Intelligent coal blending system and method based on artificial intelligence

Abstract

The intelligent coal blending system is composed of a basic data layer, a data platform layer and an intelligent application layer, and basic data of four external systems including a coal blending operation system, a raw coal tracking system, a coal yard management system and an inspection and test system are collected; data acquisition, data transmission, data preprocessing, data storage and model creation, model training, model verification, model use and API interface opening are carried out. The intelligent coal blending method comprises six steps of raw coal proportioning abnormity monitoring, raw coal quality fluctuation analysis, coke quality automatic early warning, coke quality fluctuation analysis, conflict coal type combination identification and coke quality process tracing. After the inspection and test indexes of the raw coal, the coal type mixing proportion and the target coke quality are given, coke quality prediction and coke quality change early warning are automatically carried out, and a new coal blending proportion or new coal type recommendation is given. The whole coal blending process can be managed and controlled, and the coke quality can be rapidly detected.

Description

Technical Field

[0001] The present invention relates to the technical field of coking coal blending, and in particular to an intelligent coal blending system and method based on artificial intelligence. Background Art

[0002] Coking production can be divided into the following process links: raw coal delivery, coal yard management, coal blending, coal preparation, and coke refining. Before coke refining, multiple single types of coal need to be mixed in a reasonable proportion to form a blended coal, which is then sent to the coke oven for high-temperature distillation to produce coke. In order to ensure the stability of coke quality and meet the requirements of cost reduction and efficiency improvement to the greatest extent, it is necessary to control the quality and proportion of single types of coal throughout the entire coking production process. However, due to the influence of multiple factors such as the large number of single types of coal, temporary replacement of single types of coal, limited space in the coal yard, multiple coal preparation links, and long quality inspection cycles, the quality of coke is at great risk. Coke is an important raw material for blast furnace ironmaking, and fluctuations in coke quality will have a significant impact on the quality of molten iron. Therefore, how to effectively control coking coal blending and quickly predict coke quality has become a technical problem that needs to be solved in the field of coking coal blending. Summary of the invention

[0003] The present invention aims to provide an intelligent coal blending system and method based on artificial intelligence, which can achieve the purpose of stabilizing the quality of coke by controlling the whole process of coking coal blending and quickly predicting the quality of coke.

[0004] To achieve the above object, the present invention provides the following technical solutions: An intelligent coal blending system based on artificial intelligence includes a basic data layer, a data platform layer and an intelligent application layer. The basic data layer includes four external systems: coal blending operation system, raw coal tracking system, coal yard management system, and inspection and testing system; the data platform layer includes a server, two clients and an Ethernet switch, as well as four information processing units: data acquisition, data transmission, data preprocessing, and data storage. It is also equipped with five model building units: model creation, model training, model verification, model use, and API interface opening; the intelligent application layer includes six functional modules: raw coal ratio abnormal monitoring, raw coal quality fluctuation analysis, coke quality automatic warning, coke quality fluctuation analysis, conflicting coal combination identification, and coke quality process traceability.

[0005] The six functional modules of the intelligent application layer are: Abnormal coal blending monitoring module: It provides one-click manual query or regular automatic query of the system history and actual coal blending data of the current shift, and compares the actual coal blending ratio of each type with the planned ratio. If deviation is found, it will immediately issue an abnormal operation reminder. The coal blending data includes the name, ratio and weight of the coal type; Raw coal quality fluctuation analysis module: by checking the test data, the raw coal quality problems are found, and the operator is reminded to change the coal type or adjust the ratio in time; Coke quality automatic early warning module: In each link of the coal blending production process, the coke quality prediction function is automatically executed to detect coke quality fluctuations in advance and give early warning of coke quality fluctuations; Coke quality fluctuation analysis module: Automatically conduct statistical analysis on coke quality data and raw coal inspection data of each batch according to shifts. If abnormal trends are found, timely warnings are issued and specific reasons are analyzed; Conflicting coal combination identification module: real-time identification of conflicting coal combinations in the coal yard inventory, and warnings in the form of alarms if any abnormality is found, to avoid the simultaneous use of conflicting coal types; Coke quality process traceability module: After discovering abnormal coke quality, it is used to quickly find the location, weight and other information of a single type of coal with abnormal test results in different process links based on the data recorded in each system, and trace the cause of the quality accident.

[0006] An intelligent coal blending method based on artificial intelligence, using the above intelligent coal blending system, includes the following steps: (1) Monitoring of abnormal raw coal ratio: By obtaining the operation ratio from the coal blending operation system, one-click manual query or regular automatic query of system historical data and actual coal blending data of the current shift is provided, and the actual ratio of each type of coal used is compared with the planned ratio. If deviation is found, an abnormal operation reminder will be issued immediately; (2) Raw coal quality fluctuation analysis: After obtaining the inspection data of each batch from the inspection system, trend analysis and display are performed, and raw coal quality problems are discovered at the first time, reminding operators to change coal types or adjust the proportions in time; (3) Automatic early warning of coke quality: By collecting the single-type coal inspection and testing data, flow data, and weight data of each location of the coal yard, hopper, and silo from the coal yard management system and inspection and testing system in real time, the coke quality early warning function is automatically executed when new single-type coal arrives at the coal yard, the inspection and testing indicators of single-type coal are released, before the coal extraction function is started, before the coal preparation operation is started, and when the coal blending performance is inconsistent with the plan; (4) Coke quality fluctuation analysis: Automatically analyze the historical quality data of multiple furnace coke products and the raw coal test data of each batch according to the shift, and statistically show the trend of changes. If an abnormal trend occurs, an alarm will be issued in time. (5) Identification of conflicting coal combinations: Based on the raw coal's ash content, volatile matter, sulfur content, bonding index, maximum colloid layer thickness, final shrinkage and other laboratory indicators, the conflicting coal combinations in the coal yard inventory can be identified in real time. If an abnormality is found, an alarm message will be used to remind the user to avoid the simultaneous use of conflicting coal types. (6) Coke quality process traceability: After discovering abnormal coke quality, the system can quickly find the location, weight and other information of the single type of coal with abnormal test results in different process links based on the data recorded in each system, and trace the cause of the quality accident.

[0007] Furthermore, the specific steps of the above step (3) of automatic warning of coke quality are as follows: 1) Warning of abnormal test indicators of new batches of raw coal entering the coal yard; 2) Warning of abnormal raw coal testing indicators before taking coal from the coal yard; 3) Before taking coal, an abnormal warning is issued when insufficient raw coal inventory is found in the stockyard; 4) After taking coal from the coal yard, it is found that the raw coal is stacked in the wrong position in the yard; 5) After taking coal from the coal yard, it is found that the raw coal is in the wrong position in the yard; 6) After the raw coal reaches the hopper, an abnormal warning is issued if the raw coal inspection and testing indicators are found; 7) Warning of abnormal weight when taking raw coal from the hopper to the coal tower; 8) After the raw coal arrives at the coal tower, the coal blending operation data is used to detect abnormal proportions such as raw coal weight, flow rate, and moisture; 9) After the raw coal arrives at the coal tower, the inspection and testing system detects abnormalities in the raw coal testing indicators; 10) Before the coke oven production is completed and the coke quality inspection data is available, an abnormal ratio warning is issued.

[0008] Principle of the invention: The intelligent coal blending system of the present invention is modeled based on graph network models, knowledge computing and solvers. Model training is performed after the original production data is collected and preprocessed. After the model training is completed, the model accuracy is verified based on the test set data, and finally high-precision coke quality prediction and coal blending ratio optimization are achieved. The basic principles are as follows: Artificial intelligence algorithms support 50+ raw coal feature dimensions, far exceeding traditional coal blending methods. Dimensions include raw coal industrial analysis data, G value, X value, Y value, coal rock vitrinite maximum reflectivity, coal rock reflectivity range, single coal coking data, ash component minerals, Kiel fluidity data, Oa expansion, etc. Since raw coal is a complex mixture, based on the rich raw coal feature dimensions, the raw coal properties can be more accurately characterized, making the coal blending results more reliable and reducing the risk of coke quality fluctuations.

[0009] Based on the graph network model, the accuracy, generalization, migration and interpretability of the model are improved end-to-end in an automatic manner, solving the problem that traditional AI is difficult to scale and industrialize. The graph network large model innovatively divides the AI ​​model into functional areas and collaborative areas. The functional area realizes the fully automatic parallel learning optimization of a large number of base models through data difference and model difference training; the collaborative area realizes efficient collaboration of models based on the graph network, so that the accuracy and generalization of the model can be optimized. At the same time, it also ensures the rapid migration capability and interpretability of the model, and deeply explores the correlation between the characteristics of raw coal and the compatibility between different types of coal.

[0010] Knowledge computing focuses on the effective and full use of industry knowledge, and organically combines industry knowledge with AI technology. Knowledge computing solutions can effectively transform the conceptual knowledge and procedural knowledge of coal blending experts into mathematical models, and can pass on the abstract experience that previously existed in the mind in a digital way. In the scenario of coal blending optimization, artificial intelligence technology integrates the industrial mechanism model of coal blending, solves the problem of model accuracy caused by dynamic changes in the production process under industrial production environment, and also increases the interpretability of knowledge computing models, so that knowledge computing can better assist industrial production and improve production quality and efficiency.

[0011] The main features and beneficial effects of the present invention are as follows: through the creation of an intelligent coal blending system, it is possible to control the entire coal blending process and quickly predict the quality of coke, thereby achieving the goal of stabilizing the quality of coke. The intelligent coal blending system triggers coke quality prediction for each coal stacking task and coal taking task of each shift, and the quality change of incoming batches of raw coal in real time, presents the prediction results in real time, discovers quality risks in time, and adjusts the coal type or ratio; at the same time, it can identify conflicting coal combinations in inventory, and use alarm information to remind, prevent the simultaneous use of certain types of coal, and avoid coke quality accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is the architecture diagram of the smart coal blending system.

[0013] Figure 2 This is the real-time early warning flow chart for coke quality. DETAILED DESCRIPTION

[0014] The present invention will be further described below in conjunction with the embodiments of the accompanying drawings.

[0015] Reference Figure 1:A smart coal blending system based on artificial intelligence, composed of a basic data layer, a data platform layer and an intelligent application layer. The basic data layer includes four external systems: coal blending operation system, raw coal tracking system, coal yard management system, and inspection and testing system; the data platform layer includes a server, two clients and an Ethernet switch, as well as four information processing units: data acquisition, data transmission, data preprocessing, and data storage. It is also equipped with five model building units: model creation, model training, model verification, model use, and API interface opening; the intelligent application layer includes six functional modules: raw coal ratio abnormal monitoring, raw coal quality fluctuation analysis, coke quality automatic early warning, coke quality fluctuation analysis, conflicting coal combination identification, and coke quality process traceability.

[0016] Reference Figure 1 :The basic data layer of the intelligent coal blending system includes four external systems: coal blending operation system, raw coal tracking system, coal yard management system, and inspection and testing system. The specific functions of each system are as follows: Coal blending system: The coal blending system includes a control system consisting of hoppers, batching scales and belts. Each hopper stores different types of coal. The control system controls the speed of each batching belt scale according to the coal blending ratio to achieve fine coal blending. The coal on the batching belts of each hopper flows to the large belt below to form blended coal and is sent to the coke oven separately.

[0017] Raw coal tracking system: used to track the whole process of single-type coal blending in real time. The production process of single-type coal blending includes storage in the coal yard, storage in the bucket, weighing of coal for batching, entering the coal tower, entering the coke oven and other links. A coking enterprise has hundreds of coking operations a day, and the coke oven only takes finished coke for testing once a day, and the test results can only be obtained in 8 hours. If there are multiple ratios for coking in a day, they are only tested once. Relying on coke testing will miss many batches of quality problems. At the same time, the inspection and testing of finished coke products also has a large time lag. In the complete process of coking operations, the links and positions of single-type coal include: storage in the coal yard, storage in the bucket, weighing of coal for batching, entering the coal tower, and entering the coke oven. The management of single-type coal in various locations and time periods is conducive to the subsequent early warning of coke quality at each stage. Especially before coking is completed, without testing the coke indicators, the coke quality can be predicted through the AI ​​smart coal blending system. This can accurately locate the specific furnace where the quality fluctuation occurs and the quality deterioration, and remind the operator to change the coal type in time to avoid economic losses caused by unqualified coke products.

[0018] Coal yard management system: used to manage the multi-layer stacking and orderly storage and retrieval of multiple types of single coal in the yard space. The three-dimensional shape, starting and ending positions, stack height, and height of each layer of the raw coal stack are constructed through three-dimensional scanning. The material type, stacking angle, and stacking quantity are constructed through radar scanning and laser joint modeling. The three-dimensional coordinate data of the raw coal stack in the coal yard are collected to calculate the volume of the coal yard stack and build a three-dimensional model of the coal yard. The three-dimensional coal yard stack graphics are intuitively visualized and real-time changes are visualized, supporting the tracing of the historical coal storage situation in the coal yard and the prediction of future coal storage and clearance. The digital coal yard system realizes the precise management of the starting and ending distances of the storage of a single type of coal, the precise management of the starting and ending heights of the stacking layer, inventory management, and the remaining weight of each type of coal. The stacker-reclaimer in the coal yard realizes layered stacking and reclaiming, and accurate stacking and reclaiming is performed through the stacker-reclaimer travel position, cantilever horizontal swing, cantilever pitch angle, and cantilever length. The "layered and segmented material reclaiming method" is adopted. The specific methods are as follows: (1) Automatically search for the storage location and storage quantity of materials according to the stockpiling conditions (quality, capacity), material types, and material quantities of the coal yard. Send the stockpiling operation instruction to the stockpiling and reclaiming machine PLC. The stockpiling and reclaiming machine automatically moves to the specified position according to the operation instruction, adjusts the stockpiling equipment to the stockpiling point, and then starts the stockpiling to complete the operation. (2) According to the stockpiling conditions (quality, capacity), material types, and material quantities of the coal yard, the storage location of the required materials is given, and the material reclaiming entry point is calculated. Send the reclaiming operation instruction to the stockpiling and reclaiming machine PLC. The stockpiling and reclaiming machine automatically moves to the specified position according to the operation instruction, adjusts the reclaiming equipment to the reclaiming entry point, and then starts the reclaiming operation.

[0019] Inspection and testing system: used to test the quality of each batch of raw coal and coke products. The test indicators of raw coal include: ash content (Ad), volatile matter (Vdaf), sulfur content (Std), bonding index (G), maximum gelatin layer thickness (Y), final shrinkage (X), etc. Every day, a furnace of coke is sampled for index testing. The test indicators of coke include: ash content (Ad), volatile matter (Vdaf), sulfur content (Std), wear resistance (M10), crushing strength (M25, M40), reactivity (CRI), strength after thermal reaction (CSR), moisture, etc.

[0020] The four external systems of the basic data layer, namely the coal blending system, raw coal tracking system, coal yard management system, and inspection and testing system, all have mature system integration solutions; the production process data, raw coal inspection data, blended coal data, and coal data generated by these four external systems. The platform layer includes a server, two clients, and an Ethernet switch, as well as four information processing units: data acquisition, data transmission, data preprocessing, and data storage. It is also equipped with five model building units: model creation, model training, model verification, model use, and API interface opening. The data platform layer can be personalized according to the mature product technology of the existing industrial Internet platform or server.

[0021] Reference Figure 1 :The hardware and software parts of the data platform layer of the smart coal blending system are: The hardware part includes a server, two clients and an Ethernet switch. A server uses a CPU with a main frequency performance of no less than 2.1GHz / 26-Core / 35.75MB / 150W, including 8 100M network ports and 2 computing cards. The computing power should be greater than 256 TFLOPS and support parallel computing of multiple models; the two clients are deployed in different locations of coal blending operation and management, and have functions such as data monitoring, data query, coal blending operation and operation functions; an Ethernet switch uses an 8-port industrial-grade Ethernet switch, configured with Ethernet communication protocol, and uses Category 5 twisted pair cables or optical fiber connections to establish network connections between the server, client and four external systems; The software part includes four information processing units: data acquisition, data transmission, data preprocessing, and data storage. Among them, data acquisition mainly collects basic data such as production process data, raw coal inspection data, blended coal data, coke quality data, and coal yard inventory data from four external systems: coal blending operation system, raw coal tracking system, coal yard management system, and inspection and testing system. Data transmission uses 8 industrial-grade highly reliable industrial-grade Ethernet switches, Ethernet cables, and network configuration to establish a network connection between the basic data layer and the data platform layer for data transmission. Data preprocessing includes data cleaning, data fusion, data transformation, and other capabilities. Data storage supports the storage of original data, data labels, and other information to facilitate rapid backtracking and training and updating of big data models, and supports storage capacity for more than 2 years of historical data. The software part is also equipped with five model building units: model creation, model training, model verification, model use, and API interface opening. Among them, model creation is to store the aggregated and processed data on the service according to the coking full-process production process, and then select the appropriate algorithm for algorithm development; model training is the process of using the labeled and processed basic data to let the model learn the patterns in the data, update the model's weights and parameters, and optimize the model performance; model verification is to use a part of the actual coking production data as the input of the model to verify the model accuracy after the model training is completed, and compare the predicted results with the actual recorded result indicators, and refresh and iterate the model in the feedback system to ensure that the accuracy of the model can meet the actual production use; model use is to deploy the model on the data platform layer after determining that the model can meet the coking production use, and use the knowledge it has learned to predict or generate responses to new input data; API interface opening is to open the API call interface for upper-level applications to provide data support and solution verification for the life cycle benefit analysis and process optimization of the coking full process.

[0022] Reference Figure 1 :The intelligent application layer of the smart coal blending system includes six functional modules: raw coal ratio abnormal monitoring module, raw coal quality fluctuation analysis module, coke quality automatic warning module, coke quality fluctuation analysis module, conflicting coal combination identification module, and coke quality process traceability module. The specific implementation plans of each functional module are as follows: Raw coal ratio abnormality monitoring module: Traditional coking coal blending operations require operators to manually input a large amount of configuration data, which is time-consuming, labor-intensive and error-prone. The intelligent coal blending system obtains the operation ratio from the coal blending operation system, provides one-click manual query or regular automatic query of the system history and the actual coal blending data of the current shift (such as the name, ratio, weight, etc. of the coal type), and compares the actual ratio of each coal type with the planned ratio. If a deviation is found, an abnormal operation reminder will be issued immediately.

[0023] Raw coal quality fluctuation analysis module: Raw coal is stacked in different batches when it enters the factory. The quality of different batches of the same coal may also fluctuate. The intelligent coal blending system obtains the inspection and testing data of each batch from the inspection and testing system (the raw coal enters the coal yard for inspection, and it takes one or two days to get the inspection and testing results). It then performs trend analysis and display, discovers raw coal quality problems at the first time, and reminds operators to change the type of coal or adjust the blending ratio in time to avoid economic losses caused by fluctuations in coke quality.

[0024] Coke quality automatic warning module: The intelligent coal blending system collects the single-type coal inspection and testing data, flow data, and weight data of each position of the coal yard stack, hopper (coal blending trough), and silo from the coal yard management system and the inspection and testing system in real time. When new single-type coal arrives at the coal yard, the inspection and testing indicators of single-type coal come out, before the coal-taking function is started, before the coal preparation operation is started, and when the coal blending performance is inconsistent with the plan, it automatically performs the coke quality prediction function, discovers the coke quality fluctuation in advance, and gives a prompt alarm to prompt the operator to make corrections according to the specific reasons. After the coke oven production is completed, the intelligent coal blending system finds that the historical coal blending performance queried by date and shift is inconsistent with the planned ratio. In the absence of coke quality inspection and testing results, the system can calculate the coke quality fluctuation and promptly notify the downstream blast furnace production link to take corrective measures. For example, if it is predicted that the thermal mass of coke is insufficient, the blast furnace is notified in advance to increase the injection of coal to ensure the normal furnace temperature.

[0025] Coke quality fluctuation analysis module: Traditional coking coal blending is a discrete single prediction. The coke quality prediction result data is not saved, and the historical data is not aggregated, spliced ​​and analyzed, which is not conducive to the operator's control of the coke quality trend. The intelligent coal blending system automatically analyzes the multiple inspection and testing indicators in the historical multi-furnace coke product quality data and the raw coal inspection and testing data of each batch according to the shift, and statistically presents the trend of changes. When raw coal quality problems are found, early warnings are issued in time to remind operators to change coal types or adjust the ratio in time to avoid the impact of quality fluctuations of different batches of single raw coal on coke quality.

[0026] Conflicting coal combination identification module: The test indicators of raw coal include ash content (Ad), volatile matter (Vdaf), sulfur content (Std), bonding index (G), maximum colloid layer thickness (Y), final shrinkage (X), etc. The test indicators of different raw coal types are different. If a certain indicator of multiple coal types decreases at the same time, using them at the same time will cause serious deterioration of coke quality. Therefore, when selecting the combination of coal types for blending, pay attention to avoid such conflicts. The intelligent coal blending system makes use of recommendations for different categories of single coal with abnormal quality based on the historical use of single coal types. The coal preparation workshop maintains a coal type conflict configuration file in real time based on the historical use of different coal types. The file records the information of conflicting coal types. The intelligent coal blending system identifies the conflicting coal type combinations in the coal yard inventory in real time based on the coal type conflict combination configuration file provided by coking production. If an abnormality is found, it will be reminded with an alarm message to avoid the simultaneous use of conflicting coal types.

[0027] Coke quality process traceability module: After discovering abnormal coke quality, it is used to quickly find the location, weight and other information of a single type of coal with abnormal test results in different process links based on the data recorded in each system, and trace the cause of the quality accident.

[0028] An intelligent coal blending method based on artificial intelligence, using the above-mentioned intelligent coal blending system, comprises the following steps: (1) Monitoring of abnormal raw coal ratio: By obtaining the operation ratio from the coal blending operation system, one-click manual query or regular automatic query of system historical data and actual coal blending data of the current shift is provided, and the actual ratio of each type of coal used is compared with the planned ratio. If deviation is found, an abnormal operation reminder will be issued immediately.

[0029] Specifically, the coal blending operation of each shift will issue the manually reviewed proportion and tonnage. The coal blending operation system executes the material taking, weighing and conveying actions of different types of coal through the control system of the hopper, batching scale and belt. The raw coal proportion abnormality monitoring module obtains the type information of the raw coal actually taken by monitoring and recording the control command of the hopper coal release, and finds out the type of raw coal actually taken is wrong; obtains the weight information of the raw coal actually taken by monitoring and recording the weighing process data of the batching scale, and finds out the weight error of the raw coal actually taken; obtains the transportation path of the raw coal actually taken and the arrival silo information by monitoring and recording the belt opening and closing time, speed, and connection process data between belts, and finds out the transportation path of the raw coal actually taken and the flow direction of the raw coal arriving at the silo is wrong.

[0030] (2) Raw coal quality fluctuation analysis: After obtaining the inspection data of each batch from the inspection system, trend analysis and display are performed, and raw coal quality problems are discovered at the first time, reminding operators to change coal types or adjust the proportions in time; Specifically, the raw coal quality fluctuation analysis module connects to the raw coal test data in the inspection and testing system in real time, presents each batch curve for each index such as ash content (Ad), volatile matter (Vdaf), sulfur content (Std), bonding index (G), maximum colloid layer thickness (Y), final shrinkage (X), etc., and intuitively displays the index change trend. By setting the alarm threshold, the user is automatically reminded of the deterioration information of the raw coal index; Furthermore, the raw coal quality fluctuation analysis module is connected to the procurement, logistics, and coal yard systems in real time, and provides detailed information such as the raw coal production unit, shipping location, shipping date, arrival date, coal yard stacking location, coal collection location, coal collection date, and inspection date for batches of raw coal with deteriorating quality.

[0031] (3) Automatic early warning of coke quality: By collecting the single-type coal inspection and testing data, flow data, and weight data of each location of the coal yard, hopper, and silo from the coal yard management system and inspection and testing system in real time, the coke quality early warning function is automatically executed when new single-type coal arrives at the coal yard, the inspection and testing indicators of single-type coal are released, before the coal extraction function is started, before the coal preparation operation is started, and when the coal blending performance is inconsistent with the plan; Specifically, in each of the above links, the automatic early warning module of coke quality uses the real-time collected coal type and ratio data as input, automatically triggers the coke quality prediction function, and outputs the coke quality prediction results.

[0032] (4) Coke quality fluctuation analysis: Automatically analyze the historical quality data of multiple furnace coke products and the raw coal test data of each batch according to the shift, and statistically show the trend of changes. If an abnormal trend occurs, an alarm will be issued in time. Specifically, the coke quality fluctuation analysis module connects to the coke quality test data in the inspection and testing system in real time, presents the curve of each furnace for each index such as ash content (Ad), volatile matter (Vdaf), sulfur content (Std), abrasion resistance (M10), crushing strength (M25, M40), reactivity (CRI), strength after thermal reaction (CSR), moisture, etc., and intuitively displays the trend of index changes. By setting the alarm threshold, the user is automatically reminded of the deterioration information of the coke index. Furthermore, the raw coal quality fluctuation analysis module connects with the procurement, logistics, coal yard system, and coal blending system in real time, and provides detailed information such as the raw coal production unit, shipping place, shipping date, arrival date, coal yard stacking location, coal collection location, coal collection date, inspection and testing date, and coking operation shift for batches of coke quality deterioration.

[0033] (5) Identification of conflicting coal combinations: Based on the raw coal's ash content, volatile matter, sulfur content, bonding index, maximum colloid layer thickness, final shrinkage and other laboratory indicators, the conflicting coal combinations in the coal yard inventory can be identified in real time. If an abnormality is found, an alarm message will be used to remind the user to avoid the simultaneous use of conflicting coal types. Specifically, the conflict coal combination identification module stores the list of common conflict coal types entered by the customer and the corresponding raw coal inspection and testing index data such as ash content (Ad), volatile matter (Vdaf), sulfur content (Std), bonding index (G), maximum colloid layer thickness (Y), final shrinkage (X), etc.; the conflict coal combination identification module connects to the raw coal and inspection and testing data stored in the customer's coal yard in real time; the conflict coal combination identification module regularly checks the coal types currently stored in the coal yard and the conflict coal type list; if it is found that there are 2 or more coal types stored in the coal yard in the conflict coal type list, a conflict information reminder will be given; Furthermore, the conflict coal combination identification module automatically reminds users of the deterioration trend of coke indicators by setting alarm thresholds; Furthermore, the conflict coal combination identification module compares the inspection and testing index data of the coal types currently stored in the coal yard, such as ash content (Ad), volatile matter (Vdaf), sulfur content (Std), bonding index (G), maximum colloid layer thickness (Y), final shrinkage (X), etc., with the raw coal inspection and testing information in the conflict coal type list. If the similarity of the test indicators is high, it can also be regarded as a conflict coal type.

[0034] Furthermore, the conflicting coal type combination identification module provides the user with a setting of the ratio of similarity of the assay indicators.

[0035] (6) Coke quality process traceability: After discovering abnormal coke quality, the location, weight and other information of the single type of coal with abnormal test results in different process links can be quickly found based on the data recorded in each system, so as to trace the cause of the quality accident; Specifically, the coke quality process traceability module connects to the procurement, logistics, coal yard system, and coal blending system in real time, and provides detailed information such as the raw coal production unit, shipping place, shipping date, arrival date, coal yard stacking location, coal collection location, coal collection date, collection weight, inspection and testing date, and coking operation shifts for the batches of coal involved in the anomalies found in the above modules.

[0036] See attached Figure 1 , Figure 2 :The intelligent coal blending system provides a real-time early warning function for the quality of coke with a single type of coal in each link of the coking production process. The specific steps are as follows: 1) Warning of abnormal inspection indicators of new batches of raw coal entering the coal yard: As the batches of raw coal entering the coal yard change, if the quality fluctuates according to the raw coal inspection data, the intelligent coal blending system will combine the production plan to predict the production furnace number, time, and coke quality fluctuations of the batch of raw coal, and issue a quality warning if it is found that the coke quality requirements are not met; it will promptly detect abnormal coal blending ratios and adjust the coal blending ratio or change the coal type in a timely manner to avoid economic losses caused by unreasonable raw coal blending ratios.

[0037] 2) Warning of abnormal raw coal testing indicators before taking coal from the coal yard: The intelligent coal blending system automatically predicts the changing trends of multiple indicators such as coke ash, volatile matter, sulfur, M10, M25, M40, CRI, CSR, etc. in the coke product quality according to the shift. If the preset target is not met, an alarm will be issued in time, and new coal blending suggestions will be given based on the coal type data in the coal yard inventory.

[0038] The system combines the three-dimensional model of the coal yard to obtain the location and number of layers of abnormal single type of coal in the yard. The system combines the production plan to predict the production furnace number and time of using the batch of raw coal, as well as the fluctuation of coke quality, and timely adjusts the coal blending ratio or changes the coal type.

[0039] 3) Abnormal warning when insufficient raw coal inventory is found in the stockpile before coal is taken: The intelligent coal blending system combines the raw coal inventory information and production plan of the coal yard management system to predict the specific furnace and time when insufficient inventory will cause quality fluctuations, trigger the purchase plan of the coal type, adjust the coal blending ratio or change the coal type.

[0040] 4) After taking coal from the coal yard, it is found that the raw coal is stacked in the wrong position in the yard. Warning: The intelligent coal blending system combines the coal taking production operation records, obtains the raw coal taking operation time from the raw coal tracking system, and can locate the interval and position of the wrong coal type in the bucket trough, the interval and position of the wrong coal type in the coal tower, and the furnace and time when the wrong coal type participates in coking. Combined with the inspection and test results and the proportion, the quality fluctuation of coke production in the future is predicted. If the raw coal has entered the bucket trough, the system adjusts the coal blending formula according to the actual coal storage situation of each bucket trough to eliminate the coke quality problem; if the raw coal has entered the coal tower, the system can predict the specific furnace where the quality fluctuation occurs and the corresponding coke quality in combination with the production plan to provide management decisions.

[0041] 5) After taking coal from the coal yard, it is found that the raw coal is taken from the wrong position in the yard. Warning: The intelligent coal blending system combines the coal taking production operation records, obtains the time of the raw coal taking operation from the raw coal tracking system, and can locate the interval and position of the wrong coal in the bucket, the interval and position of the wrong coal in the coal tower, the furnace and time of the wrong coal in coking, and the fluctuation of coke quality. If the raw coal has entered the bucket, the system adjusts the coal blending formula according to the actual coal storage situation of each bucket to eliminate the coke quality problem; if the raw coal has entered the coal tower, the system can predict the specific furnace where the quality fluctuation occurs and the corresponding coke quality in combination with the production plan, and provide management decisions.

[0042] 6) Warning of abnormal raw coal test indicators after the raw coal arrives at the hopper: The intelligent coal blending system can predict the quality fluctuation of coke and the specific furnace, time and coke quality fluctuation by obtaining the interval position of abnormal single coal in the hopper from the raw coal tracking system. The system adjusts the coal blending formula according to the actual coal storage situation in each hopper to eliminate coke quality problems.

[0043] 7) Warning of abnormal weight during the process of taking raw coal from the hopper to the coal tower: The intelligent coal blending system obtains abnormal information on the weight, flow, moisture or proportion of raw coal from the coal blending operation system. It can predict the quality fluctuations of coke and the specific furnaces, time and coke quality fluctuations, and remind operators to standardize coal taking operations, change coal types or adjust the blending ratio in time to eliminate coke quality problems.

[0044] 8) After the raw coal arrives at the coal tower, the coal blending operation data is used to detect abnormal proportions such as raw coal weight, flow rate, and moisture content, and provide early warning: The intelligent coal blending system obtains the interval position of abnormal raw coal in the coal tower from the raw coal tracking system, predicts the quality fluctuations of coke and the specific furnace number, time, and coke quality fluctuations, and reminds operators to change coal types or adjust the proportions in time to eliminate coke quality problems.

[0045] 9) After the raw coal arrives at the coal tower, the inspection and testing system will detect abnormal raw coal testing indicators and issue an early warning: The intelligent coal blending system obtains the interval position of abnormal raw coal in the coal tower from the raw coal tracking system, predicts the quality fluctuation of coke and the specific furnace number, time and coke quality fluctuation, and reminds operators to change coal types or adjust the blending ratio in time to eliminate coke quality problems.

[0046] 10) Before the coke oven production is completed and the coke quality inspection data is available, an abnormal ratio warning is issued.

Claims

1. An intelligent coal blending system based on artificial intelligence, including a basic data layer, a data platform layer and an intelligent application layer. The basic data layer includes four external systems: coal blending operation system, raw coal tracking system, coal yard management system, and inspection and testing system. The data platform layer includes a server, two clients and an Ethernet switch. It also includes four information processing units: data acquisition, data transmission, data preprocessing, and data storage. It also includes five model construction units: model creation, model training, model verification, model use, and API interface opening. It is characterized by: The intelligent application layer includes six functional modules: abnormal proportion monitoring module, raw coal quality fluctuation analysis module, coke quality automatic warning module, coke quality fluctuation analysis module, conflicting coal combination identification module, and process quality traceability module; The raw coal ratio abnormality monitoring module provides the ability to manually query the system history and actual coal blending data of the current shift with one-click or automatically query it regularly, and compares the actual coal ratio of each type with the planned ratio. If deviation is found, an abnormal operation reminder will be issued immediately. The coal blending data includes the name, ratio and weight of the coal type. The raw coal quality fluctuation analysis module detects raw coal quality problems by checking test data, and reminds operators to change coal types or adjust the proportion in time; The coke quality automatic early warning module automatically performs the coke quality prediction function in each link of the coal blending production process, detects coke quality fluctuations in advance, and gives early warning of coke quality fluctuations; The coke quality fluctuation analysis module automatically performs statistical analysis on the coke quality data and the raw coal inspection data of each batch according to the shift. If an abnormal trend is found, it will issue a warning in time and analyze the specific cause; The conflicting coal combination identification module identifies the conflicting coal combination in the coal yard inventory in real time, and if an abnormality is found, an alarm message is used to remind, so as to avoid the conflicting coal combinations from being used at the same time; The coke quality process tracing module is used to quickly find the location and weight information of the single type of coal with abnormal test results in different process links according to the data recorded by each system after the coke quality is found to be abnormal, so as to trace the cause of the quality accident.

2. An intelligent coal blending method based on artificial intelligence, using the intelligent coal blending system according to claim 1, characterized in that The following steps are involved: (1) Monitoring of abnormal raw coal ratio: By obtaining the operation ratio from the coal blending operation system, one-click manual query or regular automatic query of system historical data and actual coal blending data of the current shift is provided, and the actual ratio of each type of coal used is compared with the planned ratio. If deviation is found, an abnormal operation reminder will be issued immediately; (2) Raw coal quality fluctuation analysis: After obtaining the inspection data of each batch from the inspection system, trend analysis and display are performed, and raw coal quality problems are discovered at the first time, reminding operators to change coal types or adjust the proportions in time; (3) Automatic early warning of coke quality: By collecting the single-type coal inspection and testing data, flow data, and weight data of each location of the coal yard, hopper, and silo from the coal yard management system and inspection and testing system in real time, the coke quality early warning function is automatically executed when new single-type coal arrives at the coal yard, the inspection and testing indicators of single-type coal are released, before the coal extraction function is started, before the coal preparation operation is started, and when the coal blending performance is inconsistent with the plan; (4) Coke quality fluctuation analysis: Automatically analyze the historical quality data of multiple furnace coke products and the raw coal test data of each batch according to the shift, and statistically show the trend of changes. If an abnormal trend occurs, an alarm will be issued in time. (5) Identification of conflicting coal combinations: Based on the raw coal's ash content, volatile matter, sulfur content, bonding index, maximum colloid layer thickness, final shrinkage and other laboratory indicators, the conflicting coal combinations in the coal yard inventory can be identified in real time. If an abnormality is found, an alarm message will be used to remind the user to avoid the simultaneous use of conflicting coal types. (6) Coke quality process traceability: After discovering abnormal coke quality, the system can quickly find the location, weight and other information of the single type of coal with abnormal test results in different process links based on the data recorded in each system, and trace the cause of the quality accident.

3. The intelligent coal blending method based on artificial intelligence according to claim 2 is characterized in that Step (3) Automatic early warning of coke quality, the specific steps are as follows: 1) Warning of abnormal test indicators of new batches of raw coal entering the coal yard; 2) Warning of abnormal raw coal testing indicators before taking coal from the coal yard; 3) Before taking coal, an abnormal warning is issued when insufficient raw coal inventory is found in the stockyard; 4) After taking coal from the coal yard, it is found that the raw coal is stacked in the wrong position in the yard; 5) After taking coal from the coal yard, it is found that the raw coal is in the wrong position in the yard; 6) After the raw coal reaches the hopper, an abnormal warning is issued if the raw coal inspection and testing indicators are found; 7) Warning of abnormal weight when taking raw coal from the hopper to the coal tower; 8) After the raw coal arrives at the coal tower, the coal blending operation data is used to detect abnormal proportions such as raw coal weight, flow rate, and moisture; 9) After the raw coal arrives at the coal tower, the inspection and testing system detects abnormalities in the raw coal testing indicators; 10) Before the coke oven production is completed and the coke quality inspection data is available, an abnormal ratio warning is issued.

Images