Assessing Polymer Pipe Viscoelasticity Using Transient Signals and Neural Networks

Talha Sino^*
*Correspondence: Talha Sino, Department of Computational Mechanics, Van Lang University, Ho Chi Minh City, Vietnam, Email:

Introduction

The assessment of viscoelastic properties in polymer pipes is crucial for ensuring their reliability and performance in various applications, including water distribution, gas transportation, and industrial fluid handling. Polymer pipes exhibit time-dependent mechanical behavior due to their viscoelastic nature, meaning that they experience both elastic and viscous responses when subjected to stress. Understanding these properties is essential for predicting long-term performance, structural integrity, and potential failure mechanisms. Traditional methods for evaluating viscoelastic parameters often involve mechanical testing, which can be time-consuming and require specialized equipment. However, recent advancements in signal processing and artificial intelligence have paved the way for more efficient and accurate assessment techniques. One such approach involves the use of transient signals and artificial neural networks to extract viscoelastic parameters from polymer pipes in a non-destructive manner. Transient signals, typically generated by pressure waves or mechanical excitations, provide valuable insights into the material properties of polymer pipes. When a transient event occurs, such as a sudden change in pressure or mechanical impact, the resulting wave propagates through the pipe system, interacting with the material’s inherent characteristics. The response of the pipe to these transient signals carries information about its viscoelastic properties, including storage and loss moduli, relaxation times, and damping behavior. By analyzing these signals, it is possible to derive meaningful parameters that describe the material’s mechanical behavior under different conditions.

Description

Artificial Neural Networks (ANNs) have emerged as powerful tools for pattern recognition and complex data analysis, making them well-suited for interpreting transient signal responses. ANNs are capable of learning intricate relationships between input and output data, allowing them to model the nonlinear behavior of viscoelastic materials with high accuracy. The process begins with data acquisition, where transient signals are recorded using pressure sensors, accelerometers, or other measurement devices placed along the polymer pipe. These signals are then preprocessed to remove noise and extract relevant features, such as amplitude variations, frequency content, and decay rates. Feature selection plays a critical role in improving the performance of ANN models. Relevant features are chosen based on their sensitivity to viscoelastic behavior, ensuring that the neural network focuses on the most informative aspects of the transient response. Once the feature set is established, the ANN is trained using a dataset consisting of known viscoelastic parameters and corresponding transient signal features. The training process involves optimizing network weights to minimize the error between predicted and actual viscoelastic properties. Common training algorithms include back propagation and gradient descent, which iteratively adjust weights to improve model accuracy [1,2].

One of the advantages of using ANNs for viscoelastic parameter assessment is their ability to generalize from training data to unseen cases. Once trained, the ANN can process new transient signal inputs and provide rapid estimations of viscoelastic parameters without requiring extensive mechanical testing. This capability is particularly beneficial for in-situ monitoring and real-time assessment of polymer pipe conditions. Additionally, ANN-based approaches can accommodate variations in pipe geometry, material composition, and environmental conditions, making them versatile for different applications. Experimental validation of ANN models is essential to ensure their reliability and accuracy. This typically involves comparing ANN predictions with results obtained from traditional mechanical testing methods, such as Dynamic Mechanical Analysis (DMA) or stress relaxation experiments. A high degree of correlation between ANN predictions and experimental results indicates the model’s effectiveness in capturing viscoelastic behaviour. Furthermore, sensitivity analysis can be performed to evaluate how different input parameters influence ANN predictions, providing insights into the robustness of the model. Despite their advantages, ANN-based approaches also face challenges. One key issue is the need for a sufficiently large and diverse training dataset to ensure model accuracy across different conditions [3-5].

Conclusion

Collecting high-quality training data can be labor-intensive, requiring extensive experimental efforts. Additionally, neural networks operate as black-box models, meaning that their decision-making process is not always interpretable. Efforts to improve model interpretability, such as using explainable AI techniques, can help address this limitation. Future research directions in this field include integrating hybrid models that combine ANNs with physics-based simulations to enhance predictive capabilities. Incorporating advanced signal processing techniques, such as wavelet transforms or machine learning-based feature extraction, can further improve the robustness of transient signal analysis. Additionally, real-time implementation of ANN models using embedded systems and edge computing could enable on-site monitoring of polymer pipes, providing immediate feedback on structural health and potential failure risks. The use of transient signals and artificial neural networks for assessing polymer pipe viscoelasticity represents a significant advancement in material characterization techniques. By leveraging the rich information contained in transient responses and the predictive power of ANNs, this approach offers a non-destructive, efficient, and accurate method for evaluating viscoelastic parameters. Continued advancements in machine learning, signal processing, and experimental validation will further refine this technique, making it a valuable tool for ensuring the long-term reliability of polymer pipe systems in various engineering applications.

Acknowledgment

None.

Conflict of Interest

None.

References

1. Siddique, Muhammad Farooq, Zahoor Ahmad and Jong-Myon Kim. "Pipeline leak diagnosis based on leak-augmented scalograms and deep learning." Eng Appl Comput Fluid Mech 17 (2023): 2225577.

   Google Scholar, Crossref, Indexed at

2. Ahmad, Zahoor, Tuan-Khai Nguyen and Jong-Myon Kim. "Leak detection and size identification in fluid pipelines using a novel vulnerability index and 1-D convolutional neural network." Eng Appl Comput Fluid Mech 17 (2023): 2165159.

   Google Scholar, Crossref, Indexed at

3. Gong, Jinzhe, Aaron C. Zecchin, Martin F. Lambert and Angus R. Simpson. "Determination of the creep function of viscoelastic pipelines using system resonant frequencies with hydraulic transient analysis." J Hydraul Eng 142 (2016): 04016023.

   Google Scholar, Crossref, Indexed at

4. Pan, B., H. F. Duan, S. Meniconi and K. Urbanowicz, et al. "Multistage frequency-domain transient-based method for the analysis of viscoelastic parameters of plastic pipes." J Hydraul Eng 146 (2020): 04019068.

   Google Scholar, Crossref, Indexed at

5. Wang, Xun, Jingrong Lin, Mohamed S. Ghidaoui and Silvia Meniconi, et al. "Estimating viscoelasticity of pipes with unknown leaks." Mech Syst Signal Process 143 (2020): 106821.

   Google Scholar, Crossref, Indexed at