Next Article in Journal
Is Precision Therapy in Infantile-Onset Epileptic Encephalopathies Still Too Far to Call Upon?
Next Article in Special Issue
Qualitative Analysis of Sleeper Supporting Condition for Railway Ballasted Tracks Using Modal Test
Previous Article in Journal
Lateral Displacement and Distance of Vehicles in Freeway Overtaking Scenario Based on Naturalistic Driving Data
Previous Article in Special Issue
A Method for Constructing a Synthesis Health Index for Metro Vehicle Wheelsets
 
 
Article
Peer-Review Record

InSAR-RiskLSTM: Enhancing Railway Deformation Risk Prediction with Image-Based Spatial Attention and Temporal LSTM Models

Appl. Sci. 2025, 15(5), 2371; https://doi.org/10.3390/app15052371
by Baihang Lyu, Ziwen Zhang * and Heinz D. Fill
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Appl. Sci. 2025, 15(5), 2371; https://doi.org/10.3390/app15052371
Submission received: 29 December 2024 / Revised: 10 February 2025 / Accepted: 15 February 2025 / Published: 23 February 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article proposes a novel approach for predicting the risk of excessive deformations in railway lines.
This approach integrates temporal and spatial information integrating three primary methods: the spatial attention encoder, the temporal risk predictor, and the feature fusion mechanism.

The article is well-written and well-structured and, in my opinion, the proposed approach is innovative. I suggest accepting it after the following minor revisions, based on the comments below:

 

  1. Expand the bibliography: the current references list includes 37 papers by East-Asian researchers out of 44. I suggest including also literature coming from other geographical areas.
  2. Considering that the paper bridges knowledge from different disciplines, I suggest expanding Section 1.1 providing more details on the operational principles of the InSAR technique, enhancing the understanding of a wider audience of readers.
  3. In Section 2.1, I suggest including a schematic representation of the proposed approach, by a more in-depth description of the modules and their interrelations.
  4. On page 2, the second bullet point claims that the proposed approach is robust to varying railway conditions and different scenarios. This aspect should be further justified in the experimental validation section.

Author Response

The article proposes a novel approach for predicting the risk of excessive deformations in railway lines. This approach integrates temporal and spatial information integrating three primary methods: the spatial attention encoder, the temporal risk predictor, and the feature fusion mechanism. The article is well-written and well-structured and, in my opinion, the proposed approach is innovative. I suggest accepting it after the following minor revisions, based on the comments below:

 

  1. Expand the bibliography: the current references list includes 37 papers by East-Asian researchers out of 44. I suggest including also literature coming from other geographical areas.

Response:

The modified part is located in Section 1.1, Line 126 130 and 132.

Thank you for your valuable suggestion. The bibliography has been expanded to include relevant research from a broader range of geographical regions, ensuring a more balanced representation of the global academic community. In particular, additional references have been incorporated from European and North American researchers who have contributed to railway deformation monitoring, InSAR applications, and deep learning-based risk prediction. These newly added references cover key topics such as advanced InSAR methodologies for large-scale infrastructure monitoring, LSTM-based time series modeling in geospatial applications, and practical implementations of predictive maintenance in railway networks worldwide. By integrating insights from diverse geographical areas, the literature review now provides a more comprehensive perspective on the state of the art in railway deformation risk prediction and strengthens the global relevance of our study.

--------------------------------------------------------------------

 

  1. Considering that the paper bridges knowledge from different disciplines, I suggest expanding Section 1.1 providing more details on the operational principles of the InSAR technique, enhancing the understanding of a wider audience of readers.

 

Response:

Thank you for your valuable suggestion. Section 1.1 has been expanded to provide a detailed explanation of the operational principles of the InSAR technique, ensuring accessibility for a wider audience.

Interferometric Synthetic Aperture Radar (InSAR) is a remote sensing technique that measures ground deformation with high precision by analyzing the phase differences between radar images captured at different times. The phase of a radar signal carries information about the distance between the radar sensor and the ground surface. By comparing the phase of two or more radar images, InSAR can detect minute changes in surface elevation, often with sub-millimeter accuracy. This capability makes it an essential tool for monitoring ground movements over large areas, even in regions that are difficult to access. InSAR techniques rely on the generation of interferograms, which are created by calculating the phase differences between corresponding pixels in two radar images. To derive accurate deformation measurements, several processing steps are required, including phase unwrapping, which resolves ambiguities in phase values, and geocoding, which converts radar coordinates into geographic coordinates. Advanced methods such as Persistent Scatterer Interferometry (PSI) and Small Baseline Subset (SBAS) further enhance InSAR's capability by focusing on stable radar reflectors or minimizing the impact of temporal and geometric decorrelation. These methods enable the detection of both gradual and abrupt deformations across a variety of terrains and infrastructures. Despite its advantages, InSAR faces challenges such as noise introduced by atmospheric effects and limitations in data availability for certain regions. However, its non-invasive nature and ability to provide continuous spatial coverage have made it an indispensable tool for infrastructure monitoring, including railway deformation analysis. By leveraging these principles, InSAR offers a robust foundation for predictive modeling in geotechnical and structural health applications.

--------------------------------------------------------------------

 

  1. In Section 2.1, I suggest including a schematic representation of the proposed approach, by a more in-depth description of the modules and their interrelations.

 

Response:

Section 2.1 provides a detailed description of the proposed approach, accompanied by a schematic representation that illustrates the framework and its components. The schematic depicts the flow of information, beginning with the input of InSAR images and temporal features, followed by their processing through the core modules, and culminating in the output of deformation risk predictions.

The Spatial Attention Encoder extracts spatial features from InSAR imagery, emphasizing regions prone to deformation risks. The Temporal Risk Predictor models sequential dependencies within deformation trends, leveraging LSTM-based structures for capturing long-term temporal patterns. The Feature Fusion Mechanism integrates the spatial and temporal features, aligning them into a cohesive representation to enhance predictive accuracy. This structure ensures that the approach effectively captures the spatial and temporal complexities inherent in deformation risk prediction.

--------------------------------------------------------------------

 

 

  1. On page 2, the second bullet point claims that the proposed approach is robust to varying railway conditions and different scenarios. This aspect should be further justified in the experimental validation section.

 

Response:

The modified part is located in Section 4.4, Page 23.

Thank you for your insightful comment. The claim regarding the model’s robustness to varying railway conditions and different scenarios has been thoroughly validated through extensive experimentation. A comprehensive evaluation was conducted using multiple datasets representing diverse railway terrains, including flat and mountainous regions, urban railway environments, and various weather conditions. The model was trained on a mixed dataset and tested across individual subsets to assess its adaptability. Experimental results confirm that the model maintains consistently high accuracy, recall, and F1-score across different conditions, with a robustness index exceeding 0.95 in all cases. The approach demonstrates strong generalization capabilities, effectively handling complex terrains and adverse weather scenarios while preserving predictive performance. These findings justify the claim that the proposed method is robust across diverse railway conditions and environmental settings.

--------------------------------------------------------------------

 

Reviewer 2 Report

Comments and Suggestions for Authors

I have some minor questions and suggestions :

1/ The loss function used to train the predictive model is defined as the mean squared error  (MSE) between the predicted and true deformation risk, but this is possible only in case of random processes. Is there any feedback to improve model in Eq. (3) by adding additional parameters for prediction that can simulate possibly existing systematic trends in residuals between predicted and observed data? Please explain.

2/ Fig. 4 might be increased. Some details are hard to read.

3/ In Eq. (15), A denotes the attention weight matrix?

Author Response

I have some minor questions and suggestions :

1/ The loss function used to train the predictive model is defined as the mean squared error  (MSE) between the predicted and true deformation risk, but this is possible only in case of random processes. Is there any feedback to improve model in Eq. (3) by adding additional parameters for prediction that can simulate possibly existing systematic trends in residuals between predicted and observed data? Please explain.

 

Response:

Thank you for your insightful comment. While the mean squared error (MSE) loss is commonly used for regression-based predictive models, we acknowledge that it assumes residuals are randomly distributed. However, in real-world railway deformation scenarios, systematic trends may exist due to geophysical factors, long-term degradation patterns, or environmental influences. To address this, we enhance the predictive model by incorporating additional parameters that capture and correct for these systematic deviations. Instead of relying solely on MSE, we introduce a trend-corrected residual component in the prediction function. This component accounts for systematic biases in the residuals by leveraging external covariates, such as seasonal effects, geological conditions, and maintenance history. By integrating this correction into the predictive model, we dynamically adjust risk estimates based on structured variations that may not be adequately captured by the original formulation. To further improve the robustness of the model, we also incorporate a regularization term that prevents overfitting while ensuring that systematic trends are properly learned. Empirical results show that this approach reduces residual autocorrelation and enhances generalization across different railway environments. By explicitly modeling systematic deviations, our method provides more reliable and accurate deformation risk predictions.

--------------------------------------------------------------------

 

2/ Fig. 4 might be increased. Some details are hard to read.

 

Response:

Thank you for your suggestion. Figure 4 has been enlarged to improve readability, ensuring that all details, including labels and structural components, are clearly visible. Additionally, the resolution has been enhanced to maintain clarity when zoomed in, allowing readers to better understand the framework's architecture. These adjustments ensure that the figure effectively conveys the necessary information without visual strain.

--------------------------------------------------------------------

 

3/ In Eq. (15), A denotes the attention weight matrix?

 

Response:

Yes, in Equation (15)( In the new manuscript, Equation (17)), A represents the attention weight matrix. It is obtained through the scaled dot-product attention mechanism, where the interaction between the query and key matrices is computed, scaled, and passed through a softmax function to normalize the weights. This ensures that the model assigns higher importance to more relevant spatial features while suppressing less informative ones. The attention weight matrix A is then applied to the value matrix to generate the fused feature representation, allowing the model to selectively focus on critical spatial regions. This mechanism enhances the representation of deformation-prone areas in the railway network, improving the effectiveness of risk prediction.

--------------------------------------------------------------------

Reviewer 3 Report

Comments and Suggestions for Authors

The article presents important aspects of the functioning of the railway infrastructure. However, in order to better receive the article, I have the following comments and suggestions for the authors:

1. Abstract - it should contain more detailed information, considering that it is to encourage reading the article. First of all, the purpose of the study has not been formulated. The authors only wrote what the presented study deals with. But what purpose they want to show the reader - they do not write about it.

2. Introduction - The authors explained well what the premises were for writing the article, but in its current form the introduction has some shortcomings. The purpose of the study is missing, which is always an integral element of the introduction. Also, at the end of the introduction there should be a description of the structure of the article (it acts as a kind of logical algorithm of the article). This description should be moved from the methodology to the introduction.

3. Literature - the literature review is divided into sections, which I think is good. The selection of literature is appropriate and in my opinion sufficient.

4. Methodology - in my opinion, the general description of the methodology, found in section 2.1, would be good to additionally present in the form of an algorithm.

5. There is a lack of discussion. It is worth introducing a discussion based on similar works. It is also worth addressing the problem of commercialization of the model - what is this model for, for whom, how can it be used in practice, etc.?

6. Conclusions - I propose to develop conclusions about the benefits for transport systems. What stakeholders may be interested in the model?

Author Response

 

The article presents important aspects of the functioning of the railway infrastructure. However, in order to better receive the article, I have the following comments and suggestions for the authors:

  1. Abstract - it should contain more detailed information, considering that it is to encourage reading the article. First of all, the purpose of the study has not been formulated. The authors only wrote what the presented study deals with. But what purpose they want to show the reader - they do not write about it.

 

Response:

Thanks for your feedback, here is the revised version:

Railway infrastructure faces significant operational threats due to ground deformation risks from natural and anthropogenic sources, posing serious challenges to safety and maintenance. Traditional monitoring methods often fail to capture the complex spatiotemporal patterns of railway deformation, leading to delayed responses and increased risks of infrastructure failure. To address these limitations, this study introduces InSAR-RiskLSTM, a novel framework that leverages the high-resolution and wide-coverage capabilities of Interferometric Synthetic Aperture Radar (InSAR) to enhance railway deformation risk prediction. The primary objective of this study is to develop an advanced predictive model that accurately captures both temporal dependencies and spatial susceptibilities in railway deformation processes. The proposed InSAR-RiskLSTM framework integrates Long Short-Term Memory (LSTM) networks with spatial attention mechanisms to dynamically prioritize high-risk regions and improve predictive accuracy. By combining image-based spatial attention for deformation hotspot identification with advanced temporal modeling, the approach ensures more reliable and proactive risk assessment. Extensive experiments on real-world railway datasets demonstrate that InSAR-RiskLSTM achieves superior predictive performance compared to baseline models, underscoring its robustness and practical applicability. The results highlight its potential to contribute to proactive railway maintenance and risk mitigation strategies by providing early warnings for infrastructure vulnerabilities. This work advances the integration of image-based methods within cyber-physical systems, offering practical tools for safeguarding critical railway networks. The dataset and code are available at https://github.com/LyuBaihang2024/InSAR-RiskLSTM.git.

--------------------------------------------------------------------

 

  1. Introduction - The authors explained well what the premises were for writing the article, but in its current form the introduction has some shortcomings. The purpose of the study is missing, which is always an integral element of the introduction. Also, at the end of the introduction there should be a description of the structure of the article (it acts as a kind of logical algorithm of the article). This description should be moved from the methodology to the introduction.

Response:

Thank you for your feedback, we have made the following additions and modifications to the Introduction:

Railway infrastructure is vulnerable to deformation risks caused by both natural and anthropogenic factors, including soil displacement, hydrological changes, and seismic activity. These deformations pose serious threats to railway safety and operational reliability, making timely risk assessment and predictive maintenance critical for mitigating potential failures. Traditional railway monitoring techniques, such as ground-based sensors and manual inspections, often fail to capture the complex spatiotemporal dynamics of deformation processes due to their limited spatial coverage and temporal resolution. As a result, these conventional approaches may lead to delayed responses and increased infrastructure vulnerability.

To address these challenges, remote sensing technologies, particularly Interferometric Synthetic Aperture Radar (InSAR), have emerged as powerful tools for large-scale deformation monitoring. InSAR provides high-resolution, long-term observations of ground displacement, enabling the detection of subtle deformations that may precede infrastructure failures. However, raw InSAR data alone does not directly translate into actionable risk assessments. The complex spatial dependencies and temporal trends in railway deformation demand advanced analytical methods that can effectively integrate both dimensions.

The primary objective of this study is to develop an effective predictive model that leverages both the spatial and temporal aspects of InSAR data to improve railway deformation risk assessment. To achieve this, we propose InSAR-RiskLSTM, a novel framework that integrates Long Short-Term Memory (LSTM) networks with spatial attention mechanisms. This approach enables the model to dynamically prioritize high-risk regions while capturing long-term temporal dependencies in deformation processes. By combining image-based spatial attention with sequential modeling, our method enhances predictive accuracy and robustness, facilitating proactive railway maintenance and risk mitigation.   Extensive experiments conducted on real-world railway datasets demonstrate that InSAR-RiskLSTM significantly outperforms baseline models in predictive performance, reinforcing its practical applicability for infrastructure monitoring.

The remainder of this paper is organized as follows. Section 2 presents the methodological framework, including the problem formulation and detailed description of the InSAR-RiskLSTM model. Section 3 outlines the experimental setup, including datasets, evaluation metrics, and comparative analyses. Section 4 discusses the results, highlighting key findings and implications for railway maintenance. Finally, Section 5 concludes the paper with a summary of contributions and directions for future research.

--------------------------------------------------------------------

 

  1. Literature - the literature review is divided into sections, which I think is good. The selection of literature is appropriate and in my opinion sufficient.

Response:

Thank you for your positive feedback. We appreciate your recognition of the literature review's structure and the selection of relevant references.

--------------------------------------------------------------------

 

  1. Methodology - in my opinion, the general description of the methodology, found in section 2.1, would be good to additionally present in the form of an algorithm.

Response:

Thank you for your insightful suggestion. To enhance the clarity of the methodology, we have structured the general description of the proposed approach in the form of an algorithm. Algorithm 1 provides a step-by-step breakdown of the InSAR-RiskLSTM workflow, detailing how spatial and temporal features are extracted, fused, and utilized for risk prediction.

The algorithm begins with data preprocessing, followed by spatial attention encoding and temporal sequence modeling. The extracted features are then fused into a unified representation, which is subsequently used to generate deformation risk scores. This structured representation clarifies the logical flow of our methodology and demonstrates how the model effectively integrates spatial and temporal dependencies for accurate risk assessment. By presenting the methodology in algorithmic form, we provide a clearer, more structured understanding of the proposed framework.

 

--------------------------------------------------------------------

 

  1. There is a lack of discussion. It is worth introducing a discussion based on similar works. It is also worth addressing the problem of commercialization of the model - what is this model for, for whom, how can it be used in practice, etc.?

Response:

The modified part is located in Section 4, Page 24.

Thank you for your valuable suggestion. A discussion section has been introduced to compare the proposed approach with existing works and to provide a broader perspective on its significance.

The proposed InSAR-RiskLSTM model introduces an advanced approach to railway deformation risk prediction by integrating spatial attention mechanisms with LSTM-based temporal modeling. Compared to traditional machine learning methods, such as decision trees and support vector machines, which primarily rely on handcrafted features and struggle with capturing sequential dependencies, our model effectively learns both spatial correlations and long-term deformation trends. Similarly, while existing deep learning approaches like CNNs and standard LSTMs have been applied to geospatial risk assessment, they often fail to fully utilize the spatial heterogeneity present in InSAR data. InSAR-RiskLSTM addresses this limitation by incorporating spatial attention, which dynamically prioritizes high-risk areas, leading to improved predictive accuracy and robustness across varying railway conditions. Beyond theoretical advancements, the model offers practical applications for railway infrastructure monitoring and risk mitigation. It can be utilized by railway operators, infrastructure maintenance agencies, and government regulators to enhance predictive maintenance strategies. By providing early warnings for potential deformation, the model enables more efficient resource allocation, reducing maintenance costs and minimizing disruptions in railway operations. Additionally, the model's ability to integrate real-time InSAR data allows for continuous monitoring, making it suitable for large-scale deployment in national railway networks.

From a commercialization perspective, InSAR-RiskLSTM has potential applications in predictive infrastructure management platforms, where it can be integrated with existing railway monitoring systems. Companies specializing in geospatial analytics, engineering consulting, and transportation safety could leverage this model to offer advanced risk assessment services. Furthermore, with increasing investments in smart infrastructure and digital twin technologies, this model can contribute to automated railway health monitoring systems, supporting decision-making in both public and private sectors. The adaptability of the framework also allows for potential extension to other critical infrastructure domains, such as highways, bridges, and pipelines, where deformation monitoring is essential for long-term structural safety.

--------------------------------------------------------------------

 

  1. Conclusions - I propose to develop conclusions about the benefits for transport systems. What stakeholders may be interested in the model?

 

Response:

Thank you for your feedback, we have made the following additions and modifications to the Conclusions:

The model is particularly beneficial for stakeholders involved in railway infrastructure management and maintenance. Railway operators can use InSAR-RiskLSTM to optimize maintenance schedules and allocate resources more efficiently, reducing both costs and downtime. Government agencies and transportation authorities can leverage its predictive capabilities to enhance railway safety regulations and implement early warning systems for infrastructure failures. Engineering and geospatial analytics firms can integrate the model into existing risk assessment platforms, providing advanced monitoring solutions for large-scale transportation networks. Additionally, insurance companies and investment firms involved in railway projects can use the model to quantify infrastructure risks and make informed financial decisions. Beyond railway systems, the framework’s adaptability allows for broader applications in transport infrastructure monitoring, including highways, bridges, and pipelines. By leveraging high-resolution InSAR data and deep learning techniques, InSAR-RiskLSTM contributes to the development of smarter, more resilient transportation networks, supporting long-term sustainability and safety in global transport systems.

--------------------------------------------------------------------

Back to TopTop