Application of Artificial Neural Network to Predicting Hardenability of Gear Steel

The prediction of the hardenability and chemical composition of gear steel was studied using artificial neural networks. A software was used to quantitatively forecast the hardenability by its chemical composition or the chemical composition by its hardenability. The prediction result is more precise than that obtained from the traditional method based on the simple mathematical regression model.

Experimental and numerical study on plasma nitriding of AISI P20 mold steel

In this study, plasma nitriding was used to fabricate a hard protective layer on AISI P20 steel, at three process temperatures（450℃, 500℃, and 550℃） and over a range of time periods（2.5, 5, 7.5, and 10 h）, and at a fixed gas N2：H2 ratio of 75vol%：25vol%. The morphology of samples was studied using optical microscopy and scanning electron microscopy, and the formed phase of each sample was determined by X-ray diffraction. The elemental depth profile was measured by energy dispersive X-ray spectroscopy, wavelength dispersive spectroscopy, and glow dispersive spectroscopy. The hardness profile of the samples was identified, and the microhardness profile from the surface to the sample center was recorded. The results show that ε-nitride is the dominant species after carrying out plasma nitriding in all strategies and that the plasma nitriding process improves the hardness up to more than three times. It is found that as the time and temperature of the process increase, the hardness and hardness depth of the diffusion zone considerably increase. Furthermore, artificial neural networks were used to predict the effects of operational parameters on the mechanical properties of plastic mold steel. The plasma temperature, running time of imposition, and target distance to the sample surface were all used as network inputs; Vickers hardness measurements were given as the output of the model. The model accurately reproduced the experimental outcomes under different operational conditions; therefore, it can be used in the effective simulation of the plasma nitriding process in AISI P20 steel.

Forecasting and optimal probabilistic scheduling of surplus gas systems in iron and steel industry

To make full use of the gas resource, stabilize the pipe network pressure, and obtain higher economic benefits in the iron and steel industry, the surplus gas prediction and scheduling models were proposed. Before applying the forecasting techniques, a support vector classifier was first used to classify the data, and then the filtering was used to create separate trend and volatility sequences. After forecasting, the Markov chain transition probability matrix was introduced to adjust the residual. Simulation results using surplus gas data from an iron and steel enterprise demonstrate that the constructed SVC-HP-ENN-LSSVM-MC prediction model prediction is accurate, and that the classification accuracy is high under different conditions. Based on this, the scheduling model was constructed for surplus gas operating, and it has been used to investigate the comprehensive measures for managing the operational probabilistic risk and optimize the economic benefit at various working conditions and implementations. It has extended the concepts of traditional surplus gas dispatching systems, and provides a method for enterprises to determine optimal schedules.

Whole process prediction model of silicon steel strip on transverse thickness difference based on Takagi-Sugeno fuzzy network

The hot rolling and cold rolling control models of silicon steel strip were examined.Shape control of silicon steel strip of hot rolling was a theoretical analysis model,and the shape control of cold rolling was a data-based prediction model.The mathematical model of the hot-rolled silicon steel section,including the crown genetic model,inter-stand crown recovery model,and hot-rolled strip section prediction model,is used to control the shape of hot-rolled strip.The cold rolling shape control is mainly based on Takagi-Sugeno fuzzy network,which is used to simulate and predict the transverse thickness difference of cold-rolled silicon steel strip.Finally,a predictive model for the transverse thickness difference of silicon steel strips is developed to provide a new quality control method of transverse thickness of combined hot and cold rolling to improve the strip profile quality and increase economic efficiency.The qualified rate of the non-oriented silicon steel strip is finally obtained by applying this model,and it has been steadily upgraded to meet the needs of product quality and flexible production.

Modeling effects of alloying elements and heat treatment parameters on mechanical properties of hot die steel with back-propagation artificial neural network

Materials data deep-excavation is very important in materials genome exploration.In order to carry out materials data deep-excavation in hot die steels and obtain the relationships among alloying elements,heat treatment parameters and materials properties,a 11×12×12×4 back-propagation（BP）artificial neural network（ANN）was set up.Alloying element contents,quenching and tempering temperatures were selected as input;hardness,tensile and yield strength were set as output parameters.The ANN shows a high fitting precision.The effects of alloying elements and heat treatment parameters on the properties of hot die steel were studied using this model.The results indicate that high temperature hardness increases with increasing alloying element content of C,Si,Mo,W,Ni,V and Cr to a maximum value and decreases with further increase in alloying element content.The ANN also predicts that the high temperature hardness will decrease with increasing quenching temperature,and possess an optimal value with increasing tempering temperature.This model provides a new tool for novel hot die steel design.

Strength Prediction of Aluminum–Stainless Steel-Pulsed TIG Welding–Brazing Joints with RSM and ANN

Pulsed TIG welding–brazing process was applied to join aluminum with stainless steel dissimilar metals. Major parameters that affect the joint property significantly were identified as pulsed peak current, base current, pulse on time,and frequency by pre-experiments. A sample was established according to central composite design. Based on the sample,response surface methodology（RSM） and artificial neural networks（ANN） were employed to predict the tensile strength of the joints separately. With RSM, a significant and rational mathematical model was established to predict the joint strength.With ANN, a modified back-propagation algorithm consisting of one input layer with four neurons, one hidden layer with eight neurons, and one output layer with one neuron was trained for predicting the strength. Compared with RSM, average relative prediction error of ANN was ／10% and it obtained more stable and precise results.