, the basic index score, coking index score and coal rock index score are the sum of the scores of each individual indicator.

[0010] The method comprehensively considers multiple constraints, combines the coal blending coking mechanism, constructs a self-learning coke quality prediction model, and then uses a big data optimization algorithm to optimize and calculate multiple groups of coal blending schemes, including: comprehensively considering the main coal system ratio, coking coal inventory, blending coal quality, and coke quality constraints, combining the multiple constraints with the self-learning coke quality prediction model, and using a big data optimization algorithm to optimize and calculate multiple groups of coal blending schemes, wherein the big data optimization algorithm includes one or more of a differential evolution algorithm, a genetic algorithm, and a simulated annealing algorithm.

[0011] The optimization calculation of multiple groups of coal blending schemes includes: coal blending schemes with optimal cost, optimal quality and optimal overall quality.

[0012] The coal and coke resource library module is used to obtain the quality of blended coal, quality of coke, and small coke oven test data; The quality of the blended coal includes moisture, ash, volatile matter, sulfur, fineness, bonding index, and maximum thickness of the colloidal layer; Coke quality includes moisture, ash, volatile matter, fixed carbon, sulfur, reactivity, post-reaction strength, crushing strength, and abrasion resistance; The test data of small coke oven include coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance.

[0013] The coal-coke corresponding module is used to automatically track the measured quality indicators of blended coal and coke, and incorporate them into the coal blending model sample library, including: associating the coking coal data corresponding to the coal blending plan, the measured blended coal data after blending, and the coke data obtained after production, and constructing data samples corresponding to the coking coal data, blended coal data, and coke quality data, wherein the coking coal data includes coal type, blending ratio, and coking coal indicators, the measured blended coal data includes ash content, volatile matter, sulfur content, bonding index, and maximum thickness of the gelatinous layer, and the coke data includes ash content, sulfur content, reactivity, post-reaction strength, crushing strength, and wear resistance.

[0014] The self-learning coke quality prediction model is constructed according to the following method: the coal blending model sample is input into the machine learning algorithm to obtain a prediction model for the coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance indicators. As the data samples continue to accumulate, the machine learning algorithm performs iterative model training and tuning, wherein the machine learning algorithm includes one or more of a decision tree, a support vector machine, a k-nearest neighbor, a random forest, and a transformer.

[0015] The present invention provides a coking coal blending expert system and its use method. It has the following beneficial effects: 1. In the present invention, by providing multi-dimensional data management functions for incoming coal data management, silo management, and coal preparation production management, key information of each link is recorded, and a coal quality evaluation system is constructed to conduct in-depth evaluation of coal types from multiple dimensions such as quality and price. This can provide detailed and reliable data support for the procurement department, facilitate auxiliary procurement decision-making, and optimize the procurement process.

[0016] 2. In the present invention, a self-learning coke quality prediction model is adopted. The model uses a machine learning algorithm to deeply mine a large number of coal blending model samples, so as to clearly understand the intrinsic relationship between various factors and coke quality. With the continuous accumulation of rich data samples, the model is continuously iterated and trained and tuned to adapt to the various changes in production, thereby achieving a more accurate coke quality prediction effect.

[0017] 3. In the present invention, a self-learning coke quality prediction model is adopted. The model uses a machine learning algorithm to clearly understand the complex relationship between various factors and coke quality. At the same time, combined with a big data optimization algorithm, it can optimize and calculate multiple groups of coal blending plans among many possibilities, providing rich decision-making references for coal blending experts. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is the system architecture diagram of a coking coal blending expert system. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0020] Please see the attached Figure 1 An embodiment of the present invention provides a coking coal blending expert system, which includes an incoming coal data management module, a silo management module, a coal preparation production management module, a coal quality evaluation system module, an optimized coal blending and quality prediction module, a coal coke resource library module, and a coal coke corresponding module.

[0021] Specifically, the incoming coal data management module enables full traceability of incoming coal to ensure production continuity and stability; the silo management module enables adaptive use of coal resources and optimizes resource allocation; the coal preparation production management module provides a basis for the rational arrangement of production processes and improves the efficiency and accuracy of coal preparation production; the coal quality evaluation system module provides a reference for procurement and production departments to reduce procurement costs and improve product quality; the coal blending optimization and quality prediction module accurately predicts coke quality and optimizes coal blending plans; the coal coke resource library module provides timely and accurate data support for coal blending to ensure product quality; the coal coke corresponding module optimizes the coal blending model, improves coal blending accuracy, and continuously improves product quality.

[0022] The incoming coal data management module is used to collect statistics and manage the quality indicators, incoming coal conditions, and inventory levels of incoming coal, thus achieving full traceability of the quality, weight, and inventory levels of incoming coal. The silo management module is used to monitor the material level, weight and consumption of the silo in real time, track the inventory of coal types in the silo and the expected number of days of use, and achieve adaptive use of coal resources by issuing early warnings on the expected number of days of use; The coal preparation production management module is used to manage the on-site coal supply plan, coal storage conditions in the coal yard, and coal loading / unloading in the silo. It generates instructions for stacking incoming coal based on the inventory of each type of coal and the coal procurement plan. It also monitors the silo loading / unloading coal production process in real time, and implements statistics and management of incoming coal and coal resource consumption in the coal yard. The coal quality evaluation system module is used to classify coal types by solid standards and application, build a coal source index evaluation system, derive quality index scores, price index scores, and cost-effectiveness scores for each coal type, and generate a coal type evaluation report for reference by procurement and production departments. The coal blending optimization and quality prediction module is used to build a self-learning coke quality prediction model based on coking coal industrial analysis, petrographic analysis, and small coke oven test data. It comprehensively considers various constraints and combines the coal blending coking mechanism to build a self-learning coke quality prediction model. It then uses a big data optimization algorithm to optimize and calculate multiple coal blending plans. After the plans are submitted, verified, and reviewed, they are automatically sent to the silo basic automation PLC control system to execute coal preparation production. The coal and coke resource library module is used to obtain the quality of blended coal, coke quality, and small coke oven test data, and to issue early warnings for quality indicators that exceed standard values, thereby providing timely and accurate data support for coal blending. The coal-coke matching module is used to automatically track the measured quality indicators of blended coal and coke, incorporate them into the coal blending model sample library, compare the actual coke values with the predicted values, analyze the deviation between the two, and identify the possible coal types that may cause the coke to not meet expectations; The quality indicators of incoming coal include: coking coal quality indicators and coking coal rock. The coking coal quality indicators include industrial indicators, caking indicators, fluidity indicators, and ash composition indicators. Industrial indicators include moisture, ash, sulfur, and volatile matter of coking coal. Caking indicators include G value, Y value, X value, CRI, CSR, and Oya expansion. Fluidity indicators include Kiel fluidity. Ash composition indicators include SiO2, Al2O3, Fe2O3, TiO2, and CaO. Coking coal rock includes coal rock reflectivity and optical structure; Coal delivery status includes real-time monitoring of coal type, weight, shipping unit, delivery time, and number of departure sections, as well as statistical management of gross and net weight within a selected time range based on coal delivery route or coal type. Coal delivery routes include rail, truck, and ship. The inventory of coal entering the factory is calculated using the following formula: Today's inventory = yesterday's inventory + today's inventory - today's consumption.

[0023] Specifically, the incoming coal data management module achieves full traceability of incoming coal through statistics and management of incoming coal quality indicators, coal supply conditions, and inventory levels. This module provides clear information on each batch of coal and can quickly identify the root cause when quality issues or inventory allocation anomalies occur, allowing for timely and effective measures to ensure production continuity and stability. The silo management module monitors silo material level, weight, and consumption in real time, tracks coal inventory and expected usage days, and uses early warnings to adapt coal resource usage. This allows for advance planning of coal resource usage, avoiding production stagnation due to insufficient inventory of a certain type of coal, preventing resource waste caused by excessive backlogs, and optimizing resource allocation. The coal preparation production management module manages on-site coal incoming plans, coal storage in the coal yard, and coal loading / unloading in the silo. It generates instructions for the stacking of incoming coal, monitors the production process in real time, and compiles relevant data. This provides a basis for enterprises to rationally arrange production processes, improves the efficiency and accuracy of coal preparation production, and reduces confusion and errors in production. The coal quality evaluation system module classifies coal types, constructs an evaluation system to derive quality, price, and cost-effectiveness scores, and generates reports. This provides a reference for procurement and production departments, helping them select cost-effective coal types and reduce procurement costs. It also helps production departments rationally arrange production based on coal quality and improve product quality. The coal blending optimization and quality prediction module builds a self-learning coke quality prediction model based on a variety of data. It uses a big data optimization algorithm to derive multiple coal blending plans and issue them for execution. It can accurately predict coke quality, optimize coal blending plans, improve the stability of coke quality, reduce production costs, and enhance the company's market competitiveness. The coal and coke resource library module obtains data on the quality of blended coal, coke, and small coke oven tests, provides early warnings for quality indicators that exceed standards, and provides timely and accurate data support for coal blending. This allows for timely detection of quality anomalies, adjustment of coal blending plans, and guaranteed production of products that meet quality standards. The coal-coke matching module automatically tracks the measured quality indicators of blended coal and coke, incorporates them into the sample library, and compares and analyzes deviations to identify coal types that affect coke quality. This helps to continuously optimize the coal blending model, improve coal blending accuracy, and continuously improve product quality. The quality indicators of incoming coal are divided into coking coal quality indicators and coking coal rock. The coking coal quality indicators include industrial, caking, fluidity, ash content indicators, as well as the coal rock reflectivity and optical structure of coking coal rock. This can comprehensively and accurately evaluate the characteristics of incoming coal. During the procurement process, these indicators can be used to screen out coal types that meet production needs, avoiding production problems caused by coal quality inconsistencies. During coal blending production, different coal types can be reasonably matched according to their various indicators, improving the stability and consistency of coke product quality and reducing production costs. By real-time monitoring of the coal type, weight, shipping unit, shipping time, and number of departure sections of incoming coal routes, and statistics of the gross weight and net weight within the selected time range by coal route or coal type, the dynamics of incoming coal can be monitored in real time, which is conducive to the rational arrangement of the receiving process, the preparation of the site and equipment in advance, and the improvement of the efficiency of receiving incoming coal. At the same time, the production plan can be flexibly adjusted according to the coal incoming situation to ensure production continuity and avoid production delays caused by untimely coal arrival or unclear information. In addition, different shipping units and coal routes can be evaluated to optimize procurement channels and ensure a stable supply of coal. By using the formula "today's inventory = yesterday's inventory + today's inventory - today's consumption" to calculate the inventory of coal entering the factory, a clear and accurate inventory calculation method is provided. It can grasp the inventory changes in real time, plan procurement plans in advance, avoid inventory backlogs occupying funds or insufficient inventory affecting production. In production scheduling, it can reasonably arrange coal blending production based on inventory data, optimize resource utilization, improve production operation efficiency, and ensure that production activities are carried out in an orderly manner.

[0024] Track the inventory of coal types in the silo and calculate the expected number of days of use using the following formula: Silo coal inventory = (volume of coal in the silo at the end of coal feeding - volume of coal in the silo at the start of coal feeding) * bulk density of coal - silo coal consumption; Estimated number of days of use = silo coal inventory / silo coal consumption; Adaptive use of coal resources can be achieved by issuing early warnings for the expected number of days of use, including manually setting an early warning value for the minimum number of days of use, and issuing early warning prompts for the type of coal in the silo when the expected number of days of use is lower than the early warning value for the minimum number of days of use.

[0025] Specifically, the formula "Silo coal inventory = (volume of coal in the silo at the end of coal feeding - volume of coal in the silo at the start of coal feeding) * coal type bulk density - silo coal type consumption" can accurately calculate the actual inventory of each coal type in the silo. This helps companies clearly understand the storage conditions of different coal types in the silo, thereby rationally planning the subsequent coal blending production process. When formulating production plans, accurate coal inventory can be used to avoid production interruptions caused by insufficient inventory of a certain coal type, ensuring production continuity. It can also prevent over-purchasing due to unclear inventory, reduce capital waste, and optimize resource allocation. Using the formula "estimated number of days of use = silo coal inventory / silo coal consumption," one can intuitively understand how long each type of coal can remain usable at the current consumption rate. This provides a time dimension reference for production scheduling and enables early estimation of production progress. When a certain type of coal is expected to have a short number of days of use, the company can coordinate with the procurement department to replenish supplies or adjust the production plan to prioritize the consumption of that type of coal, ensuring that the entire production process is not affected by coal supply and improving production planning and stability. Manually set the minimum usage days warning value, and issue a warning prompt for the coal type in the silo when the expected usage days are lower than this value. This can timely detect the lack of coal inventory. When the warning is triggered, countermeasures can be taken to speed up the procurement process, adjust the coal blending plan, and give priority to the use of coal types with tight inventory. This effectively avoids production stagnation caused by sudden shortage of coal types, ensures smooth production, reduces the risk of economic losses due to supply problems, and also improves resource utilization efficiency.

[0026] Management of on-site coal supply plan, coal storage situation in coal yard, and coal loading / unloading situation in silo includes: The on-site coal plan obtains the coal resource's weighing order number, expected arrival time, weight, coal type, quality index, and freight information, and matches the weight with the quality index based on the coal type's weighing order number; Coal storage situation in the coal yard: Obtain the coal outbound / inbound quantity, inventory, storage period, and storage time. Set different storage periods for different types of coal according to the coalification characteristics of different types of coal to prevent individual batches of coal from being left unused for a long time and causing oxidation and deterioration. The silo loading / unloading coal blending status obtains the silo loading operation status including the loading batch number, coal type information, coal yard number, loading speed, loading time, and loading weighing value. The silo unloading operation status includes the coal type, coal yard number, blending ratio, unloading speed, unloading time, unloading weighing value, target coke oven number, and target coal silo.

[0027] Specifically, by obtaining the coal resource's weighing order number, expected arrival time, weight, coal type, quality indicators, and freight information, and matching the weight and quality indicators based on the weighing order number, the company can comprehensively control the incoming coal dynamics, allowing for advance preparations for receiving coal, reasonable arrangements for sites and equipment, and ensuring that incoming coal can enter the plant efficiently and orderly. At the same time, accurate incoming coal information helps to timely adjust the coal blending plan based on production needs, avoiding production delays or coal blending errors caused by unclear incoming coal information, ensuring the smooth progress of the production process, and improving overall production efficiency. Obtaining the incoming / outgoing quantity, inventory, storage period, and storage time of each type of coal, and setting corresponding storage periods based on the coalification characteristics of each type of coal, can effectively ensure coal quality. Based on this data, the coal yard space can be rationally planned to improve coal yard utilization, prevent oxidation and deterioration of coal due to long-term storage, reduce resource waste, and lower production costs. At the same time, clear inventory data provides the production department with a basis for coal blending, enabling the formulation of reasonable production plans to ensure production continuity and stability. Recording the loading and unloading operation status of the silo can monitor the coal blending production process in real time and promptly detect abnormal situations in production. When the coal blending ratio is inaccurate or the loading / unloading speed is abnormal, measures can be taken to make adjustments. This not only ensures the stability of coal blending quality and improves the quality of coke products, but also reduces equipment damage and production stagnation caused by production abnormalities, reduces production costs, and improves economic benefits.

[0028] The cost-effectiveness score of coal types is divided into quality index score and price index score. The quality index score includes basic index score, coking index score and coal rock index score. The weight, basic score, benchmark quality index, change unit and plus / minus score of each item in the quality index score and price index score are set. The cost-effectiveness score of coal types is finally obtained through model calculation. The basic index score includes individual indicators of ash content, sulfur content and volatile matter. The coking index score includes individual indicators of reactivity and post-reaction strength. The coal rock index score includes individual indicators of standard deviation and vitrinite reflectance. Among them: Quality index score = basic index score * basic index weight + coking index score * coking index weight + coal rock index score * coal rock index weight Price index score = price index score * price index weight Single indicator score = single basic score + (benchmark quality indicator - actual quality indicator) * (single item plus or minus score / change unit) Among them, the basic index score, coking index score and coal rock index score are the sum of the scores of each individual indicator.

[0029] Specifically, the cost-effectiveness score of coal types is divided into quality index score and price index score, and the value of coal types is comprehensively considered from the two dimensions of quality and price, so that the relationship between the two can be weighed when purchasing and using coal types, avoiding excessive pursuit of low prices and ignoring quality or excessive emphasis on quality leading to excessive costs, and achieving optimal resource allocation. At the same time, the quality index score is divided into basic index score, coking index score and coal rock index score, so as to comprehensively evaluate the quality of coal types, judge the performance of coal types in various production links, and provide a basis for production plan formulation and process adjustment. Moreover, by setting the weights, basic scores, benchmark quality indicators, change units and addition and subtraction scores of each item in the quality index score and price index score, and using the above formula to calculate the final cost-effectiveness score of coal types, it can provide a method for quantitative evaluation of coal types, and flexibly adjust various parameters according to their own production needs and cost tolerance, so that the coal type evaluation is accurate, the efficiency of procurement decision-making is improved, the procurement cost is reduced, and the stability of production quality is guaranteed.

[0030] Taking into account various constraints and combining with the coal blending coking mechanism, a self-learning coke quality prediction model is constructed, and then the big data optimization algorithm is used to optimize and calculate multiple groups of coal blending schemes, including: comprehensively considering the main coal system ratio, coking coal inventory, blending coal quality, and coke quality constraints, combining various constraints with the self-learning coke quality prediction model, and using the big data optimization algorithm to optimize and calculate multiple groups of coal blending schemes, among which the big data optimization algorithm includes one or more of the differential evolution algorithm, genetic algorithm, and simulated annealing algorithm.

[0031] Specifically, a self-learning coke quality prediction model is constructed, and continuously learning from constantly updated data, gradually improving the accuracy of coke quality prediction. As time goes by and the amount of data accumulates, the relationship between coal blending and coke quality can be more clearly grasped, and the coke quality can be estimated in advance. At the same time, by comprehensively considering the main coal system ratio, coking coal inventory, blending coal quality, and coke quality constraints, a more comprehensive coal blending plan can be formulated to ensure that the plan meets core production needs, rationally utilizes resources, and ensures that coke quality meets standards, and then an economical and quality-assured coal blending plan is formulated to improve the overall efficiency of production operations. At the same time, one or more big data optimization algorithms including differential evolution algorithms, genetic algorithms, and simulated annealing algorithms are used to optimize and calculate multiple groups of coal blending plans among massive coal blending possibilities. Various parameter combinations can be explored to find the optimal or better coal blending plan that meets multiple constraints. Compared with traditional methods, the time for finding a suitable coal blending plan is greatly shortened, decision-making efficiency is improved, and more options are provided for complex and changing production environments and market demands.

[0032] The optimization calculation results in multiple coal blending schemes, including the coal blending scheme with the best cost, the best quality and the best overall performance.

[0033] Specifically, through optimization, multiple groups of coal blending plans with optimal cost, optimal quality and overall optimality are calculated. Among them, the cost-optimal coal blending plan is conducive to minimizing the cost expenditure in raw material procurement and production process while ensuring basic production quality requirements, thereby increasing the company's profit margin, effectively controlling costs and enhancing price competitiveness; the quality-optimal coal blending plan produces high-quality coke and increases product added value; the overall optimal coal blending plan takes into account both cost and quality, which can not only ensure stable product quality but also reasonably control costs, provide the most practical and sustainable coal blending strategy options, and enhance overall operational efficiency and market adaptability.

[0034] The coal and coke resource library module is used to obtain the quality of blended coal, coke quality, and small coke oven test data; The quality of the blended coal includes moisture, ash, volatile matter, sulfur, fineness, bonding index, and maximum thickness of the colloidal layer; Coke quality includes moisture, ash, volatile matter, fixed carbon, sulfur, reactivity, post-reaction strength, crushing strength, and abrasion resistance; The test data of small coke oven include coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance.

[0035] Specifically, the coal and coke resource library module obtains data on the quality of blended coal, including indicators such as moisture, ash, volatile matter, sulfur, fineness, caking index, and maximum thickness of the colloidal layer. This data provides information for the initial stage of coal blending. Through this data, the characteristics of the blended coal can be accurately understood. During the blending process, the proportions of various coal types can be reasonably adjusted according to different indicators. This optimizes the coal blending plan, improves the uniformity and stability of the blended coal, lays a good foundation for subsequent coke production, and helps to improve the consistency of coke quality and reduce quality fluctuations. Obtaining coke quality data, including moisture, ash, volatile matter, fixed carbon, sulfur, reactivity, post-reaction strength, crushing strength, and abrasion resistance, enables direct assessment of final product quality, thereby determining whether the coke meets production standards. If any coke quality indicators are found to be abnormal, the data can be traced back to the coal blending process, identifying the cause and adjusting the coal blending plan in a timely manner to ensure stable product quality and meet the needs of coke of different quality grades. Obtain small coke oven test data, including coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength, and wear resistance, providing a reference in a simulated production environment. Small coke oven tests can verify the feasibility of coal blending schemes and coke quality in advance on a smaller scale. By analyzing these test data, the coal blending scheme can be optimized before large-scale production, reducing the risks and costs of large-scale production.

[0036] The coal-coke corresponding module is used to automatically track the measured quality indicators of blended coal and coke, and incorporate it into the coal blending model sample library, including: associating the coking coal data corresponding to the coal blending plan, the measured blended coal data after blending, and the coke data obtained after production, and constructing data samples corresponding to coking coal data, blended coal data, and coke quality data. Among them, coking coal data includes coal type, blending ratio, and coking coal indicators. The measured blended coal data includes ash content, volatile matter, sulfur content, bonding index, and maximum thickness of the gelatinous layer. The coke data includes ash content, sulfur content, reactivity, post-reaction strength, crushing strength, and wear resistance.

[0037] Specifically, the data samples constructed in the coal-coke corresponding module are conducive to the continuous optimization of the model. As the data of coking coal, blending coal and coke are continuously incorporated into the coal blending model sample library, the model can capture more data characteristics and potential rules, thereby improving the accuracy of coke quality prediction, providing a reliable basis for the formulation of coal blending plans, and improving the efficiency of quality traceability. When there is a problem with coke quality, based on the associated data samples, the root cause can be traced from multiple dimensions of coal blending plans, various coking coal data and measured blending coal data, and the factors affecting coke quality can be found in time, and effective improvement measures can be taken to ensure stable product quality, improve production efficiency and reduce production costs.

[0038] The self-learning coke quality prediction model is constructed according to the following method: the coal blending model samples are input into the machine learning algorithm to obtain the prediction models of coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance. As the data samples continue to accumulate, the machine learning algorithm performs iterative model training and tuning. Among them, the machine learning algorithm includes one or more of decision trees, support vector machines, k-nearest neighbors, random forests, and transformers.

[0039] Specifically, the self-learning coke quality prediction model construction method processes the coal blending model samples through the machine learning algorithm, which can mine the intrinsic correlation between the data, thereby establishing a prediction model to provide a basis for coke quality prediction. At the same time, as the data samples accumulate, the machine learning algorithm is iteratively trained and tuned to make the model adapt to the new data characteristics and improve the prediction accuracy. In addition, by using one or more machine learning algorithms such as decision tree, support vector machine, k-nearest neighbor, random forest, and transformer, the most suitable algorithm can be selected according to its own data characteristics and production needs, thereby improving the stability and generalization ability of the prediction model, better controlling the coke quality, and reducing the production risks and cost losses caused by quality fluctuations.

[0040] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A coking coal blending expert system, characterized in that: It includes the incoming coal data management module, silo management module, coal preparation production management module, coal quality evaluation system module, coal blending optimization and quality prediction module, coal coke resource library module, and coal coke corresponding module.

2. A coking coal blending expert system according to claim 1, characterized in that: The incoming coal data management module is used to collect statistics and manage the quality indicators, incoming coal conditions, and inventory of incoming coal, so as to achieve full traceability of the quality, weight, and inventory of incoming coal. The silo management module is used to monitor the material level, weight and consumption of the silo in real time, track the inventory of coal types in the silo and the expected number of days of use, and achieve adaptive use of coal resources by issuing early warnings on the expected number of days of use; The coal preparation production management module is used to manage the on-site coal supply plan, coal storage conditions in the coal yard, and the loading and unloading of coal in the silo. It generates a stacking guide for incoming coal based on the inventory of each type of coal and the coal procurement plan. It also monitors the silo loading and unloading coal production process in real time, and realizes the statistics and management of the incoming coal situation and the coal resource consumption in the coal yard. The coal quality evaluation system module is used to classify coal types by solid standards and application, build a coal source index evaluation system, derive quality index scores, price index scores, and cost-effectiveness scores for each type of coal, and generate a coal type evaluation report for reference by procurement and production departments. The optimized coal blending and quality prediction module is used to build a self-learning coke quality prediction model based on coking coal industrial analysis, petrographic analysis, and small coke oven test data, taking into account various constraints and combining the coal blending coking mechanism. It then uses a big data optimization algorithm to optimize and calculate multiple sets of coal blending plans. After the plans are submitted, verified, and reviewed, they are automatically sent to the silo basic automation PLC control system to execute coal preparation production; The coal and coke resource library module is used to obtain the quality of the blended coal, the quality of the coke, and the test data of the small coke oven, and to issue early warnings for quality indicators that exceed the standard values, thereby providing timely and accurate data support for coal blending; The coal-coke corresponding module is used to automatically track the measured quality indicators of blended coal and coke, incorporate them into the coal blending model sample library, compare the actual coke value with the predicted value, analyze the deviation between the two, and find out the possible coal types that may cause the coke to not meet expectations; The quality indicators of the incoming coal include: coking coal quality indicators and coking coal coal rock, wherein the coking coal quality indicators include industrial indicators, bonding indicators, fluidity indicators, and ash composition indicators. The industrial indicators include moisture, ash, sulfur, and volatile matter of the coking coal. The bonding indicators include G value, Y value, X value, CRI, CSR, and Oya expansion. The fluidity indicators include Kiel fluidity. The ash composition indicators include SiO2, Al2O3, Fe2O3, TiO2, and CaO. Coking coal rock includes coal rock reflectivity and optical structure; The coal arrival status includes real-time monitoring of the coal type, weight, shipping unit, shipping time, and number of departure sections of the coal arrival channels, and statistical management of the gross weight and net weight within a selected time range according to the coal arrival channel or coal type. The coal arrival channels include rail, truck, and ship transportation; The inventory of coal entering the factory is calculated using the following formula: Today's inventory = yesterday's inventory + today's inventory - today's consumption.

3. A coking coal blending expert system according to claim 1, characterized in that: The inventory and expected usage days of the coal types in the tracking silo are calculated using the following formula: Silo coal inventory = (volume of coal in the silo at the end of coal feeding - volume of coal in the silo at the start of coal feeding) * bulk density of coal - silo coal consumption; Estimated number of days of use = silo coal inventory / silo coal consumption; The adaptive use of coal resources by issuing an early warning on the expected usage days includes: manually setting an early warning value for the minimum usage days, and issuing an early warning prompt for the type of coal in the silo when the expected usage days are lower than the early warning value for the minimum usage days.

4. A coking coal blending expert system according to claim 1, characterized in that: The management of on-site coal supply plan, coal storage situation in coal yard, and coal loading / unloading situation in silo includes: The on-site coal plan obtains the coal resource's weighing order number, expected arrival time, weight, coal type, quality index, and freight information, and matches the weight with the quality index based on the coal type's weighing order number; Coal storage situation in the coal yard: Obtain the coal outbound / inbound quantity, inventory, storage period, and storage time. Set different storage periods for different types of coal according to the coalification characteristics of different types of coal to prevent individual batches of coal from being left unused for a long time and causing oxidation and deterioration. The silo loading / unloading coal blending status obtains the silo loading operation status including the loading batch number, coal type information, coal yard number, loading speed, loading time, and loading weighing value. The silo unloading operation status includes the coal type, coal yard number, blending ratio, unloading speed, unloading time, unloading weighing value, target coke oven number, and target coal silo.

5. A coking coal blending expert system according to claim 1, characterized in that: The cost-effectiveness score of coal types is divided into quality index score and price index score, wherein the quality index score includes basic index score, coking index score and coal rock index score. The weight, basic score, benchmark quality index, change unit and plus / minus score of each item in the quality index score and price index score are set, and the cost-effectiveness score of the coal type is finally obtained through model calculation. The basic index score includes individual indicators of ash content, sulfur content and volatile matter, the coking index score includes individual indicators of reactivity and post-reaction strength, and the coal rock index score includes individual indicators of standard deviation and vitrinite reflectance, wherein: Quality index score = basic index score * basic index weight + coking index score * coking index weight + coal rock index score * coal rock index weight Price index score = price index score * price index weight Single indicator score = single basic score + (benchmark quality indicator - actual quality indicator) * (single item plus or minus score / change unit) Among them, the basic index score, coking index score and coal rock index score are the sum of the scores of each individual indicator.

6. A coking coal blending expert system according to claim 1, characterized in that: The method comprehensively considers multiple constraints, combines the coal blending coking mechanism, constructs a self-learning coke quality prediction model, and then uses a big data optimization algorithm to optimize and calculate multiple groups of coal blending schemes, including: comprehensively considering the main coal system ratio, coking coal inventory, blending coal quality, and coke quality constraints, combining the multiple constraints with the self-learning coke quality prediction model, and using a big data optimization algorithm to optimize and calculate multiple groups of coal blending schemes, wherein the big data optimization algorithm includes one or more of a differential evolution algorithm, a genetic algorithm, and a simulated annealing algorithm.

7. A coking coal blending expert system according to claim 6, characterized in that: The optimization calculation of multiple groups of coal blending schemes includes: coal blending schemes with optimal cost, optimal quality and optimal overall quality.

8. The coking coal blending expert system according to claim 1, characterized in that: The coal and coke resource library module is used to obtain the quality of blended coal, quality of coke, and small coke oven test data; The quality of the blended coal includes moisture, ash, volatile matter, sulfur, fineness, bonding index, and maximum thickness of the colloidal layer; Coke quality includes moisture, ash, volatile matter, fixed carbon, sulfur, reactivity, post-reaction strength, crushing strength, and abrasion resistance; The test data of small coke oven include coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance.

9. The coking coal blending expert system according to claim 1, characterized in that: The coal-coke corresponding module is used to automatically track the measured quality indicators of blended coal and coke, and incorporate them into the coal blending model sample library, including: associating the coking coal data corresponding to the coal blending plan, the measured blended coal data after blending, and the coke data obtained after production, and constructing data samples corresponding to the coking coal data, blended coal data, and coke quality data, wherein the coking coal data includes coal type, blending ratio, and coking coal indicators, the measured blended coal data includes ash content, volatile matter, sulfur content, bonding index, and maximum thickness of the gelatinous layer, and the coke data includes ash content, sulfur content, reactivity, post-reaction strength, crushing strength, and wear resistance.

10. The coking coal blending expert system according to claim 1, characterized in that: The self-learning coke quality prediction model is constructed according to the following method: the coal blending model sample is input into the machine learning algorithm to obtain a prediction model for the coke ash content, sulfur content, reactivity, post-reaction strength, crushing strength and wear resistance indicators. As the data samples continue to accumulate, the machine learning algorithm performs iterative model training and tuning, wherein the machine learning algorithm includes one or more of a decision tree, a support vector machine, a k-nearest neighbor, a random forest, and a transformer.

Images