scholarly journals Application Study of Sigmoid Regularization Method in Coke Quality Prediction

Complexity ◽  
2020 ◽  
Vol 2020 ◽  
pp. 1-10
Author(s):  
Shaohong Yan ◽  
Hailong Zhao ◽  
Liangxu Liu ◽  
Qiaozhi Sang ◽  
Peng Chen ◽  
...  

Coke is an indispensable and vital flue for blast furnace smelting, during which it plays a key role as a reducing agent, heat source, and support skeleton. Models of prediction of coke quality based on ANN are established to map the functional relationship between quality parameters Mt, Ad, Vdaf, St,d, and caking property (X, Y, and G) of mixed coal and quality parameters Ad, St,d, coke reactivity index (CRI), and coke strength after reaction (CSR) of coke. A regularized network training method based on Sigmoid function is designed considering that redundancy of network structure may lead to the learning of undesired noise, in which weights having little impact on performance and leading to overfitting are removed in terms of computational complexity and training errors. The cascade forward neural network with validation is found to be the most suitable one for coke quality prediction, with errors around 5%, followed by feedforward neural network structure and radial basis neural networks. The cascade forward neural network may play a guiding role during the coke production.

2020 ◽  
Author(s):  
O.Yu. Sidorov ◽  
N.A. Aristova

The article shows the application of a neural network for modeling coke quality indicators Coke Reactivity Index (CRI) and Coke Strength after Reaction (CSR). Two optimization methods were used to train the neural network. The influence of the number of neurons on the simulation results was studied. The difference between experimental and calculated data on average does not exceed 2 %. The conclusion is made about the prospects of using a neural network to predict the values of CRI and CSR of coke. Keywords: artificial neural network, coke, coke reactivity index, coke strength after reaction


Energies ◽  
2021 ◽  
Vol 14 (12) ◽  
pp. 3401
Author(s):  
Michał Rejdak ◽  
Andrzej Strugała ◽  
Aleksander Sobolewski

Coke is an integral component of the blast furnace charge; therefore, it plays an important role in the integrated steelmaking process. Achieving the required coke quality parameters by producers requires the use of a high proportion of the highest quality coking coals (hard coking coals) in the coking blends, which significantly increases the unit production costs. Approximately 75% of these costs are constituted by the cost of the coal blend’s preparation. There is a deficit in the best quality coking coals on the world market and their supply are characterized by large fluctuations in quality parameters. Therefore, from the point of view of the economics of coke production, it is advantageous to produce high-quality coke from a coke blend with the highest possible content of cheaper coals. The paper presents the results of the influence of coal charge bulk density and semi-soft coking coal content in the coking blend on the textural and structural parameters of coke, which determine its quality. Research has shown that the application of increased density influences the parameters of the texture and structure of the coke, which shape its quality parameters. The use of stamp-charging technology contributes to the improvement of the coke quality or enables the production of coke of a predetermined quality from blends containing cheaper semi-soft coals.


2010 ◽  
Vol 297-301 ◽  
pp. 1290-1294 ◽  
Author(s):  
A.N. Dmitriev ◽  
Yu.A. Chesnokov ◽  
G.Yu. Arzhadeeva

A mathematical model for the forecasting of the basic metallurgical properties of coke depending on the quality indicators of coals and carbonization parameters is developed. The base of the model are the theoretical and statistical dependences of the physico-chemical and petrographic properties of initial coals on the reactivity and mechanical strength of the coke. These indicators depend on diffusion processes during carbonization. On the basis of initial data, the material and heat balance of the coking process are made and the product price is calculated. The full-function program includes: the initial database interactively formed on the basis of the coals quality certificates and concentrates; the calculating unit of forecasting of the coke properties, allowing to fulfil the mathematical model adapting for the concrete industrial conditions of carbonization process; the unit to calculate results. The calculated values, i.e. the Coke Reactivity Index (CRI) and the Coke Strength after Reaction (CSR) can be used further for technical and economical calculations in the balance logic-statistic model of the blast furnace smelting operation. Or using the feedback, it is possible to calculate necessary relations of the burden components for the carbonization.


2015 ◽  
Vol 60 (2) ◽  
pp. 625-644
Author(s):  
Krystian Probierz ◽  
Marek Marcisz

Abstract The characteristics of variation of the CRI (Coke Reactivity Index) and CSR (Coke Strength after Reaction) indices as well as the variation of the petrographic composition of coking coal in the Pniówek deposit (SW part of the Upper Silesian Coal Basin) have been presented. The area in which the research results have been obtained has a fundamental meaning to the Polish coking coal reserves, which are characterized by high variation both in quality and coalification. So far, no research related to the determination of the CRI and CSR variation in deposits that would be based on pillar samples collected from active workings has been performed for the Polish coking coal deposits. The samples have been obtained from 6 coal seams (Załęże Beds, a part of Mudstone Series-Westphalian A), at depths between −500 and −700 m.a.s.l. The variation of CRI and CSR values has been presented both along the depth of the deposit (vertically) as well as isolines maps (horizontal variations). The relationships between the CRI and CSR index values and the parameters which are fundamental for their values, that is the R vitrinite reflectance and the petrographic composition (content of the Vtmmf vitrinite, Lmmf liptinite, and Immf inertinite macerals) have been analyzed. The examined coking coal of the Pniówek coal mine is characterized by the following values of the analyzed parameters: CRI = 19.9-60.8% (mean of 33.4%), CSR = 24.4-65.3% (mean of 49.5%), R = 0.98-1.14% (mean of 1.08%), Vtmmf = 60-81% (mean of 74%), Lmmf = 4-11% (mean of 7%), Immf = 13-31% (mean of 19%). The analysis of the variation of the coal quality parameters has not indicated evident and distinct vertical variation tendencies. When considered together with the horizontal variation in the E-W direction, in the view of the tectonics of the deposit (strike, dip, course of the main faults), it indicated a relation between the quality parameters and the direction of bed dips. In the deposit of the Pniówek coal mine, presence of coals of various quality has been confirmed. In the east, at greater depths, less coalified coal characterized by lower CRI values and higher CSR values is present. Such coal has a lesser vitrinite content and a high inertinite content. In the western direction (opposite to the dip direction), higher coalified coals, with higher CRI values and lower CSR values occur-these coals have a high content of vitrinite and low part of inertinite. Inversion of coalification has been demonstrated, as the smaller the depth the lower the reflectance of the coal should be, whereas the case in the Pniówek coal mine is the opposite. Such inversion may be related, as it has been demonstrated numerous times, to the occurrence of thermal metamorphism which modified the regional structure of coalification. No evident relationship of CSR and CRI values and the petrographic content of coal has been found, which is exhibited by low values of correlation indices. High content of inertinite in the samples characterized by relatively low values of CRI, relatively high CSR values and the lowest reflectance, however, draws attention. This runs against expectations, as usually the coal with better coking properties is characterized by the lowest content of inertinite macerals. The explanation of this relation requires further research on the inertinite macerals, especially the typically inert macerals (that is fusinite, micrinite and sclerotinite). The found relationship between the CSR and CRI values does not deviate from the data provided in literature from around the world. The correlation of the CSR index and the Pniówek coal mine vitrinite reflectance, however, is only partially consistent with the results relating to other coals. This confirms the difference of the coal in the examined area, which was exhibited many times and which should be connected to a very specific course of the coalification processes, especially the effect of thermal metamorphism.


2020 ◽  
Vol 2020 (17) ◽  
pp. 2-1-2-6
Author(s):  
Shih-Wei Sun ◽  
Ting-Chen Mou ◽  
Pao-Chi Chang

To improve the workout efficiency and to provide the body movement suggestions to users in a “smart gym” environment, we propose to use a depth camera for capturing a user’s body parts and mount multiple inertial sensors on the body parts of a user to generate deadlift behavior models generated by a recurrent neural network structure. The contribution of this paper is trifold: 1) The multimodal sensing signals obtained from multiple devices are fused for generating the deadlift behavior classifiers, 2) the recurrent neural network structure can analyze the information from the synchronized skeletal and inertial sensing data, and 3) a Vaplab dataset is generated for evaluating the deadlift behaviors recognizing capability in the proposed method.


2016 ◽  
Vol 7 (2) ◽  
pp. 105-112
Author(s):  
Adhi Kusnadi ◽  
Idul Putra

Stress will definitely be experienced by every human being and the level of stress experienced by each individual is different. Stress experienced by students certainly will disturb their study if it is not handled quickly and appropriately. Therefore we have created an expert system using a neural network backpropagation algorithm to help counselors to predict the stress level of students. The network structure of the experiment consists of 26 input nodes, 5 hidden nodes, and 2 the output nodes, learning rate of 0.1, momentum of 0.1, and epoch of 5000, with a 100% accuracy rate. Index Terms - Stress on study, expert system, neural network, Stress Prediction


Sign in / Sign up

Export Citation Format

Share Document