With the supply-side reform and structural adjustment of the upstream coal industry and downstream steel industry in my country, the pressure on corporate profits has increased. The introduction of digital transformation to reduce costs and increase efficiency will be an inevitable choice for coal coking companies.
Huawei Cloud has joined hands with China Coking Industry Association, Coal Research Institute, Shougang Research Institute, Xinlei Group, and Yuqi Network to jointly build a smart coal blending solution for the coking industry. Huawei Cloud combines advanced algorithms and sufficient computing power with coking companies' processes, production data, and expert experience in the coking industry to provide high-precision coke quality predictions and high-efficiency coal blending optimization capabilities to help coking companies reduce costs and increase efficiency.
trend analysis
1. Cost reduction and efficiency enhancement: Artificial intelligence technology brings new possibilities for coal blending optimization and accurate prediction of coke quality
By optimizing coal blending through artificial intelligence, it can ensure the quality of coke, rationally use coal resources, save high-value coking coal, expand coking coal resources, and achieve the quality and cost reduction of coke production.
2. Refined management: upgrade the information system to improve the level of informatization
Coking companies use information integration and big data technology to combine, analyze and visualize production process data, provide data support for business decision-making, and realize refined management of the company.
3. Equipment predictive maintenance: Internet of Things + artificial intelligence to realize predictive maintenance of production equipment and increase equipment revenue
Obtain equipment operating status data through industrial protocol interfaces, monitor operating status in real time, use artificial intelligence technology to extract key features in equipment data, build equipment failure analysis models and failure prediction models, and achieve rapid fault location and maintenance recommendations.
4. Intelligent environmental protection monitoring: apply IoT and cloud computing technology to identify environmental risks in time
In recent years, the national environmental protection requirements have been increasing year by year, and the emission control of pollutants has become a key task of coking enterprises. By connecting the exhaust vent sensing device to the IoT platform for data collection and using cloud computing to analyze the data, it is possible to discover possible environmental risks in time.
Business challenge
1. The scarcity of high-quality coking coal resources and high prices lead to high production costs for enterprises
The use of coking feedstock coal accounts for more than 80% of the total coking cost. Due to the scarcity of domestic high-quality coking coals, the prices of high-quality coking coals such as main coking coal and fat coal are relatively high, resulting in high production costs for coking companies.
2. It is difficult to predict the quality of coke, causing surplus production indicators and further increasing costs
The traditional coal blending method currently widely used in China is based on the industrial composition information of raw coal for coal blending production. For key indicators such as heat and cold strength of coke, only range prediction can be made, and accurate quality prediction cannot be achieved.
3. The coal blending scene is difficult to blend, and production accidents that fail to meet the coke index are likely to occur
With the introduction of relevant national policies on the grading utilization of coal resources, it has become more and more common for coal mines and coal washing plants to achieve comprehensive utilization of coal through coal blending. Traditional coal blending based on industrial composition cannot cope with coal blending scenarios.
4. Coal blending experience depends on manual accumulation and is difficult to pass on
At present, most of the domestic coking enterprises adopt traditional coal blending and rely on the empirical formula formed by the accumulation of years of experience of coal blending experts to produce coal. It is difficult to inherit and evolve coal blending technology.
Solution scenario
Coke quality prediction
Coke quality prediction is used to solve the problem of inaccurate coke quality prediction in traditional coal blending scenarios. The current coke quality prediction accuracy has exceeded 95%. The coke quality is predicted based on the parameters of coal preparation, coal blending, and carbonization chamber (coke oven process).
- Raw coal storage yard The raw coal storage yard is associated with the raw coal inventory data of the coking enterprise, and all information about the raw coal for coking of the enterprise can be obtained in real time. The coal blender selects the various raw coals that need to be used by checking, and can accurately see the various characteristic indicators and other information of the selected coal when selecting.
- Coal blending is based on raw coal, and the proportion of each raw coal is adjusted according to actual needs.
- After confirming the coal required for coking in the carbonization chamber, the coking process needs to be confirmed, mainly to determine the coking furnace temperature and the length of coking time.
- Coke quality prediction takes the ratio of raw coal and coking process data as the input of the model to call the model for coke quality prediction. The prediction results include the quality indicators of coke ash content A d, sulfur content St, d, crushing strength M40, and wear resistance M10, reactive CRI, post-reaction strength CSR, volatile content Vdaf and several key indicators make result predictions. In addition, the yields of the four major products of coke, coal gas, tar and crude benzene will be predicted.
Optimization of coal blending ratio
The plan presents the coal blending operation as a software, and the raw coal inventory system of the associated company is already in the background, which is a good auxiliary system for coal blenders to perform coal blending operations. During operation, the coal blender selects the raw coal that needs to be used, and then inputs the upper and lower limits of the coal ratio of the coal type raw material, and then enters the required coke index and process information, and the system will give a suggested blended coal ratio.
- The yard coal preparation raw coal storage yard is associated with the raw coal inventory data of the coking enterprise, and all the information of the raw coking coal of the enterprise can be obtained in real time. The coal blender selects the various raw coals that need to be used by checking, and can accurately see the various characteristic indicators and other information of the selected coal when selecting.
- Coal blending users revise the maximum and minimum proportions of each coal source they use (the minimum proportion is 0 by default, and the maximum proportion is 50). At the same time, the inventory of each raw coal can be set to participate in model optimization.
- Coke target quality The user determines the target quality of coke, including the quality indicators of coke ash content A d, sulfur content St, d, crushing strength M25, abrasion resistance strength M10, reactive CRI, post-reaction strength CSR, and volatile content Vdaf. And the maximum and minimum Vdaf volatile content of coal into the furnace.
- The carbonization chamber confirms the coal blending amount and the coking process, mainly to determine the coking furnace temperature and the length of coking time.
- Coke uses coal source inventory and characteristic indicators of raw coal, target coke quality indicators, and coking process data as the input of the model to call the model for smart coal blending recommendation, and the system will give suggested coal blending recommendations. In addition, the quality index of coke and the yield of the product will be predicted.
Solution architecture
Smart coal blending solution
According to the data collection template, the production data of the past two years under the stable state of the process is fed back. HUAWEI CLOUD service analyzes the feedback data and retrains the model to adapt to the new customer scenario, and guides the customer to perform data verification and small coke oven effect verification , To ensure that the actual production needs of customers are met.
Architecture advantage
Model accuracy meets the production needs of coking enterprises
Provides two basic functions: coke quality prediction and coal blending optimization recommendation. Modeling is based on the actual production history data of the enterprise. The model has high accuracy. After adaptation, it can be used to assist coal blenders in completing daily coal blending work more efficiently.
Comprehensive benefit evaluation and optimization of the whole process
Based on historical coal blending data, coke, coal gas, tar and other major product yield data, combined with prices to optimize the blending ratio.
Meet the habits of coal blenders
The plan does not distinguish between coal types and mines, but only looks at the indicators, and the optimization results are closer to the actual situation of the customer, and conform to the coal blending habits of coal blenders.
Support full process optimization
Through the whole process of raw coal data, with coal and coke data, the model on the cloud is automatically refreshed to maintain the high precision of the model. In the later stage, the enterprise production data can be uploaded to the cloud, and the model can be seamlessly connected. All data such as coke ovens can be integrated into the model to achieve full Process Optimization.
Solution advantage
1. Recommendation of low-cost coal blending schemes, effectively reducing enterprise production costs
Combining the coking raw material coal data, production process data and coke quality targets used by the company, it provides a cost-effective coal blending program that meets the quality of coke for coal blending experts to choose and use, and reduce the company's production costs.
2. High-precision coke quality prediction helps coal blending experts solve the problem of difficult coke quality prediction
Modeling is based on the actual production data of the enterprise and the coal blending mechanism to provide high-precision coke quality prediction capabilities and help coal blending experts solve the problem of difficult to accurately predict coke quality.
3. Integrate advanced coal blending theory to support coal blending experts to calmly deal with the blending scene
Huawei Cloud works with industry experts such as the China Coking Association and the Chinese Academy of Coal Sciences to jointly build solutions, combined with innovative coal blending methods such as coal blending theory, to help coal blending experts calmly cope with raw coal blending.
4. Online optimization of models to help coking companies realize the precipitation and evolution of coal blending expert experience
Based on the online iterative update capability of the HUAWEI CLOUD big data platform and AI linkage support model, it ensures that the system continues to provide high precision capabilities.