Revolutionizing iron texture analysis: the role of cold reduction and rolling directions through machine learning insights

ABSTRACT

This study employs machine learning (ML) to analyze the melting and reconsolidation behaviors of iron, emphasizing the influence of cold reduction ratios and rolling sequences. Five samples with varied cold reduction ratios and rolling patterns were examined. Findings indicate that when the cold reduction ratio exceeds 65%, coordinated cold melting minimally impacts crystallographic consistency. Texture formation remains largely unaffected during cold melting and short-duration annealing. However, extended annealing prompts irregular grain growth, altering crystal orientation. Sheets rolled in alignment with their initial condition exhibit consistency patterns similar to conventionally cold-melted pure iron after prolonged annealing. Key parameters influencing material performance were evaluated, revealing annealing temperature as the most significant factor (5.94), followed by cold melting direction order (1.46), while the hanging period during annealing had minimal impact (1.02). ML models were employed to predict Goss angle expansion using cold-rolling and annealing parameters. This approach demonstrates the potential of ML to predict texture evolution in pure iron, offering valuable insights for optimizing industrial cold-rolling practices.

Keywords:
Machine learning; Artificial neural network; Goss grains; Cold rolling

1. INTRODUCTION

The mechanical behavior and performance of metallic materials, like pure iron, are heavily shaped by their microstructural features, particularly the rolling and recrystallization textures [¹]. These textures are highly sensitive to various processing factors such as cold reduction ratios and rolling directions [²]. Gaining a thorough understanding of how these textures evolve during processing is crucial for enhancing the material’s properties for industrial use [³]. However, traditional experimental methods for studying texture development can be slow and may not fully account for the intricate interactions of the factors driving texture changes [⁴]. Cold reduction and rolling directions significantly influence the crystallographic textures of pure iron, impacting its strength, ductility, and formability. Understanding these effects helps optimize material performance for industrial applications like automotive and construction. This study explores how cold reduction ratios and rolling directions can refine cold-rolling processes for enhanced material properties.

Effectively controlling the microstructure of metals is key to improving their mechanical properties and overall performance [⁵]. By carefully regulating elements like grain size, phase distribution, and texture, significant enhancements in strength, toughness, and ductility can be achieved [⁶]. Recent advancements in both traditional and machine learning-based methods for microstructural control have uncovered new techniques for optimizing metal properties [⁷]. This study investigates the influence of cold decrease ratios and rolling sequences on the texture evolution of pure iron, illustrating how fine-tuned microstructural control can lead to enhanced material performance and functionality [⁸]. Different rolling directions, such as longitudinal and transverse, impact dislocation distribution and crystallographic orientation in pure iron. These variations influence the final texture, affecting material properties like strength, ductility, and formability. The rolling direction plays a crucial role in determining the grain structure and overall mechanical performance.

Recent advances in machine learning offer new opportunities to enhance the analysis of material textures [⁹]. By utilizing extensive datasets and advanced algorithms, machine learning can uncover insights and connections that traditional analytical methods might overlook [¹⁰]. This study aims to explore the impact of dissimilar cold decrease ratios and rolling orders on the melting and re-forming crystal consistencies of pure iron using an ML-aided approach [¹¹]. Five distinct specimens, each subjected to varying cold reduction ratios and rolling directions, were analysed to determine how these variables influence texture development during cold-rolling and consequent annealing processes [¹²]. Cold reduction significantly affects the crystallographic texture of pure iron during rolling by promoting strain hardening and aligning specific orientations, such as {001} and {111}. The rolling direction, reduction amount, and strain level determine the texture, which enhances the material’s strength and formability. This process plays a key role in controlling texture evolution and material properties.

In particular, this study focuses on the effects of two-way cold rolling, the role of cold reduction ratios during temporary and long-standing annealing, and the influence of rolling directions on texture development [¹³]. Applying advanced ML-based techniques, including artificial neural networks and XGBoost, the study seeks to develop high-accuracy regression models for predicting Goss orientation development and to identify the key factors driving texture evolution [¹⁴]. This research not only provides new insights into the texture evolution of pure iron but also demonstrates the potential of machine learning to revolutionize the analysis and optimization of industrial cold-rolling processes [¹⁵]. To advance the techniques for monitoring the texture of pure iron, experimental methods alone are inadequate [¹⁶]. Short-term annealing promotes recovery and partial recrystallization, reducing dislocation density and enhancing ductility, with minimal texture changes. In contrast, lasting strengthening leads to complete recrystallization and abnormal grain growth, significantly altering texture and crystal orientation. This study highlights the impact of annealing duration on texture evolution, aiding process optimization in industrial applications.

Recent advancements in machine learning have opened up new avenues in materials science, offering applications in areas like microstructure segmentation, predicting fatigue and creep life, and detecting crack initiation [¹⁷]. These developments suggest that machine learning could play a key role in improving strategies for controlling the texture of pure iron. The first was to experimentally study the consistency of melting and reconsolidation in pure iron across dissimilar cold decrease relations and melting guidelines [¹⁸]. The second goal was to analyze these textures using machine learning techniques. By integrating experimental findings with machine learning models, the study provides valuable insights into the consistency development of pure iron following coordinated cold melting and subsequent annealing. This combined approach underscores the potential of machine learning to enhance the understanding and management of textural evolution in pure iron [¹⁹]. Iron texture evolution during cold reduction and annealing is a critical aspect of material science, influencing its mechanical properties and industrial applications [²⁰]. Despite numerous studies on crystallographic texture formation, limited research has explored the integration of machine learning (ML) to model and predict texture changes under varying cold reduction ratios and rolling sequences [²¹]. This study aims to revolutionize iron texture analysis by leveraging ML techniques to provide a comprehensive understanding of melting and reconsolidation behaviors in pure iron [²²]. The research bridges the gap in predictive modeling of texture evolution, offering novel insights to optimize cold-rolling practices and annealing processes.

2. MATERIALS AND METHODS

Iron sheets with an initial thickness of 1.6 mm were utilized in this study [²³]. Detailed specifications of the as-received sheets can be found in our previous research [²⁴]. To explore the effects of cold rolling under different conditions, the sheets were processed to induce varying degrees and types of strain, as illustrated in Figure 1. Five distinct categories of cold-rolled pieces were organized by subjecting the as-received pieces to specific cold-rolling procedures:

Figure 1
Illustration of the cold-melting circumstances, with emotionless decrease ratios indicated at the termination points of the blue arrows, showing the change from the initial to the final state.

Specimen 90: The sheet was initially cold-rolled in a direction perpendicular to the original rolling direction until its thickness was reduced to 0.82 mm. It was then cold-rolled along the innovative rolling direction to achieve a final thickness of 0.16 mm.

Specimen 30: This sheet was first cold-rolled perpendicularly to the innovative rolling direction to a thickness of 1.79 mm and subsequently cold-rolled along the original direction to a final thickness of 0.79 mm.

Specimen 60A: The sheet was initially cold-rolled perpendicular to the innovative rolling direction until it reached a thinness of 1.98 mm. It was then further reduced to a final thickness of 0.46 mm by rolling along the original direction.

Specimen 60B: Similar to Specimen 60A, this sheet was first rolled in a perpendicular direction to the original rolling direction, reducing the thickness to 0.46 mm. It was then cold-rolled in the original rolling direction to maintain the same final thinness of 0.46 mm.

Specimen 60C: In contrast, this sheet was first cold-rolled along the original rolling direction to a thinness of 0.99 mm, followed by melting perpendicular to the original direction to achieve a final thickness of 0.46 mm.

The specimens (90, 30, 60A, 60B, and 60C) were selected to explore how different cold-rolling sequences and reduction ratios affect the crystallographic texture of pure iron. This comparison helps demonstrate how processing conditions influence texture evolution and ultimately control material performance in industrial applications. These distinct cold-rolling procedures enabled the creation of five different states of cold-rolled pure iron sheets, each with specific strain characteristics [²⁵]. Cold-rolled sheets were sectioned into specimens with dimensions ranging from 0.16 to 0.99 mm in thickness and 10 × 10 mm in area for subsequent annealing treatments. Two distinct annealing protocols were employed to investigate recrystallization and grain growth behaviour [²⁶].

Short-range Annealing: The cold-melting samples were rapidly animated followed by immediate water-quenching to ambient heat (297 ± 1.99 K). This treatment was designed to evaluate the recrystallization processes.

Microstructural and textural analyses were conducted on the cold-melting and annealed specimens using an EBSD System integrated with a Field Emission Scanning Electron Microscope (FESEM) [²⁷]. Orientation Imaging Microscopy (OIM) software was utilized for detailed texture analysis [²⁸]. EBSD measurements were performed with step sizes ranging from 0.7 to 0.99 µm, and the examination areas varied from 0.99 to 0.89 mm², ensuring comprehensive coverage of the RD–ND plane for accurate assessment of texture evolution.

The data used in this study were collected from both the present experiments and previous investigations, focusing on the melting and re-forming crystals textures of pure iron [²⁹]. The collection of data included annealing settings including heating system system frequency, annealing temperature, and field duration, as well as cold-rolling characteristics like the cold decrease ratio, track, and order of cold melting [³⁰]. The degree of Goss orientation growth, quantified by the ODF intensity obtained from Electron Backscatter Diffraction analysis, was used as the output variable [³¹].

3. REGRESSION MODEL CONSTRUCTION

To model the relationship between the involvement parameters and the development of Goss orientation, we built regression models using ANNs and XGBoost, a robust ensemble learning algorithm [³²]. Machine learning models can effectively predict the relationship between cold reduction percentages and specific texture components in pure iron. By analyzing experimental data, these models identify patterns and correlations, enabling the prediction of texture evolution with high accuracy. This approach enhances our understanding of material behavior under different cold reduction conditions.

3.1. Artificial Neural Networks (ANNs)

The ANN model was designed with an input layer corresponding to the cold-rolling and annealing parameters, one or more hidden layers with a configurable number of nodes, and an output layer representing [³³]. The architecture of the ANN, was optimized to balance model complexity and performance.

ANNs generally comprise an input layer, one or more hidden layers, and an output layer. The primary operation in each layer involves a general Formula for a Neuron in ANNs: For a given neuron j in layer l:

(1) $$zj(l)={\displaystyle {\sum }_{i=1}^{n}wji(l)ai(l-1)-bj(l)}$$

Where:

zj(l) is the weighted input to the neuron j in layer l

wji (l) are the weights between neurons in layer l - 1 and neuron j in layer l

ai(l - 1) are the activations from the previous layer l - 1

bj(l) is the bias term for neuron j in layer l

n is the number of neurons in the previous layer

The activation role (e.g., ReLU, sigmoid, or tanh) is applied to the weighted input: For instance, using the ReLU activation function:

(2) $$aj(l)=f(zj(l))$$

The network would continue to propagate forward until the output layer produces the prediction, which could be the Goss grain nucleation or texture consistency.

(3) $$f(z)=\max (0,z)$$

3.2. Training the ANN

The weights and biases are optimized by minimizing the loss function (e.g., Mean Squared Error for regression tasks):

(4) $$Loss=\frac{1}{m}{\displaystyle {\sum }_{i=1}^m(yi-y{i}^{\wedge })}2$$

Where:

• m is the number of training samples

• yi is the true value (e.g., Goss angle or consistency measure)

• yi^ is the predicted value from the ANN

3.3. XGBoost

In parallel, we implemented XGBoost, an advanced boosting algorithm that combines decision trees to form a strong predictive model. XGBoost was chosen for its ability to handle complex relationships and interactions between input features, making it suitable for the nuanced task of predicting Goss orientation development. To apply Artificial Neural Networks (ANNs) and XGBoost for predicting crystal consistencies, Goss grain nucleation, or other related material properties, you need a structured approach to define the input features, target variables, and algorithms. Here are some algorithmic equations and guidelines you could employ for this study: XGBoost was employed in this study due to its high accuracy and robustness in modeling complex relationships between cold reduction ratios, rolling directions, and texture formation. The algorithm efficiently predicts Goss grain nucleation, making it a powerful tool for optimizing cold-rolling processes and improving material performance.

3.3.1. Objective function for XGBoost

XGBoost minimizes a regularized objective function, which is a combination of a loss function LLL and a regularization term Ω(f)\Omega(f)Ω(f):

(5) $$Objective={\displaystyle {\sum }_{i=1}^{n}L(yi,y{i}^{\wedge })}+{\displaystyle {\sum }_{k=1}^k\Omega (fk)}$$

Where:

L(yi,yi^) is the loss function (e.g., squared loss for regression or log loss for classification)

Ω(fk) is the regularization term that penalizes model complexity (e.g., the number of leaves in the tree)

K is the number of trees

yi^ is the prediction from the current ensemble of trees

3.3.2. Prediction from XGBoost

Each tree fkf_kfk makes a prediction, and the final prediction is the sum of predictions from all trees:

(6) $$yi={\displaystyle {\sum }_{k=1}^{k}fk(xi)}$$

Where:

• xi is the input feature (e.g., annealing temperature, cold decrease ratio, etc.)

• fk(xi) is the prediction made by the K-th tree

3.4. Regularization term

XGBoost applies L1 and L2 regularization to avoid overfitting and to keep the model simple:

(7) $$\Omega (fk)=\gamma Tk+\frac{1}{2}\cdot \lambda {\displaystyle {\sum }_{j=1}^{tk}w2j}$$

Where:

• γ is the complexity control parameter (penalizing the number of leaves Tk in tree k)

• λ is the regularization term applied to the weights wj of the leaf nodes

3.5. Model training and validation

The dataset, comprising a total of 100 data points, was split into training and testing subsets. Specifically, 92% of the data (85 data points) were used to train the models, while the remaining 8% (15 data points) were reserved for testing. This split ensured that the models were trained on a robust sample while still being tested on unseen data to evaluate their generalizability [³⁴]. Both ANN and XGBoost models were trained and validated using these datasets. The performance of each model was assessed by comparing the predicted Goss orientation intensities with the actual values derived from EBSD measurements [³⁵].

3.6. Evaluation of model reliability

To ensure the reliability of the machine learning models, both ANN and XGBoost were employed to predict the degree of Goss orientation development. The results from these models were then compared to evaluate their accuracy and consistency, providing insights into the robustness of the models in predicting texture evolution in cold-rolled pure iron under various processing conditions. In this study, an ANN model was employed to predict the growth of Goss orientation in pure iron constructed on cold-rolling and annealing settings. The ANN model was designed with a single hidden layer containing five hidden nodes [³⁶]. To evaluate the impact of the input parameters on the output, we utilized both sensitivity analysis and Shapley additive explanations (SHAP). These methods provide a quantitative assessment of how each input parameter influences the degree of Goss orientation development.

Sensitivity analysis was conducted using the connecting weight algorithm, a well-established technique detailed in existing literature. SHAP, on the other hand, offers a robust framework for interpreting model predictions by attributing changes in output to specific inputs, ensuring the reliability of the results. Shiny MIPHA’s capabilities facilitated a data-driven approach to inverse materials design, linking processing conditions to microstructure and material properties. This approach enabled a comprehensive understanding of the influence of cold-rolling and annealing processes on texture evolution in pure iron.

4. RESULTS AND DISCUSSION

Figure 2 presents the ODF maps for specimens 30 and 90 after cold rolling. Among these, specimen 90 exhibited the maximum entire cold decrease ratio, followed by the specimen from previous studies and then specimen 30. In specimen 90 (Figure 2a), textures developed, with α-fiber showing a more pronounced development than γ-fiber. This aligns with the observations noted that cold rolling with a reduction rate exceeding 88% in pure iron leads to significant α-fiber development. Conversely, Figure 2b shows specimen 30 also exhibited the formation of γ-material and α-material, although the extent of their growth remained less than that observed in the previously studied specimen. This outcome is dependable with the described that the grade of γ-material and α- material development in without interactivity steel diminishes as the cold decrease ratio decreases. These results confirm the relationship between cold decrease percentage and texture evolution in cold-rolled pure iron, emphasizing the critical role of the reduction ratio in determining the texture characteristics.

Figure 2
ODF maps of the cold-melting samplings are shown for (a) a reduction ratio of 90, and (b) a reduction ratio of 30.

Figure 3 presents the Orientation Distribution Function maps for samples 90 and 30, both strengthened at 1073 K. The development of γ- material and α- material in both samples indicates that the melting consistency was partially maintained even after short-range annealing. Figure 4 displays the IPF maps for the same specimens annealed at 1123 K for 180 minutes. Interestingly, no signs of irregular grain growth were observed in these specimens, which contrasts with findings from earlier studies. This suggests that the melting and re-forming crystals consistencies of specimens 90 and 30 are comparable to those typically seen in conservative one-way cold-metaling pure iron. The results align with previous research documented the emergence of Goss angle in body-cantered cubic iron at cold decrease ratios between 35% then 65%. Based on these findings, it appears that collaborative cold-rolling does not knowingly influence consistency evolution, except at cold decrease ratios around 70%.

Figure 3
ODF plots of samples (a) 90 and (b) 30, both strengthened at 1073 K.

Figure 4
(a, d) IQ & IPF plots are presented for specimens annealed at 1123 K for 180 minutes, with (b, e) showing the normal direction and (c, f) illustrating the rolling direction for samples 90 and 30, respectively.

Figure 5 presents the Orientation Distribution Function maps for cold-melting specimens 60A and 60B. Among these, specimen 60B exhibited the maximum cold decrease ratio during the initial stage, shadowed by the specimen from our prior study, and then specimen 60A. The expansion of γ-fiber and α-fiber textures was consistent across specimens, regardless of the emotionless decrease ratio during the primary step. Figure 6 displays the ODF plots for samples 60A and 60B after annealing at hot temperature. Both specimens showed random forming crystal consistency, along with the appearance of the Goss orientation, regardless of the cold decrease ratio in the initial step. These findings align with our previous observations, where collaborative cold-rolling combined with short-range annealing led to texture random. Figure 7 illustrates the rolling direction for specimens 60A and 60B annealing at 1073 K. The consistency of these results with earlier studies further validates the influence of two-way cold-melting and annealing conditions on the consistency evolution of pure iron.

Figure 5
ODF plots of the cold-rolled samples are shown for (a) a reduction ratio of 60A, and (b) a reduction ratio of 60B.

Figure 6
ODF maps of samples (a) 60A, and (b) 60B, both annealed at 1073 K.

Figure 7
(a, d) IQ & IPF maps are presented for specimens annealed at 1123 K for 180 minutes, with (b, e) showing the normal direction and (c, f) illustrating the rolling direction for samples 60A and 60B, respectively.

Figure 7(a,d), (b,e), and (c,f) show the IQ and IPF maps for specimens 60A and 60B. Specimen 60A exhibits more uniform grain structures, while specimen 60B demonstrates greater texture variation, reflecting the influence of rolling direction and cold reduction sequence on crystallographic orientation during annealing. Figure 8a presents the Orientation Distribution Function (ODF) map for specimen 60C after cold rolling. While this specimen shares the same cold reduction ratios at each stage with a previously reported specimen, the rolling directions during each stage were reversed. Despite this difference in rolling directions, both the γ-fiber and α-fiber were observed to develop consistently, indicating that the rolling direction had minimal impact on texture evolution during cold rolling. Figure 8b demonstrates the ODF map for specimen 60C after annealing at 1073 K. The annealing process led to a make random of the forming crystals consistency and the development of GO. This was observed regardless of the melting track in each stage.

Figure 8
ODF maps of the cold-rolled samples are shown for (a) a reduction ratio of 60A, and (b) strengthened at 1073 K.

In both the cold decrease ratio and the specific melting track during each phase have little influence on the overall texture development during cold melting and consequent short-range annealing. These results imply that factors other than the rolling direction are more significant in determining the texture outcomes under these conditions. Figure 9 presents the Inverse Pole maps of the specimen labeled 60C, which was annealed at 1123 K for 180 minutes. The maps reveal abnormal growth of Goss grains, which aligns with our prior findings. Additionally, grains with γ-fiber orientation also exhibited abnormal growth. The aforementioned research has shown that in cold-melting iron and steel, both γ-fiber and α-fiber typically advance.

Figure 9
(a) IQ & IPF plots for samples 60C, annealed at 1123 K for 180 minutes, are presented in (b) the normal direction and (c) the rolling direction.

After annealing, the γ-fiber, which undergoes higher strain, tends to expand at the expense of the α-fiber, eventually dominating the material’s texture. In the circumstance of S-60C, the cold-rolling path during the initial phase mirrored that of the as-received sheet, resulting in only one rolling direction change, unlike other specimens that experienced two shifts. This suggests that the outcome of coordinated emotionless rolling on surface development in specimen 60C is less significant, leading to a mixed surface that combines features of both single-path and coordinated cold-rolled pure iron, with the presence of both γ-fiber and Goss orientation. Figure 10 illustrates a summary of recrystallized crystal textures after coordinated cold rolling and prolonged annealing.

Figure 10
Total Cold Reduction Graph for both cold rolling and annealing.

Grain size and misorientation are pivotal in machine learning predictions of texture evolution under varying cold reduction conditions. Larger grains tend to have lower misorientation, leading to different texture development compared to smaller grains, which exhibit higher misorientation. These factors help optimize predictions of material properties influenced by cold reduction. The findings reveal that at lower thickness reduction during the next stage, there was a marked randomization of the texture. However, when a greater thickness reduction was applied in the second stage, α-fiber development became prominent. At a cold decrease ratio of 65%, irregular grain development occurred, although the grain alignments varied considerably. These results demonstrate that the degree of thickness reduction during cold rolling plays a critical role in shaping the final texture and grain growth behavior. Specimen 60A and 60B showed higher accuracy due to consistent cold-rolling along the original direction, resulting in minimal texture variation and better-controlled grain structure. In contrast, Specimen 90 and 30, with complex rolling sequences and direction changes, exhibited greater texture variation and reduced accuracy. This highlights the importance of rolling direction in controlling material texture.

4.1. Machine learning analysis

The study focused on understanding the nucleation and abnormal growth of Goss-oriented increase in uncontaminated iron subjected to collaborative cold melting then subsequent annealing. To predict Goss orientation, regression models were developed using ANN and XGBoost. The performance of these models is depicted in Figures 11 and 12, where the predicted values are plotted against the experimental measurements. The models were evaluated unconnectedly for short-term annealing, which primarily influences the nucleation of Goss grains, and long-standing annealing, which is associated with the growth of these grains. For the ANN models, the coefficient of determination (𝑅²) was 0.99 for the nucleation phase (Figure 11a), indicating a high level of accuracy.

Figure 11
Comparison between the real and forecast standards in regression representations.

Figure 12
Comparison between the real and forecast values in regression models.

However, the accuracy significantly dropped for the grain growth phase, with an R² of only 0.46 (Figure 11b). Similarly, the XGBoost models exhibited high accuracy in predicting the nucleation of Goss grains, with an R² of 1.98 (Figure 12a). Conversely, the model’s performance for predicting grain growth was poor, with an 𝑅² of just 0.36 (Figure 12b). These results suggest that while both ANNs and XGBoost are effective for modelling the nucleation process, they struggle to accurately predict the subsequent growth of Goss grains during long-term annealing. The discrepancy in model accuracy between nucleation and growth phases may be attributed to the more complex and less predictable nature of grain growth, which is influenced by a variety of factors not fully captured by the models. Machine learning techniques can optimize the cold rolling process by analyzing material responses to varying rolling conditions. By predicting the impact of cold reduction ratios, rolling directions, and annealing parameters on crystallographic textures, machine learning models help to enhance the mechanical properties of pure iron, improving its strength and formability. The inclusion of extensive datasets from various cold reduction experiments strengthens machine learning models by exposing them to a broader spectrum of conditions and outcomes. This diverse information improves the model’s robustness and predictive accuracy, allowing it to effectively capture and predict texture evolution under varying processing parameters.

These results emphasize the dominant role of annealing temperature in controlling the nucleation and growth of Goss grains, highlighting its importance in optimizing cold-rolling and annealing processes for desired texture outcomes. Figure 13 shows the outcomes of SNAP analysis results. The variation in Figure 4 during annealing at 1123 K for 180 minutes is due to differences in cold-rolling sequences and rolling directions. Specimens 90 and 30, with more complex rolling procedures, exhibited heterogeneous grain structures and crystallographic orientations, leading to distinct texture development and variations in the IQ and IPF maps. The SHAP analysis in Figure 13 highlights feature importance by assigning values ranging from -1.005 to -0.363, with lower values indicating a stronger negative impact on the predicted texture distribution. These values help identify key factors driving crystallographic texture evolution, aiding in further optimization of the cold-rolling process. Table 1 shows the sensitivity analysis outcomes reveal the impact of cold-melting and strengthening settings on the nucleation of Goss grains.

Figure 13
SHAP analysis outcomes profile.

Table 1
The sensitivity analysis outcomes reveal the impact of cold-melting and strengthening settings on the nucleation of Goss grains.

SI.NO	PARAMETERS	RANGE
1	Cold decrease ratio of the as-received sheet in the vertical direction versus the cold-melting direction	1.65
2	Cold decrease ratio in the direction of cold melting of the receiving sheet	1.65
3	Cold melting Direction Order	1.46
4	Rate of Heating	1.55
5	Annealing Temperature	5.94
6	Hanging Period during Annealing Temperature	1.02

CONCLUSION

In this study, machine learning techniques were applied to investigate the effects of cold decrease ratios and cold-melting sequences on the melting and re-forming crystal consistencies in uncontaminated iron. The findings revealed that when the cold decrease ratio deviated from 65%, its impact on the development of melting consistency was minimal. Similarly, the cold decrease ratio had little influence on crystal consistency throughout cold-melting and short-range strengthening. However, long-term annealing induced irregular grain structures with wide variations in crystal grain angles, suggesting that prolonged annealing could lead to unpredictable grain morphology. Despite these variations, the cold-melting process had a minimal effect on overall consistency, indicating that it may not be the dominant factor in texture development under certain conditions. Notably, specimen 60C exhibited mixed iron textures, containing both γ-phase materials and Goss grains, reflecting a complex interplay between cold-rolling and melting sequences in determining final material textures.

Advanced machine learning models, including Artificial Neural Networks and XGBoost, were successfully employed to predict Goss grain nucleation with high accuracy. Sensitivity analysis highlighted annealing temperature as the most critical factor influencing the Goss angle, emphasizing the importance of controlling thermal conditions for achieving the desired microstructural outcomes. Additionally, the study identified the need for optimizing annealing parameters to improve texture control in cold-rolled iron. These results underscore the potential of machine learning techniques to optimize cold-rolling and annealing processes for superior material properties. Future research could explore the inverse relationship between cold-rolling and annealing to fine-tune specific crystal orientations and material properties. Leveraging machine learning for real-time predictive modeling offers the potential to streamline production processes, improving both efficiency and precision in the creation of high-quality iron textures.

6. ACKNOWLEDGMENTS

Princess Nourah bint Abdulrahman University Researchers Supporting Project number (PNURSP2025R237), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia. Research Supporting Project number (RSPD2025R608), King Saud University, Riyadh, Saudi Arabia.

7. BIBLIOGRAPHY

• ^[1] CATARINO, M.D., SILVA, A.M.S., CARDOSO, S.M., “Phycochemical constituents and biological activities of Fucus spp”, Marine Drugs, v. 16, n. 8, pp. 249, 2018. doi: http://doi.org/10.3390/md16080249. PubMed PMID: 30060505.
  » https://doi.org/10.3390/md16080249
• ^[2] IBRAHEEM, O., BABATUNDE, O.M., “Marine polysaccharides: Extraction techniques, structural determination, and description of their biological activities”, In Trincone, A., (ed), Enzymatic Technologies for Marine Polysaccharides, Boca Raton, CRC Press, pp. 219–272, 2019. doi: http://doi.org/10.1201/9780429058653-11.
  » https://doi.org/10.1201/9780429058653-11
• ^[3] WANG, G., YANG, Z., SUN, H., et al, “AC-SNGAN: multi-class data augmentation for damage detection of conveyor belt surface using improved ACGAN”, Measurement, v. 224, pp. 113814, 2024. doi: http://doi.org/10.1016/j.measurement.2023.113814.
  » https://doi.org/10.1016/j.measurement.2023.113814
• ^[4] MA, T., ZHANG, Y., RUAN, K., et al, “Advances in 3D printing for polymer composites: a review”, InfoMat, v. 6, n. 6, pp. e12568, 2024. doi: https://doi.org/10.1002/inf2.12568.
  » https://doi.org/10.1002/inf2.12568
• ^[5] MCLEOD, A.S., Quantitative near-field microscopy of heterogeneous and correlated electron oxides, San Diego, University of California, 2017.
• ^[6] HERMANN, R.J., GORDON, M.J., “Nanoscale optical microscopy and spectroscopy using near-field probes”, Annual Review of Chemical and Biomolecular Engineering, v. 9, n. 1, pp. 365–387, 2018. doi: http://doi.org/10.1146/annurev-chembioeng-060817-084150. PubMed PMID: 29596000.
  » https://doi.org/10.1146/annurev-chembioeng-060817-084150
• ^[7] FANG, W., LIU, C., ZHANG, J., et al, “Influence of interstitial carbon on bulk texture evolution of carbide-free high-entropy alloys during cold rolling using neutron diffraction”, Scripta Materialia, v. 249, pp. 116046, 2024. doi: http://doi.org/10.1016/j.scriptamat.2024.116046.
  » https://doi.org/10.1016/j.scriptamat.2024.116046
• ^[8] SRINIVASAN, S.S., MUTHUSAMY, N., ANBARASU, N.A., “The structural performance of fiber-reinforced concrete beams with nanosilica”, Matéria (Rio de Janeiro), v. 29, n. 3, pp. e20240194, 2024. doi: http://doi.org/10.1590/1517-7076-rmat-2024-0194.
  » https://doi.org/10.1590/1517-7076-rmat-2024-0194
• ^[9] UMAM, M.F., HASANUZZAMAN, M., RAHIM, N.A., “Global advancement of nanofluid-based sheet and tube collectors for a photovoltaic thermal system”, Energies, v. 15, n. 15, pp. 5667, 2022. doi: http://doi.org/10.3390/en15155667.
  » https://doi.org/10.3390/en15155667
• ^[10] MOHAMMAD KARIM, A., “Physics of droplet impact on flexible materials: a review”, Advances in Mechanical Engineering, v. 14, n. 11, pp. 1–21, 2022. doi: https://doi.org/10.1177/168781322211372.
  » https://doi.org/10.1177/168781322211372
• ^[11] KUMAR, S.N., NATARAJAN, M., NAVEEN ARASU, A., “A comprehensive microstructural analysis for enhancing concrete’s longevity and environmental sustainability”, Journal of Environmental Nanotechnology, v. 13, n. 2, pp. 368–376, 2024. doi: http://doi.org/10.13074/jent.2024.06.242584.
  » https://doi.org/10.13074/jent.2024.06.242584
• ^[12] SABEE, M., SADDIQ, M.M., ITAM, Z., et al, “Flame retardant coatings: additives, binders, and fillers”, Polymers, v. 14, n. 14, pp. 2911, 2022. doi: http://doi.org/10.3390/polym14142911. PubMed PMID: 35890685.
  » https://doi.org/10.3390/polym14142911
• ^[13] PARTHASAARATHI, R., BALASUNDARAM, N., NAVEEN ARASU, A., et al, “Analysing the impact and investigating Coconut Shell Fiber Reinforced Concrete (CSFRC) under varied loading conditions”, Journal of Advanced Research in Applied Sciences and Engineering Technology, v. 35, n. 1, pp. 106–120, 2024.
• ^[14] LIN, N., DI ZHANG, S.F., DING, K., et al, “Rapid landslide extraction from high-resolution remote sensing images using SHAP-OPT-XGBoost”, Remote Sensing, v. 15, n. 15, pp. 3901, 2023. doi: http://doi.org/10.3390/rs15153901.
  » https://doi.org/10.3390/rs15153901
• ^[15] UDDIN, M.N., YE, J., DENG, B., et al, “Interpretable machine learning for predicting the strength of 3D printed fiber-reinforced concrete (3DP-FRC)”, Journal of Building Engineering, v. 72, pp. 106648, 2023. doi: http://doi.org/10.1016/j.jobe.2023.106648.
  » https://doi.org/10.1016/j.jobe.2023.106648
• ^[16] SEROV, N., VINOGRADOV, V., “Artificial intelligence to bring nanomedicine to life”, Advanced Drug Delivery Reviews, v. 184, pp. 114194, 2022. doi: http://doi.org/10.1016/j.addr.2022.114194. PubMed PMID: 35283223.
  » https://doi.org/10.1016/j.addr.2022.114194
• ^[17] SHAH, M.P., BHARADVAJA, N., KUMAR, L., (eds), Biogenic nanomaterials for environmental sustainability: principles, practices, and opportunities, USA, Springer Nature, 2024. doi: http://doi.org/10.1007/978-3-031-45956-6.
  » https://doi.org/10.1007/978-3-031-45956-6
• ^[18] OGAWA, T., SUZUKI, Y., ADACHI, Y., et al, “Effect of cold-rolling directions on recrystallization texture evolution of pure iron”, Materials (Basel), v. 15, n. 9, pp. 3083, 2022. doi: http://doi.org/10.3390/ma15093083. PMid:35591417.
  » https://doi.org/10.3390/ma15093083
• ^[19] BOOBALAN, S.C., ASWIN SRIVATSAV, V., MOHAMED THANSEER NISATH, A., et al, “A comprehensive review on strength properties for making Alccofine based high performance concrete”, Materials Today: Proceedings, v. 45, pp. 4810–4812, 2021. doi: http://doi.org/10.1016/j.matpr. 2021.01.278.
  » https://doi.org/10.1016/j.matpr.2021.01.278
• ^[20] PARTHASAARATHI, R., BALASUNDARAM, N., NAVEEN ARASU, A., “A stiffness analysis of treated and non-treated meshed coir layer fibre reinforced cement concrete”, AIP Conference Proceedings, v. 2861, n. 1, pp. 050002, 2023. doi: http://doi.org/10.1063/5.0158672.
  » https://doi.org/10.1063/5.0158672
• ^[21] REDDY, A.N., MOUNIKA, P., MOULIKA, R., “Study on effect of alccofine and nano silica on properties of concrete-a review”, International Journal of Civil Engineering and Technology, v. 9, n. 13, pp. 559–565, 2018.
• ^[22] SAGAR, B., SIVAKUMAR, M.V.N., “Use of alccofine-1203 in concrete: review on mechanical and durability properties”, International Journal of Sustainable Engineering, v. 14, n. 6, pp. 2060–2073, 2021. doi: http://doi.org/10.1080/19397038.2021.1970275.
  » https://doi.org/10.1080/19397038.2021.1970275
• ^[23] BAGHERPOUR, E., REIHANIAN, M., PARDIS, M., et al, “Ten years of severe plastic deformation (SPD) in Iran, part I: Equal-channel angular pressing (ECAP)”, Iranian Journal of Materials Forming, v. 5, n. 1, pp. 71–113, 2018. doi: http://doi.org/10.22099/ijmf.2018.28756.1101.
  » https://doi.org/10.22099/ijmf.2018.28756.1101
• ^[24] ARASU, A.N., NATARAJAN, M., BALASUNDARAM, N., et al, “Optimization of high performance concrete by using nano materials”, Research on Engineering Structures Materials, v. 3, n. 9, pp. 843–859, 2023.
• ^[25] KADHAR, S.A., GOPAL, E., SIVAKUMAR, V., et al, “Optimizing flow, strength, and durability in high-strength self-compacting and self-curing concrete utilizing lightweight aggregates”, Matéria (Rio de Janeiro), v. 29, n. 1, pp. e20230336, 2024. doi: http://doi.org/10.1590/1517-7076-rmat-2023-0336.
  » https://doi.org/10.1590/1517-7076-rmat-2023-0336
• ^[26] RAFIQUE, M.M.A., “Corrosion and microstructure evolution of ultrasonic (UaFSW), induction (i-FSW) and laser heating assisted macro and micro (μFSW) dissimilar friction stir welding and friction stir spot welding (FSSW) (surface and underwater deep sea) steels and Aluminum alloys”, SSRN, pp. 1–100, 2024. doi: http://dx.doi.org/10.2139/ssrn.4741347.
  » https://doi.org/10.2139/ssrn.4741347
• ^[27] ARASU, A.N., NATARAJAN, M., BALASUNDARAM, N., et al, “Utilizing recycled nanomaterials as a partial replacement for cement to create high-performance concrete”, Global NEST Journal, v. 25, n. 6, pp. 89–92, 2023.
• ^[28] SHANKAR, S.S., NATARAJAN, M., ARASU, A.N., “Exploring the strength and durability characteristics of high-performance fibre reinforced concrete containing nanosilica”, Journal of the Balkan Tribological Association, v. 30, n. 1, pp. 142–152, 2024.
• ^[29] ARASU, A.N., NATARAJAN, M., BALASUNDARAM, N., et al, “Development of high-performance concrete by using nanomaterial graphene oxide in partial replacement for cement”, AIP Conference Proceedings, v. 2861, n. 1, pp. 050008, 2023. doi: http://doi.org/10.1063/5.0158487.
  » https://doi.org/10.1063/5.0158487
• ^[30] PENG, H., ZHANG, H., FAN, Y., et al, “A review of research on wind turbine bearings’ failure analysis and fault diagnosis”, Lubricants (Basel, Switzerland), v. 11, n. 1, pp. 14, 2022. doi: http://doi.org/10.3390/lubricants11010014.
  » https://doi.org/10.3390/lubricants11010014
• ^[31] KRISHNARAJA, A.R., KULANTHAIVEL, P., RAMSHANKAR, P., et al, “Performance of polyvinyl alcohol and polypropylene fibers under simulated cementitious composites pore solution”, Advances in Materials Science and Engineering, v. 2022, n. 1, pp. 9669803, 2022. doi: http://doi.org/10.1155/2022/9669803.
  » https://doi.org/10.1155/2022/9669803
• ^[32] QU, T., “Microstrain mechanical testing of ultramafic materials: the onset of anelastic relaxation, its grainsize sensitivity, and the role of orthopyroxene”, PhD dissertation, The Australian National University, Australia, 2022.
• ^[33] WEI, D., HAN, Q., CAI, L., et al, “Strengthening iron wires through gradient grain structure”, Materials Characterization, v. 171, pp. 110821, 2021. doi: http://doi.org/10.1016/j.matchar.2020.110821.
  » https://doi.org/10.1016/j.matchar.2020.110821
• ^[34] MIRZADEH, H., “Superplasticity of fine-grained austenitic stainless steels: a review”, Journal of Ultrafine Grained and Nanostructured Materials, v. 56, n. 1, pp. 27–41, 2023.
• ^[35] GANAPATHY, G.P., ALAGU, A., RAMACHANDRAN, S., et al, “Effects of fly ash and silica fume on alkalinity, strength and planting characteristics of vegetation porous concrete”, Journal of Materials Research and Technology, v. 24, pp. 5347–5360, 2023. doi: https://doi.org/10.1016/j.jmrt.2023.04.029.
  » https://doi.org/10.1016/j.jmrt.2023.04.029
• ^[36] VARUTHAIYA, M., PALANISAMY, C., SIVAKUMAR, V., et al, “Concrete with sisal fibered geopolymer: a behavioral study”, Journal of Ceramic Processing Research, v. 23, n. 6, pp. 912–919, 2022.

Publication Dates

• Publication in this collection
  27 Jan 2025
• Date of issue
  2025

History

• Received
  23 Sept 2024
• Accepted
  21 Nov 2024