Pavement Engineering Research Articles

Toward Climate Resilient Asphalt Binder Selection in Canada

Climate change and extreme weather events present a fundamental challenge to pavement engineering and planning practice, given that transportation infrastructure has traditionally been planned and designed using historical climate data under the implicit assumption that climate is stationary and future conditions will resemble past ones. Transportation professionals should consider future risks scenarios in road design, material selection and system operations. This will be challenging given the inherent uncertainties in any climate projections. However, changes might be needed since transportation professionals are expected to deliver cost-effective and climate-resilient transportation infrastructure. The material selection plays an important role in the durability of flexible roads; therefore, it is important to consider the whole road design life cycle in terms of climate change and extreme weather events impact on asphalt selection as well as to consider different traffic speeds. This led to the development of the NRC Climate Adaptation and Asphalt Selection Tool (CAAST).

Predicting resilient modulus for pavement design: a comprehensive review of artificial neural network applications

ABSTRACT The Resilient Modulus (M R) is an essential parameter describing the mechanical behaviour of pavement materials, but the tests to obtain it are time-consuming and costly. Therefore, machine learning, especially Artificial Neural Networks (ANN), has recently been an effective alternative for predicting M R. Although there has been an increase in articles on ANN for predicting M R, no systematic review has yet categorised the publications, algorithm structures, predictive parameters, and model accuracy. This review provides a comprehensive overview of studies using ANN to predict M R for pavement materials. It examines various ANN subtypes and model structures, addressing a gap in understanding these models and identifying prevalent predictive parameters. Twenty-five peer-reviewed articles published in English-language journals were identified using keywords related to neural networks and M R. For fine soils, all models reviewed use moisture content as a predictive parameter. In contrast, granular soils models typically include stress state and physical characteristics of the materials. For recycled or stabilised materials, there is significant variability in model inputs. This review underscores the potential of ANN in enhancing predictive models for M R in pavement engineering, pointing towards future research and application refinement in this area of study.

Asphalt pavement raveling recognition based on deep learning algorithms

Abstract Pavement raveling can lead to the development of issues such as potholes. Monitoring and providing real-time early warnings for road surface developments is crucial for timely maintenance planning and effectively controlling the occurrence and progression of surface issues. To achieve intelligent identification of raveling distress on asphalt pavement surfaces, we utilized a portable laser-based pavement surface scanning device to collect raveling images, followed by histogram equalization to eliminate noise caused by environmental factors. Additionally, data enhancement strategies such as rotation and mirroring were used to enrich training data and expand the sample size. A Convolutional Neural Network (CNN) was employed to perform multi-layer convolution and pooling operations on the images to quickly extract features of distress. Finally, five pre-trained transfer learning models—ResNet-18, DenseNet, VGG16, Inception, and EfficientNet—were introduced to identify the asphalt surface disease data and the normal surface data. The results showed that the accuracy and loss of the training sets of several models exhibited a positive trend, with the EfficientNet model demonstrating stable and superior performance under the same training conditions. The precision matrix, confusion matrix, precision, recall, and F1 score were used as evaluation indexes to assess the effectiveness of the models for identifying pavement raveling. The EfficientNet and Inception models had the best accuracy and comprehensive performance, with accuracy rates of 0.96 and 0.95, respectively. The precision, recall, and F1 score are 0.95, 0.96, and 0.96 for the EfficientNet model respectively. A total of 146 images correctly predicted raveling disease and only 4 normal images were incorrectly predicted as diseased. For example, the EfficientNet model’s identification accuracy for pockmark disease on the PC platform was 0.963, verifying the model’s feasibility in practical applications. The principal contribution of this paper lies in the efficient and precise identification of asphalt pavement raveling, which adversely affects vehicular traveling comfortability. This provides theoretical and technical support for the intelligent identification of asphalt pavement defects and enhancing road asset maintenance management in the pavement engineering industry.

Innovative Pavement Solutions: A Comprehensive Review from Conventional Asphalt to Sustainable Colored Alternatives

Climate change significantly impacts transportation infrastructure, particularly asphalt pavements. Similarly, the heat absorption of paved surfaces, especially conventional black pavements, significantly intensifies the urban microclimate. Paved surfaces, including asphalt pavements, account for over 30% of the covered surfaces and are vulnerable to rising temperatures, which cause not only pavement distress, such as rutting and cracking, but also urban heat islands (UHI). Sustainable pavement solutions, specifically colored pavements, have been investigated for their potential to mitigate these effects. This review presents an extensive overview of current pavement technologies, emphasizing conventional asphalt’s economic, environmental, and functional characteristics. A discussion of the benefits and challenges of colored pavements is also provided, including their ability to reduce UHI, enhance safety, and contribute to sustainable urban growth. This paper discusses advancements in pavement material science, the use of recycled materials, and the application of reflective coatings, providing insights into sustainable infrastructure development. Transitioning from conventional black pavements to sustainable colored alternatives is not merely a matter of material choice but a strategic transition toward resilient urban planning. Increasing demand for environmentally friendly infrastructure could prompt the construction industry to adopt colored pavements as a tool to promote environmental stewardship.

The State of the Art on Phase Change Material-Modified Asphalt Pavement

During the construction and maintenance of asphalt pavement, a lot of non-renewable resources are consumed, which discharge a variety of waste gasses and smoke, causing a serious impact on the environment. Reducing society’s reliance on non-renewable resources is therefore key to improving sustainability. It is found that phase change materials (PCMs), as environmentally friendly materials, can spontaneously store and release heat energy by changing the phase state, thus reducing the adverse effect of temperature on asphalt pavement, reducing the occurrence of high-temperature stress, minimizing the cost of road construction and maintenance, and saving resources. In order to promote the application of PCMs in asphalt pavement, to promote self-controlling temperature technology for asphalt pavement, and to improve the sustainable development of asphalt pavement, this paper reviews the research status of PCMs in asphalt pavement, both domestically and abroad. The results show that the thermal conductivity of the modified asphalt binder can reach 0.29–0.39 W/mK, and the thermal diffusivity can reach 0.2–0.3 mm2/s, but the influence on the viscosity of the asphalt is limited, and both are less than 2000CP. The durability and thermal stability of the modified asphalt mixture are improved, and the maximum temperature can be lowered by 9 °C, which effectively reduces the occurrence of hightemperature stress. This review will help to better understand the function of PCMs and promote the sustainable development of green and environmentally friendly asphalt pavement.

Evaluating the effects of cement asphalt emulsion paste as a grouting material on the fatigue and thermal contraction properties of semi-flexible asphalt composite pavement

Semi-flexible asphalt (SFA) composite pavement is widely used in pavement engineering for its high load-bearing capacity, enhanced durability, impermeability, and resistance to fuel and oil spillage. However, its cracking resistance remains a concern, influenced by repeated vehicle loads and thermal contraction stresses. This study investigates the impact of cement asphalt emulsion paste as a grouting material on SFA’s fatigue and thermal contraction. SFA specimens were grouted with cement paste (CP) containing 20%, 40%, and 60% asphalt emulsion. These specimens were subjected to indirect tensile strength, resilient modulus, and fatigue tests, along with evaluation of thermal contraction behavior using coefficient of thermal contraction. Results indicated that adding asphalt emulsion to CP enhances the flexibility and ductility of SFA but reduces its indirect tensile strength and stiffness modulus. A mathematical model was introduced to predict the resilient modulus based on indirect tensile strength tests. The fatigue analysis demonstrated significant difference in fatigue resistance between SFA and AC16 materials. Incorporating asphalt emulsion also improved SFA’s thermal contraction behavior.

The potential of solid waste as a viable replacement for quartz sand in the ultra-high performance concrete: a comprehensive review

ABSTRACT Ultra-high performance concrete (UHPC) is renowned for its dense structure and exceptional mechanical properties, including durability. However, its widespread adoption is hindered by the high costs of essential fine aggregates like quartz sand and the associated environmental impact. Additionally, using industrial solid waste raises pollution concerns, contradicting global sustainability goals. Numerous studies have shown that it is feasible to produce eco-friendly UHPC with excellent performance by solid waste. This comprehensive review thoroughly investigates the workability, mechanical properties (including compressive strength, flexural strength and tensile strength), microstructure (encompassing pore structure and interfacial transition zone) and durability of UHPC incorporating solid waste in its production. The paper emphasizes the critical influence of factors, such as the solid waste substitution rate and the physical and chemical properties of solid waste. The paper also outlines the current application of UHPC with solid waste in pavement engineering and highlights the urgent need to address challenges such as high production costs, the lack of design standards and construction technique deficiencies, thereby emphasising the urgency surrounding the design and advancement of solid waste-incorporated UHPC in pavement engineering as it can reduce the environmental impact of traditional UHPC and improve the efficient use of solid waste.

Development of sustainable, high strength slag based alkali activated pavement quality concrete using agro-industrial wastes: properties and life cycle analysis

ABSTRACT Alkali-activated binder-based composites offer a sustainable alternative to cement-based systems, giving a conservational approach. This research incorporated rice-husk ash (RHA) and waste foundry sand (WFS) into ground-granulated-blast-furnace slag (GGBS)-based alkali-activated high-end pavement-quality concretes, optimising with air-curing for expressways, freeways and runway applications. This study promotes sustainable construction by efficiently utilising agro-industrial wastes (i.e., RHA, GGBS and WFS) promoting a global initiative to decrease environmental impact and to adopt eco-friendly pavement construction methods. This study systematically substitutes the primary binder-GGBS with RHA at 5% intervals. The binder was alkali-activated using measured amounts of liquid sodium silicate and sodium hydroxide flakes prepared at an activator modulus (Ms) of 1.25 by maintaining the sodium oxide dosage of 4.0% with respect to the total binder. In particular, 15% RHA (with 85% GGBS) and 20% WFS (with 80% river sand) satisfy mechanical design standards for workability and strength. A comprehensive life cycle analysis using extensive literature support assesses the environmental impact and sustainability of developed pavement quality composites, showing over 70% less energy use than Portland cement-based pavement mixes. Therefore, employing RHA and WFS in air-cured alkali-activated composite manufacture shows that agro-industrial waste components may be used to make sustainable, cost-effective pavement construction. This study effectively advocates for a shift towards environmentally responsible pavement construction, contributing to a greener, more sustainable future in concrete pavement engineering.

Cement-based and alkali-activated controlled low-strength materials made from cup lump rubber for use as road materials

This research explores the use of cup lump rubber (CLR), an agricultural by-product, as a component in controlled low-strength material (CLSM) for pavement applications in road construction. Two distinct CLSM mixtures were developed: one based on cement and the other on alkali activation. The study evaluated the workability, mechanical properties, and microstructures of both CLSM formulations. Key fresh properties, including slump flow, setting time, and bleeding, were analysed to assess their impact on the self-compaction process. Mechanical characteristics such as unconfined compressive strength, resilient modulus, and wave velocities were also measured. Some CLSM mixtures, both cement-based and alkali-activated, were found to meet the requirements for soil cement bases and subbases. Notably, the resilient modulus values showed significant improvement after 28 days, with certain mixtures achieving subbase-quality gravel standards. The study concludes by recommending the use of both cement-based and alkali-activated CLSMs in pavement design, highlighting their potential to enhance the field of pavement engineering.

Study on the low-temperature performance of chemically cross-linked polyethylene composite modified asphalt binder

The great disadvantage of polyethylene modified asphalt in terms of low-temperature performance has led to limitations in its large-scale application. The aim of this study was to investigate whether peroxides and silane coupling agents can significantly improve the low-temperature performance of modified asphalt through chemical cross-linking. Therefore, nine experimental groups were designed through orthogonal experiments, and the low-temperature properties of matrix asphalt, control asphalt, and nine modified asphalt groups were evaluated by using a bending beam rheometer, and microscopic analyses were carried out using FM, SEM, and FTIR tests. The experimental results showed that the 5 °C ductility increased to a maximum of 136.3 % and 166.2 % compared to the matrix asphalt and control asphalt, and the cross-linking between the polymer and the asphalt was enhanced microscopically. Chemical cross-linking of asphalt and polymer using peroxide and silane coupling agent provides a feasible method to expand the application of polymer in pavement engineering.

Steel slag aggregate property improvement in cold mixed asphalt mixtures through surface modification treatment

Expanding the resource utilization of solid waste steel slag (SS) is necessary to reduce the environmental impact of stockpiling and disposal. However, SS has the shortcomings of poor volume stability and poor water damage resistance, which limits its application in pavement engineering, especially in cold mixed asphalt mixtures. In this work, hexamethylene diisocyanate biuret (containing -NCO) and hydroxyl acrylic resin (containing -OH) were utilized as raw materials to prepare NCO-OH composite modifier (NOCM) by regulating NCO/OH (0.75, 1.00, 1.25, and 1.50) for SS surface modification. Surface morphology, adhesion properties, aggregate properties, and prepared mixture properties of SS before and after modification were investigated, respectively, and the curing mechanism of NOCM and action mechanism were analyzed. Results show that the cured NOCM forms a dense modified film on the SS surface, and the rough surface and large pores of SS were blocked, which effectively hindered the free oxides in SS from coming into contact with water. NOCM reduced the apparent relative density, water absorption, and wear loss of SS. NOCM with NCO/OH=1.25 was recommended to modify SS to obtain the most balanced properties, at which time MSS has a roughest surface to mechanically interlock with asphalt, and there are excess -NCO groups to chemically bond with polar components in asphalt. Mixtures prepared at the optimum NCO/OH had good water damage resistance, freeze-thaw cycle resistance, and volume stability. Applying NOCM-modified SS in environment-friendly cold mixed asphalt mixtures can effectively improve pavement durability and bring good environmental benefits.

Experimental Research on the effect of Structural glass fibre in the Tensile Strength Ratio and Retained Stability of VG40 Bituminous mix

Abstract The tensile characteristics of bituminous mix are a major concern for pavement engineers due to the cracking issues. The cracking infuses moisture into the pavement layers which affects the stability of the pavement. The measures to reduce the fatigue tensile cracks have to be taken in the design stage rather than the maintenance stage. The tensile properties and the stability are evaluated using the indirect tensile strength and retained stability tests. While fibers improve the tensile properties of pavement, the S-glass fiber possessing superior mechanical strength has been found to have less scope in past studies. Thus, the current study primarily focuses on examining how Structural Glass fiber reinforced in VG-40 bituminous concrete mix improves the tensile and moisture susceptibility characteristics. The Marshall technique of bituminous mix design was used to establish the ideal bitumen content for casting the bituminous concrete specimens. At the ideal bitumen content, the attributes of bituminous mixes were established. Tensile strength ratio (TSR), Indirect Tensile Strength (ITS), and Retained Stability of modified bituminous concrete mix were assessed. It is observed that the bituminous mix prepared by adding the S-Glass fibre of 8mm length by 3% weight of aggregates exhibited higher ITS, TSR, and Retained stability. The bituminous mix with glass fiber incorporated had higher retained stability (95.36%) than the conventional mix (91.17%). The results of the Indirect Tensile Strength Test revealed that the mix with glass fiber added had a higher Tensile Strength Ratio (92%) compared to the mix with conventional fiber (87%). A noteworthy outcome that creates opportunities for more study in this field is the rise of 5% TSR and 4.66% maintained the stability of the glass fiber-modified bituminous mix.

Innovations in recycled construction materials: paving the way towards sustainable road infrastructure

The expansive development of infrastructure has led to increased consumption of virgin aggregates in road construction, resulting in significant environmental impacts. To address this issue, there is a pressing need for sustainable alternatives that utilize recycled materials in pavement applications. This paper presents a comprehensive review of a decade-long research program focused on the development and evaluation of sustainable pavement materials, such as recycled and waste aggregates, industrial by-products, and natural fibers. The research encompassed a wide range of innovative materials and technologies, such as geopolymer-stabilized recycled aggregates, cement-stabilized waste materials, natural additive-modified cement stabilization, and recycled aggregate-geogrid reinforcement systems. The experimental framework employed a combination of mechanical testing, durability assessment, microstructural analysis, and environmental safety evaluation to assess the performance and sustainability of these materials. The key findings demonstrated the superior mechanical properties, improved durability, and environmental suitability of the recycled materials compared to conventional virgin aggregates. The successful implementation of these sustainable solutions in real-world projects highlights their potential to reduce the environmental footprint of road infrastructure development. Furthermore, the paper discussed the practical implications of the research outcomes for pavement design and construction, as well as future research directions to further advance the field of sustainable pavement engineering. The findings of this research report can be used as guidance for researchers, practitioners, and policymakers seeking to upcycle the widespread adoption of recycled materials in road application and contribute to the development of a more sustainable and resilient transportation infrastructure.

Evaluation of compatibility in bio-oil and zinc oxide modified asphalt to facilitate waste recycling

The promotion of bio-oil (Bo) waste in road asphalt modification has significant potential for application and environmental benefits. The thermal storage stability, rheological properties, and nano-microstructural characteristics of the composite modified asphalt system were investigated. The Cole-Cole diagram effectively evaluated the compatibility of modified asphalt, demonstrating high sensitivity and accuracy. Significant interaction energy was observed between zinc oxide (ZnO) and asphaltene, with asphaltene primarily acting as an electron donor during its interaction with ZnO. The three organic small molecules in Bo primarily interacted with the polar molecules in asphalt through ester groups, exhibiting varied interaction effects with asphaltene. Bo influenced the colloidal structure of asphalt mainly through diffusion and molecular intercalation. When Bo constituted 4% of the composite modified asphalt system, the organic small molecules from Bo fully diffused within the asphalt matrix, inhibiting the stacking of aromatic rings in the asphaltene structure and enhancing the aromatic activity of asphaltene. Bo improved the compatibility between ZnO and asphalt primarily through dilution and balancing effects from the perspective of asphaltene components. The research results can promote the promotion and application of bio-oil waste in asphalt pavement engineering, which is an important measure to practice sustainable development.

Resilient roads in challenging terrain: a case study of Siddhartha highway in Nepal

Nepal is a country known for its diverse and challenging topography, and it relies heavily on a robust road infrastructure network to connect its remote regions and urban centers. This study addresses the critical need for enhanced road safety and infrastructure resilience on the Siddhababa road section of the Siddhartha Highway, Nepal, notorious for its high accident rates and susceptibility to landslides. Given the road's strategic importance in connecting remote regions and its challenging topographical conditions, our research aimed to identify the most suitable pavement type to mitigate these issues. Through a detailed examination incorporating eight different soil tests, alongside evaluations of traffic loads, weather conditions, and existing pavement performance, we adopted a comparative analysis methodology to assess the viability of flexible versus rigid pavements within this unique context. Results revealed that the soil composition and environmental conditions of the Siddhababa section significantly influence pavement performance, with specific gravity, moisture content, and California Bearing Ratio (CBR) tests indicating a nuanced suitability for both pavement types under varying circumstances. Our analysis concluded that, despite the economic and staged reinforcement benefits of flexible pavements, the durability, safety, and maintenance considerations favor the adoption of rigid pavement for the Siddhababa road section. However, acknowledging the economic constraints, a hybrid approach is recommended, emphasizing rigid pavements for the most vulnerable sections and flexible pavements elsewhere. This study contributes to the pavement engineering field by providing a model for pavement type selection in mountainous regions, aiming to enhance road safety and durability amidst challenging environmental conditions.

Optimizing modified asphalt binder performance at high and intermediate temperatures using experimental and machine learning approaches

Asphalt pavements are highly susceptible to temperature-induced stresses, which can lead to significant deterioration such as rutting at high temperatures and cracking at low temperatures. These challenges are exacerbated by global warming, which raises pavement temperatures and potentially accelerates material degradation. Addressing these issues, this study utilizes two base asphalt binders, PG 64–22 and PG 76–22, and the use of phase change materials (PCMs) like styrene-butadiene-styrene, lignin, and ground tire rubber to enhance the resilience of asphalt binders. Through an extensive experimental investigation, which includes aging processes such as rolling thin film oven and pressure aging vessel, alongside a series of both traditional and advanced tests (including dynamic shear rheometer, rotational viscometer, Fourier transform infrared spectroscopy, among others), a comprehensive set of 51 samples were analysed. By incorporating these binders and PCMs into asphalt, the experimental work demonstrated a significant improvement in the thermal and mechanical properties, effectively increasing resistance to high-temperature rutting and fatigue cracking. Complementing the experimental approach, machine learning models were employed to predict complex asphalt binder properties, such as combustion value and failure temperature, based on the experimental data. This dual approach fills a significant gap in the literature by offering a more efficient method and prediction model for optimizing pavement materials amidst climate change, while also promising to significantly advance pavement technology by enhancing material design and performance.

Long-chain alkyl emulsifiers induced asphalt particle dispersion: Lipophilicity-enhancement effect

The structure of the lipophilic groups within the emulsifiers plays a pivotal role in uniformly dispersing the asphalt particles in the water regarding the storage and thermal stability of emulsified asphalt. This study combines laboratory experiments with molecular dynamics simulations, investigating the influence of alkyl chain length on emulsified asphalt dispersion and behavior at the oil/water interface. The results reveal that long-chain alkyl emulsifiers (C16TAC and C18TAC) present a lipophilicity-enhancement effect. This phenomenon increases the free volume fraction of emulsified asphalt, reducing viscosity and improving the compatibility of emulsion. This results in smaller asphalt particle sizes (D50 = 1.982μm) and more uniform dispersion. Vigorously lipophilic long-chain alkyl promotes emulsifier molecular migration towards the asphalt phase, thickening the oil/water interfacial layer and reducing interfacial tension and energy by more than 90 %. Notely, long-chain alkyl emulsifiers establish strong hydrogen bonds with water molecules, leading to water molecule aggregation into a hydration layer. Laboratory experimental results demonstrate that this lipophilicity-enhancement effect significantly improves the storage and thermal stability of emulsified asphalt. This research recommends long-chain alkyl emulsifiers in developing and designing energy-efficient pavement engineering regarding high-performance emulsified asphalt.

Statistical and machine learning models for predicting spalling in CRCP

Continuously reinforced concrete pavement (CRCP), crucial for the resilience of transportation infrastructure owing to its continuous steel reinforcement, confronts a critical challenge in the form of spalling—a distress phenomenon posing a threat to pavement durability and overall structural integrity. The detachment or breakage of concrete from the surface compromises CRCP’s functionality and raises safety concerns and escalating maintenance costs. To address this pressing issue, our study investigates the multifaceted factors influencing spalling, employing a comprehensive approach that integrates statistical and machine learning techniques for predictive modeling. Descriptive statistics meticulously profile the dataset, emphasizing age, thickness, precipitation, temperature, and traffic parameters. Regression analysis unveils key relationships, emphasizing the significance of age, annual temperature, annual precipitation, maximum humidity, and the initial International Roughness Index (IRI) as influential factors. The correlation matrix heatmap guides feature selection, elucidating intricate interdependencies. Simultaneously, feature importance analysis identifies age, Average Annual Daily Traffic (AADT), and total pavement thickness as crucial contributors to spalling. In machine learning, adopting models, including Gaussian Process Regression and ensemble tree models, is grounded in their diverse capabilities and suitability for the complex task at hand. Their varying predictive accuracies underscore the importance of judicious model selection. This research advances pavement engineering practices by offering nuanced insights into factors influencing spalling in CRCP, refining our understanding of spalling influences. Consequently, the study not only opens avenues for developing improved predictive methodologies but also enhances the durability of CRCP infrastructure, addressing broader implications for informed decision-making in transportation infrastructure management. The selection of Gaussian Process Regression and ensemble tree models stems from their adaptability to capture intricate relationships within the dataset, and their comparative performance provides valuable insights into the diverse predictive capabilities of these models in the context of CRCP spalling.

Enhancing lateritic soil for sustainable pavement subbase with polymer-modified cement: A comparative study of styrene butadiene rubber and styrene acrylic latex applications

This study evaluates styrene butadiene rubber (SBR) and styrene acrylic latex (SA) as modifiers in cement-treated subbase materials (CTSB) to enhance mechanical properties and reduce cement usage sustainably. Optimal ratios for stabilizing sub-standard lateritic soils were identified, reducing water demand and increasing mechanical strength in polymer-modified cement pastes. A 10 % SA and a 15 % SBR as cement replacement by mass significantly improved bearing strength and strain capacities in CTSB, signifying enhanced flexibility and elasticity. Despite slight changes in compaction characteristics, the study identified 1.6 % SA and 2.4 % SBR as optimal binder (i.e., polymer-cement mixture) contents, compared to 3.3 % cement for conventional CTSB with similar unconfined compressive strength standards. SBR-enriched CTSB exhibited superior resilient modulus, indicating stronger inter-particle bonding. The integration of SA and SBR reduced capillary rise and enhanced moisture stability. This sustainable approach enhances pavement durability and reduces CO2 emissions by minimizing cement use. The findings emphasize the potential of polymer-modified CTSB for cost-effective and environmentally friendly road construction, offering significant implications for advancing pavement engineering materials and promoting eco-friendly practices within the industry.

From ensemble learning to deep ensemble learning: A case study on multi-indicator prediction of pavement performance

Recently, Big data analytics approaches were combined with emerging machine learning techniques, which can provide more sophisticated insights into information-intensive activity. Compared to traditional shallow-architecture machine learning algorithms, Deep learning could excavate more potential information from the raw features. However, its powerful representational capacity relies on the support of enormous samples. The ensemble trees system performs superior on small sample problems due to better generalization capacity. To merge both benefits of deep learning and ensemble tree system, this paper developed a deep ensemble algorithm applied to multiple indicators prediction of pavement performance, including International Roughness Index and pavement 3-layer modulus. The deep ensemble algorithm is developed by merging a deep neural network (DNN) with the decision manifold property of the decision trees (TabNet) into a cascade ensemble system, combined with a sliding window algorithm to extract dependency information from raw data. During the training stage, the Bayesian Optimization Algorithm (BOA) is used to search for the optimal combination of sub-decision makers in the cascade ensemble. And equipped with GPU, it can speed up by 2.6–4.0 times. In the case study of pavement engineering, with sufficient training samples, it can achieve an average accuracy of 98.74 %, higher than DNN (97.49 %) and XGBoost (96.12 %) in predicting pavement indicators. With insufficient training samples, it can achieve an accuracy improvement of 12 % than XGBoost (75 %) and 24.5 % than DNN (62.5 %).