Prediction of FeO content in sintered ore based on ICEEMDAN and CNN-BiLSTM-AM DOI

Jinxin Fan,

Huan Yang, Xiaotong Li

et al.

Ironmaking & Steelmaking Processes Products and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 15, 2025

FeO content of sintered ore is an important reference index for measuring the performance ore. It significantly impacts ironmaking process, iron quality, and energy consumption. Aiming at current problem delayed poor accuracy detection results, this article proposes a hybrid network model that incorporates improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), convolutional neural (CNN), bidirectional long short-term memory (BiLSTM), attention mechanism (AM) prediction. First, time series were decomposed using ICEEMDAN to obtain sub-layers different frequencies. Then, features higher correlation selected by feature selection as inputs, followed predicting sequences CNN-BiLSTM-AM feature-selected variables, respectively. Finally, all predicted sublayer predictions reconstructed into final prediction summation. The proposed effectively captures essence sequence through algorithm, extracts deep from data CNN, contextual information BiLSTM, enhances extraction capability AM. experimental results show collaboration AM modelling improves accuracy. Additionally, algorithm employed enhance further, offering advantages over other techniques. MAE, MAPE, RMSE, RRMSE, R² new ICEEMDAN-CNN-BiLSTM-AM (ICBA) are 0.0751, 0.846%, 0.0937, 1.0500%, 0.9646, respectively, demonstrating significant improvement in outperforming relevant comparison models.

Language: Английский

Price forecasting through neural networks for crude oil, heating oil, and natural gas DOI Creative Commons
Bingzi Jin, Xiaojie Xu

Deleted Journal, Journal Year: 2024, Volume and Issue: 1, P. 100001 - 100001

Published: Feb. 29, 2024

Building price projections of various energy commodities has long been an important endeavor for a wide range participants in the market. We study forecast problem this paper by concentrating on four significant commodities. Using nonlinear autoregressive neural network models, we investigate daily prices WTI and Brent crude oil as well monthly Henry Hub natural gas New York Harbor No. 2 heating oil. prediction performance resulting from model configurations, including training techniques, hidden neurons, delays, data segmentation. Based investigation, relatively straightforward models are built that yield quite accurate reliable performance. Specifically, terms relative root mean square errors is 1.96%/1.81%/9.75%/21.76%, 1.96%/1.80%/8.76%/14.41%, 1.87%/1.78%/9.10%/16.97% training, validation, testing, respectively, overall error 1.95%/1.80%/9.51%/20.35% whole sample oil/Brent oil/New oil/Henry gas. The outcomes projection might be used technical analysis or integrated with other fundamental forecasts policy analysis.

Language: Английский

Citations

135
Wholesale price forecasts of green grams using the neural network DOI Creative Commons
Bingzi Jin, Xiaojie Xu

Asian Journal of Economics and Banking, Journal Year: 2024, Volume and Issue: unknown

Published: May 23, 2024

Purpose Agriculture commodity price forecasts have long been important for a variety of market players. The study we conducted aims to address this difficulty by examining the weekly wholesale index green grams in Chinese market. covers ten-year period, from January 1, 2010, 3, 2020, and has significant economic implications. Design/methodology/approach In order nonlinear patterns present time series, investigate auto-regressive neural network as forecast model. This modeling technique is able combine basic functions approximate more complex characteristics. Specifically, examine prediction performance that corresponds several configurations across data splitting ratios, hidden neuron delay counts, model estimation approaches. Findings Our turns out be rather simple yields with good stability accuracy. Relative root mean square errors throughout training, validation testing are specifically 4.34, 4.71 3.98%, respectively. results benchmark research show produces statistically considerably better when compared other machine learning models classic time-series econometric methods. Originality/value Utilizing our findings independent technical would one use. Alternatively, policy fresh insights into might achieved combining them (basic) outputs.

Language: Английский

Citations

104
Machine learning predictions of regional steel price indices for east China DOI
Bingzi Jin, Xiaojie Xu

Ironmaking & Steelmaking Processes Products and Applications, Journal Year: 2024, Volume and Issue: unknown

Published: May 16, 2024

From 1 January 2010 to 15 April 2021, this study examines the challenging task of daily regional steel price index forecasting in east Chinese market. We train our models using cross-validation and Bayesian optimisations implemented through expected improvement per second plus algorithm, utilise Gaussian process regressions validate findings. Investigated parameters as part model training involve predictor standardisation status, basis functions, kernels standard deviation noises. The that were built accurately predicted indices between 8 2019 with an out-of-sample relative root mean square error 0.57%, 0.84, absolute 0.48, correlation coefficient 99.81%.

Language: Английский

Citations

68
Pre-owned housing price index forecasts using Gaussian process regressions DOI
Bingzi Jin, Xiaojie Xu

Journal of Modelling in Management, Journal Year: 2024, Volume and Issue: unknown

Published: June 4, 2024

Purpose The purpose of this study is to make property price forecasts for the Chinese housing market that has grown rapidly in last 10 years, which an important concern both government and investors. Design/methodology/approach This examines Gaussian process regressions with different kernels basis functions monthly pre-owned index estimates ten major cities from March 2012 May 2020. authors do by using Bayesian optimizations cross-validation. Findings indices June 2019 2020 are accurately predicted out-of-sample established models, have relative root mean square errors ranging 0.0458% 0.3035% correlation coefficients 93.9160% 99.9653%. Originality/value results might be applied separately or conjunction other develop hypotheses regarding patterns residential real estate conduct further policy research.

Language: Английский

Citations

67
Contemporaneous causality among price indices of ten major steel products DOI
Bingzi Jin, Xiaojie Xu

Ironmaking & Steelmaking Processes Products and Applications, Journal Year: 2024, Volume and Issue: 51(6), P. 515 - 526

Published: May 5, 2024

The problem of dynamic relationships among the price indices 10 major steel products – rebar, wire, plate, hot rolled coil, cold galvannealed sheet, seamless tube, welded section and narrow strip is addressed in present work for Chinese market from 2011M7 to 2021M4. For examination contemporaneous causal links series, we use data on a daily basis combine vector error correction model directed acyclic graph. This analysis done using both Peter Clark linear non-Gaussian algorithms. With exception series each part cointegration according model, all save thin strips respond long-run equilibrium disturbances. method allows us achieve routes that allow innovation accounting, but algorithm prevented reaching an We categorise complex dynamics adjustment processes after shocks based impulsive responses, which tube are predominating comparison other seven items. Our findings show these three goods should get most consideration when designing long-term strategies prices.

Language: Английский

Citations

61
Palladium Price Predictions via Machine Learning DOI
Bingzi Jin, Xiaojie Xu

Materials Circular Economy, Journal Year: 2024, Volume and Issue: 6(1)

Published: June 11, 2024

Language: Английский

Citations

52
Predictions of steel price indices through machine learning for the regional northeast Chinese market DOI
Bingzi Jin, Xiaojie Xu

Neural Computing and Applications, Journal Year: 2024, Volume and Issue: 36(33), P. 20863 - 20882

Published: Aug. 19, 2024

Language: Английский

Citations

52
Forecasts of thermal coal prices through Gaussian process regressions DOI
Bingzi Jin, Xiaojie Xu

Ironmaking & Steelmaking Processes Products and Applications, Journal Year: 2024, Volume and Issue: 51(8), P. 819 - 834

Published: July 23, 2024

Given thermal coal's significance as a tactical energy source, price projections for the commodity are crucial investors and decision-makers alike. The goal of current work is to determine whether Gaussian process regressions useful this forecast problem using dataset closing prices coal traded on China Zhengzhou Commodity Exchange from January 4, 2016, December 31, 2020. This significant financial index that has not received enough attention in literature terms forecasting. Our forecasting exercises make use Bayesian optimizations cross-validation. 02, 2020, 2020 successfully predicted by generated models, with out-of-sample relative root mean square error 0.4210%. shown be problem. outcomes projection might used independent technical forecasts or conjunction other policy research entails developing viewpoints patterns.

Language: Английский

Citations

36
Thermal coal futures trading volume predictions through the neural network DOI
Bingzi Jin, Xiaojie Xu,

Yun Zhang

et al.

Journal of Modelling in Management, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 16, 2024

Purpose Predicting commodity futures trading volumes represents an important matter to policymakers and a wide spectrum of market participants. The purpose this study is concentrate on the energy sector explore volume prediction issue for thermal coal traded in Zhengzhou Commodity Exchange China with daily data spanning January 2016–December 2020. Design/methodology/approach nonlinear autoregressive neural network adopted performance examined based upon variety settings over algorithms model estimations, numbers hidden neurons delays ratios splitting series into training, validation testing phases. Findings A relatively simple setting arrived at that leads predictions good accuracy stabilities maintains small errors up 99.273 th quantile observed volume. Originality/value results could, one hand, serve as standalone technical predictions. They other be combined different (fundamental) forming perspectives trends carrying out policy analysis.

Language: Английский

Citations

33
Forecasts of China Mainland New Energy Index Prices through Gaussian Process Regressions DOI
Bingzi Jin, Xiaojie Xu

Deleted Journal, Journal Year: 2024, Volume and Issue: 01

Published: Jan. 1, 2024

Energy index price forecasting has long been a crucial undertaking for investors and regulators. This study examines the daily predicting problem new energy on Chinese mainland market from January 4, 2016 to December 31, 2020 as insufficient attention paid in literature this financial metric. Gaussian process regressions facilitate our analysis, training procedures of models make use cross-validation Bayesian optimizations. From 2, 2020, was properly projected by created models, with an out-of-sample relative root mean square error 1.8837%. The developed may be utilized investors’ policymakers’ policy analysis decision-making processes. Because results provide reference information about patterns indicated they also useful building similar indices.

Language: Английский

Citations

24