Latest open access articles published in Intelligent, Resilient, and Integrated Infrastructure Systems at https://www.acadlore.com/journals/IR2IS https://www.acadlore.com/journals/IR2IS Acadlore en Creative Commons Attribution(CC - BY) IR2IS support@acadlore.com https://www.acadlore.com/article/IR2IS/2026_1_1/ir2is010102 Spectrophotometric online sensors are widely deployed for nutrient monitoring, but their accuracy deteriorates under turbid conditions. Existing compensation algorithms, whether single-coefficient linear subtractions or log-ratio expressions embedded in the analytical formula, often introduce new errors at low turbidity or saturate beyond the calibration range. This study develops a unified residual-learning framework that retains the certified instrument output as a baseline and adds a data-driven correction. The framework is applied to three nutrient parameters: ammonia nitrogen (NH4-N), total nitrogen (TN), and total phosphorus (TP). Controlled turbidityaddition experiments yielded 72 NH4-N samples (4 concentrations × 6 nephelometric turbidity units (NTU) levels, 0–500 NTU), 56 TN samples (4 concentrations × 7 NTU levels, 0–500 NTU, on two instruments), and 219 TP samples (4 concentrations × 8 NTU levels, 0–1034 NTU). Three model families were trained on each dataset: a three-parameter polynomial (Approach A), a Ridge regression on eight engineered log-features (Approach B), and a residual eXtreme Gradient Boosting (XGBoost) with monotone constraints (Approach C). Performance was evaluated by random 5-fold and leave-one-concentration-out (LOCO) cross-validation, stratified by low (4-N. The XGBoost variant achieved 45.2%, 87.2%, and 75.6%, respectively. At the saturated 2.35 mg·L−1 NH4-N point, where the embedded log-quadratic formula fails, XGBoost reduced the MAE from 1.885 to 0.055 mg·L−1 (97.1% reduction), indicating that residual learning can correct this observed in-grid failure more effectively than the fixed-coefficient baseline tested here. LOCO testing showed analyte-specific generalization: TP retained its random-fold gains, whereas NH4-N exhibited concentrationextrapolation limits. The Ridge model offers predictable, interpretable correction; the XGBoost variant provides additional accuracy where calibration saturation dominates the error. Joint reporting of random-fold and LOCO accuracy is recommended as standard practice for AI-augmented water-quality sensors. 03-09-2026 <![CDATA[ <p>Spectrophotometric online sensors are widely deployed for nutrient monitoring, but their accuracy deteriorates under turbid conditions. Existing compensation algorithms, whether single-coefficient linear subtractions or log-ratio expressions embedded in the analytical formula, often introduce new errors at low turbidity or saturate beyond the calibration range. This study develops a unified residual-learning framework that retains the certified instrument output as a baseline and adds a data-driven correction. The framework is applied to three nutrient parameters: ammonia nitrogen (NH₄-N), total nitrogen (TN), and total phosphorus (TP). Controlled turbidityaddition experiments yielded 72 NH₄-N samples (4 concentrations × 6 nephelometric turbidity units (NTU) levels, 0–500 NTU), 56 TN samples (4 concentrations × 7 NTU levels, 0–500 NTU, on two instruments), and 219 TP samples (4 concentrations × 8 NTU levels, 0–1034 NTU). Three model families were trained on each dataset: a three-parameter polynomial (Approach A), a Ridge regression on eight engineered log-features (Approach B), and a residual eXtreme Gradient Boosting (XGBoost) with monotone constraints (Approach C). Performance was evaluated by random 5-fold and leave-one-concentration-out (LOCO) cross-validation, stratified by low (<100 NTU) and high (≥100 NTU) turbidity. Under random-fold testing, the Ridge production model reduced the mean absolute error (MAE) by 42.5% for TN, 75.6% for TP, and 7.6% for NH₄-N. The XGBoost variant achieved 45.2%, 87.2%, and 75.6%, respectively. At the saturated 2.35 mg·L⁻¹ NH₄-N point, where the embedded log-quadratic formula fails, XGBoost reduced the MAE from 1.885 to 0.055 mg·L⁻¹ (97.1% reduction), indicating that residual learning can correct this observed in-grid failure more effectively than the fixed-coefficient baseline tested here. LOCO testing showed analyte-specific generalization: TP retained its random-fold gains, whereas NH₄-N exhibited concentrationextrapolation limits. The Ridge model offers predictable, interpretable correction; the XGBoost variant provides additional accuracy where calibration saturation dominates the error. Joint reporting of random-fold and LOCO accuracy is recommended as standard practice for AI-augmented water-quality sensors.</p> ]]> Residual Machine Learning Compensates Turbidity Interference and Corrects In-Grid Saturation in Online Spectrophotometric Monitoring of Ammonia Nitrogen, Total Nitrogen, and Total Phosphorus junting peng siyu dai shuihan mo feng lin qidong yin kai he doi: 10.56578/ir2is010102 Intelligent, Resilient, and Integrated Infrastructure Systems 03-09-2026 Intelligent, Resilient, and Integrated Infrastructure Systems 03-09-2026 2026 1 1 Article 26 10.56578/ir2is010102 https://www.acadlore.com/article/IR2IS/2026_1_1/ir2is010102 https://www.acadlore.com/article/IR2IS/2026_1_1/ir2is010101 Airport service quality (ASQ) plays a central role in shaping passenger satisfaction, airport competitiveness, and the long-term performance of aviation infrastructure systems. Existing studies increasingly recognise airports not merely as service environments but as integrated infrastructure systems in which transportation networks, operational processes, information systems, and passenger-facing functions interact dynamically. However, current knowledge remains fragmented across service domains, evaluation methods, passenger characteristics, and industry benchmarking practices. This study investigates the analytical foundations and research evolution of ASQ from an infrastructure systems perspective. A systematic literature review was conducted following the PRISMA protocol using a Scopus-based corpus of 303 peer-reviewed publications published between 1976 and 2024, supplemented by major industry reports and benchmarking frameworks. The review synthesised evidence across airport service domains, service attributes, and key performance indicators (KPIs), passenger heterogeneity, survey methodologies, social media-based assessment approaches, and the interaction between airline and ASQ. The results showed that ASQ research has evolved from isolated service evaluation toward increasingly integrated and multi-dimensional assessment frameworks. Processing and non-processing service domains were found to exert asymmetric effects on passenger satisfaction, while substantial variations were identified across demographic, behavioural, geographic, and travel-related passenger profiles. The review further showed that industry benchmarking systems provide operational comparability but remain only partially aligned with academic analytical approaches. Several research gaps were identified, particularly in landside infrastructure evaluation, arrival-stage service assessment, integrated objective–subjective performance measurement, and system-level understanding of airport operations. The findings indicate that ASQ should be interpreted as an emergent property of interconnected infrastructure subsystems rather than as isolated service encounters. This study provides an integrated conceptual foundation for future research on intelligent, resilient, and evidence-based airport infrastructure management and supports more transparent and analytically grounded decision-making for airport operators, policymakers, and researchers. 03-07-2026 <![CDATA[ Airport service quality (ASQ) plays a central role in shaping passenger satisfaction, airport competitiveness, and the long-term performance of aviation infrastructure systems. Existing studies increasingly recognise airports not merely as service environments but as integrated infrastructure systems in which transportation networks, operational processes, information systems, and passenger-facing functions interact dynamically. However, current knowledge remains fragmented across service domains, evaluation methods, passenger characteristics, and industry benchmarking practices. This study investigates the analytical foundations and research evolution of ASQ from an infrastructure systems perspective. A systematic literature review was conducted following the PRISMA protocol using a Scopus-based corpus of 303 peer-reviewed publications published between 1976 and 2024, supplemented by major industry reports and benchmarking frameworks. The review synthesised evidence across airport service domains, service attributes, and key performance indicators (KPIs), passenger heterogeneity, survey methodologies, social media-based assessment approaches, and the interaction between airline and ASQ. The results showed that ASQ research has evolved from isolated service evaluation toward increasingly integrated and multi-dimensional assessment frameworks. Processing and non-processing service domains were found to exert asymmetric effects on passenger satisfaction, while substantial variations were identified across demographic, behavioural, geographic, and travel-related passenger profiles. The review further showed that industry benchmarking systems provide operational comparability but remain only partially aligned with academic analytical approaches. Several research gaps were identified, particularly in landside infrastructure evaluation, arrival-stage service assessment, integrated objective–subjective performance measurement, and system-level understanding of airport operations. The findings indicate that ASQ should be interpreted as an emergent property of interconnected infrastructure subsystems rather than as isolated service encounters. This study provides an integrated conceptual foundation for future research on intelligent, resilient, and evidence-based airport infrastructure management and supports more transparent and analytically grounded decision-making for airport operators, policymakers, and researchers. ]]> Airport Service Quality as an Integrated Infrastructure System: A Systematic Literature Review of Service Domains, Passenger Heterogeneity, Evaluation Frameworks, and Industry Benchmarking dileep dixit sanjay gupta doi: 10.56578/ir2is010101 Intelligent, Resilient, and Integrated Infrastructure Systems 03-07-2026 Intelligent, Resilient, and Integrated Infrastructure Systems 03-07-2026 2026 1 1 Article 1 10.56578/ir2is010101 https://www.acadlore.com/article/IR2IS/2026_1_1/ir2is010101