A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning
<p>Methodology.</p> "> Figure 2
<p>Feature-level fusion of features.</p> "> Figure 3
<p>Ensemble Isolation Forest model.</p> "> Figure 4
<p>Emissions by driver behavior.</p> "> Figure 5
<p>Spatial distribution of anomalies.</p> "> Figure 6
<p>Spatial emission hotspots. (<b>a</b>) CO emissions; (<b>b</b>) HC emissions; (<b>c</b>) NOx emissions.</p> "> Figure 7
<p>Emissions by hour of the day.</p> "> Figure 8
<p>Emissions by day of the week.</p> "> Figure 9
<p>Distribution of anomaly scores.</p> "> Figure 10
<p>Emissions levels: Anomaly vs. Non-anomaly.</p> "> Figure 11
<p>Emissions by hour of the day with anomalies.</p> "> Figure 12
<p>Emissions comparison after reducing speed limit.</p> ">
Abstract
:1. Introduction
- Telemetric Framework with Unsupervised Learning: We propose an advanced telemetric data-driven framework that integrates unsupervised learning techniques, such as Gaussian Mixture Models (GMMs) and Ensemble Isolation Forests (EIFs), to accurately assess vehicle emissions under real-world driving conditions. This approach allows for the identification of high-risk driving patterns and the detection of anomalies, providing a more nuanced understanding of emissions than traditional methods.
- VSP-Based Emissions Estimation: We introduce a method for estimating emissions based on Vehicle Specific Power (VSP) calculations, which allows for the measurement of pollutant levels under varying driving conditions. This method provides a more granular understanding of emissions, factoring in the instantaneous performance of vehicles during different driving scenarios.
- Scalable Framework: Our framework is designed to be scalable and adaptable to various urban environments. Through the integration of case study insights on speed-limit enforcement, we demonstrate how the framework can directly inform policy by identifying high-emission zones and providing actionable data. The findings indicate that speed regulation, as an intervention strategy, can significantly reduce emissions in urban areas. This highlights the framework’s potential to support urban planners and policymakers in implementing targeted, evidence-based strategies for improving air quality and mitigating environmental impacts.
2. Related Works
2.1. Driving Behavior and Emission
2.2. Vehicle Emissions and Environmental Impact
2.3. Vehicle Emission Assessment Through Telemetric Data
3. Methodology
3.1. Data Collection and Preprocessing
3.2. Feature Engineering
3.2.1. Road Grade
3.2.2. Vehicle Specific Power (VSP) Calculation
3.2.3. Event-Based Features
- Harsh Braking: This feature identifies instances where the vehicle experiences sudden and significant deceleration. Harsh braking is detected when the deceleration exceeds a predefined threshold, . The equation is
- Rapid Acceleration: Rapid acceleration is identified when the vehicle’s acceleration exceeds a specific threshold, . The equation is
- Overspeeding: This feature identifies instances where the vehicle exceeds the legal speed limit for the road segment. It is calculated as
3.2.4. Pattern-Based Features
- Speed Variance per Trip: This feature measures the variability in the vehicle’s speed during a trip, indicating how much the speed fluctuates. It is calculated as
- Rolling Acceleration Mean: This feature calculates the mean acceleration over a rolling window, providing insights into how the vehicle’s acceleration changes over time. It is given by
- Rolling Acceleration Standard Deviation: This feature measures the standard deviation of acceleration within a rolling window, indicating the variability in acceleration over time. It is calculated as
- Turning Rate: This feature measures the rate of change in bearing between consecutive data points, indicating sharp turns:
- Stop-and-Go Behavior: This feature is detected when the vehicle’s speed falls below 5 km/h and then increases:
3.2.5. Contextual-Based Features
- Night Driving: Trips occurring before 6 AM or after 6 PM are flagged as night driving:
- Rainy Driving: Driving in rainy conditions is detected based on precipitation data (PRCP):
- Weekend Driving: Driving on Saturday or Sunday is flagged as weekend driving.
3.3. Emission Estimation Using VSP Bins
3.3.1. VSP Range Determination
- Negative VSP values: Typically represent situations such as deceleration or driving downhill.
- Positive VSP values: Represent power demand from activities like acceleration and cruising.
3.3.2. Applying Emission Factors
3.4. Driving Pattern Clustering Using GMM
Algorithm 1: Clustering using GMM with EM Algorithm |
|
3.5. Anomaly Detection Using Ensemble Isolation Forest
- Traffic Risk Assessment: Identified anomalies are linked with dangerous driving patterns, helping to classify trips based on risk levels.
- Environmental Impact Analysis: Anomalies related to fuel consumption and emissions, such as sudden spikes in VSP or deviations in speed patterns, are flagged to better assess emission hotspots and inform mitigation strategies.
Model Architecture
- 1.
- Data Segmentation: The goal here is to select specific groups of features from the engineered features to train individual models. This involves choosing different feature subsets for different Isolation Forest models within the ensemble. The engineered features are segmented into several feature subsets, each capturing different dimensions of the data. The subsets include the following:
- (a)
- Kinematic Features: Speed, acceleration, and Vehicle Specific Power (VSP).
- (b)
- Emission Attributes: Concentrations of pollutants such as CO, HCs, and NOx.
- (c)
- Behavioral Indicators: Hard braking, rapid acceleration, and overspeeding.
- (d)
- Statistical Measures: Variance in speed and rolling statistics of acceleration.
- (e)
- Contextual Features: Temporal and environmental factors such as night driving, rainy conditions, sharp turns, and stop-and-go behavior.
- 2.
- Ensemble Isolation Forest Models: The purpose here is to train multiple Isolation Forest (IF) models, each on a different feature subset, to detect anomalies within that specific context. Below is the structure of each IS model:
- (a)
- Input Layer: This layer receives the feature subset for the particular IS model.
- (b)
- Isolation Tree Construction: The core of the Isolation Forest model is the construction of multiple binary trees, referred to as isolation trees. Each tree recursively partitions the data by randomly selecting a feature and a split point for that feature with the expectation that anomalies will be isolated in fewer splits compared to normal data points.For a dataset with n observations, each isolation tree, T, a random feature, (where k is the number of features), and a split value, , are chosen. The data are split into two subsets:This process continues until each data point is isolated in its leaf node or the maximum tree depth is reached.
- (c)
- Anomaly Scoring Mechanism: Once the trees are built, the model calculates how deeply a point is isolated in each tree. Anomalies tend to be isolated quickly, resulting in a shorter path length. The average path length for a point across all trees gives its anomaly score. Here, the expected path length, , for a point, x, in a dataset of size n is given by
- (d)
- Ensemble Aggregation: Since the Isolation Forest is an ensemble of isolation trees, the final anomaly score for each data point is the average of the scores across all trees in the ensemble:
- 3.
- Anomaly Score Aggregation: After each Isolation Forest model outputs an anomaly score, the scores from all models are aggregated to produce a final comprehensive score for each data point. Aggregation can be done through averaging or other methods like majority voting. In our work, we use score averaging:
- 4.
- Anomaly Classification: After aggregating the scores, a final threshold, , is applied to classify data points as anomalous or normal. The threshold can be dynamically adjusted based on the expected contamination rate or application-specific needs. If , the point x is classified as an anomaly. The choice of threshold is critical for controlling the trade-off between false positives and false negatives.
- : the point x is classified as an anomaly.
- : the point x is considered normal.
Algorithm 2: Ensemble Isolation Forest for Anomaly Detection |
Input: Telemetric dataset , Number of feature subsets M, Number of Isolation Forest (IF) models N, Anomaly threshold Output: Anomaly labels for each data point in D 3. Ensemble Aggregation: Aggregate the scores from all models by averaging them: 4. Anomaly Classification: Classify each data point x as anomalous if Otherwise, classify it as normal. 5. Return: Return the anomaly labels for each data point x in D. |
4. Results
4.1. Data Description
- Kinematic Features: Speed, acceleration, and Vehicle Specific Power (VSP), which capture the dynamics of vehicle operation.
- Emission Attributes: Concentrations of pollutants such as Carbon Monoxide (CO), Hydrocarbons (HCs), and Nitrogen Oxides (NOx), reflecting the environmental impact of driving behavior.
- Behavioral Indicators: Metrics for hard braking, rapid acceleration, and overspeeding, which are indicative of aggressive driving patterns.
- Statistical Measures: Variance in speed and rolling statistics of acceleration, providing insights into the consistency of driving behavior over time.
- Contextual Features: Temporal and environmental factors such as night driving, rainy conditions, sharp turns, and stop-and-go behavior, which contribute to understanding driving conditions.
4.2. Experiment Setup
- Contamination: This parameter represents the expected proportion of anomalies within the dataset, with values set at 0.1, 0.2, and 0.3.
- Number of Estimators (n_estimators): This denotes the number of decision trees included in the ensemble, with candidate values of 50, 100, and 150.
- Maximum Samples (max_samples): This determines the number of training samples for each base estimator, evaluated with ’auto’, 256, and 512.
4.3. Performance Evaluation
4.3.1. Clustering for Behavior Patterns
4.3.2. Emission Estimation
4.3.3. Anomaly Detection
4.4. Case Study
Case Study 1: Limiting Speed Limit
5. Discussion
6. Conclusions
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
References
- Wang, J.; Wang, R. The Impact of Urbanization on Environmental Quality in Ecologically Fragile Areas: Evidence from Hengduan Mountain, Southwest China. Land 2024, 13, 503. [Google Scholar] [CrossRef]
- Ghaffarpasand, O.; Beddows, D.; Ropkins, K.; Pope, F. Real-world assessment of vehicle air pollutant emissions subset by vehicle type, fuel and EURO class: New findings from the recent UK EDAR field campaigns, and implications for emissions restricted zones. Sci. Total Environ. 2020, 734, 139416. [Google Scholar] [CrossRef] [PubMed]
- Pacura, W.; Szramowiat-Sala, K.; Gołaś, J. Emissions from Light-Duty Vehicles—From Statistics to Emission Regulations and Vehicle Testing in the European Union. Energies 2023, 17, 209. [Google Scholar] [CrossRef]
- Rahman, S.A.; Fattah, I.R.; Ong, H.C.; Ashik, F.R.; Hassan, M.M.; Murshed, M.T.; Imran, M.A.; Rahman, M.H.; Rahman, M.A.; Hasan, M.A.M.; et al. State-of-the-art of establishing test procedures for real driving gaseous emissions from light-and heavy-duty vehicles. Energies 2021, 14, 4195. [Google Scholar] [CrossRef]
- Triantafyllopoulos, G.; Dimaratos, A.; Ntziachristos, L.; Bernard, Y.; Dornoff, J.; Samaras, Z. A study on the CO2 and NOx emissions performance of Euro 6 diesel vehicles under various chassis dynamometer and on-road conditions including latest regulatory provisions. Sci. Total Environ. 2019, 666, 337–346. [Google Scholar] [CrossRef]
- Hu, Z.; Lu, Z.; Song, B.; Quan, Y. Impact of test cycle on mass, number and particle size distribution of particulates emitted from gasoline direct injection vehicles. Sci. Total Environ. 2021, 762, 143128. [Google Scholar] [CrossRef]
- Smit, R.; Bluett, J. A new method to compare vehicle emissions measured by remote sensing and laboratory testing: High-emitters and potential implications for emission inventories. Sci. Total Environ. 2011, 409 13, 2626–2634. [Google Scholar] [CrossRef]
- Wong, P.k.; Vong, C.m.; Ip, W.f.; Wong, H.c. Preliminary study on telemetric vehicle emission examination. In Green Communications and Networks, Proceedings of the International Conference on Green Communications and Networks (GCN 2011), Gaudia, Spain, 26–29 September 2011; Springer: Berlin/Heidelberg, Germany, 2012; pp. 443–451. [Google Scholar]
- Ding, H.; Cai, M.; Lin, X.; Chen, T.; Li, L.; Liu, Y. RTVEMVS: Real-time modeling and visualization system for vehicle emissions on an urban road network. J. Clean. Prod. 2021, 309, 127166. [Google Scholar] [CrossRef]
- Merkisz, J.; Pielecha, J.; Radzimirski, S. New Trends in Emission Control in the European Union; Springer: Berlin/Heidelberg, Germany, 2014; Volume 4. [Google Scholar]
- Williams, M.; Minjares, R. A Technical Summary of Euro 6/VI Vehicle Emission Standards; ICCT: Washington, DC, USA, 2016. [Google Scholar]
- Hooftman, N.; Messagie, M.; Van Mierlo, J.; Coosemans, T. A review of the European passenger car regulations–Real driving emissions vs local air quality. Renew. Sustain. Energy Rev. 2018, 86, 1–21. [Google Scholar] [CrossRef]
- Dhital, N.B.; Wang, S.X.; Lee, C.H.; Su, J.; Tsai, M.Y.; Jhou, Y.J.; Yang, H.H. Effects of driving behavior on real-world emissions of particulate matter, gaseous pollutants and particle-bound PAHs for diesel trucks. Environ. Pollut. 2021, 286, 117292. [Google Scholar] [CrossRef]
- Ng, E.C.; Huang, Y.; Hong, G.; Zhou, J.L.; Surawski, N.C. Reducing vehicle fuel consumption and exhaust emissions from the application of a green-safety device under real driving. Sci. Total Environ. 2021, 793, 148602. [Google Scholar] [CrossRef] [PubMed]
- Fondzenyuy, S.K.; Turner, B.M.; Burlacu, A.F.; Jurewicz, C.; Usami, D.S.; Feudjio, S.L.T.; Persia, L. The Impact of Speed Limit Change on Emissions: A Systematic Review of Literature. Sustainability 2024, 16, 7712. [Google Scholar] [CrossRef]
- Zhang, R.; Chen, H.; Xie, P.; Zu, L.; Wei, Y.; Wang, M.; Wang, Y.; Zhu, R. Exhaust Emissions from Gasoline Vehicles with Different Fuel Detergency and the Prediction Model Using Deep Learning. Sensors 2023, 23, 7655. [Google Scholar] [CrossRef] [PubMed]
- Šarkan, B.; Jaśkiewicz, M.; Kubiak, P.; Tarnapowicz, D.; Loman, M. Exhaust Emissions Measurement of a Vehicle with Retrofitted LPG System. Energies 2022, 15, 1184–2022. [Google Scholar] [CrossRef]
- Huang, Y.; Ng, E.C.; Zhou, J.L.; Surawski, N.C.; Lu, X.; Du, B.; Forehead, H.; Perez, P.; Chan, E.F. Impact of drivers on real-driving fuel consumption and emissions performance. Sci. Total Environ. 2021, 798, 149297. [Google Scholar] [CrossRef]
- Mądziel, M. Modelling CO2 Emissions from Vehicles Fuelled with Compressed Natural Gas Based on On-Road and Chassis Dynamometer Tests. Energies 2024, 17, 1850. [Google Scholar] [CrossRef]
- Mislyuk, O.; Khomenko, E.; Yehorova, O.; Zhytska, L. Assessing risk caused by atmospheric air pollution from motor vehicles to the health of population in urbanized areas. East.-Eur. J. Enterp. Technol. 2023, 121. [Google Scholar] [CrossRef]
- Nunes, L.J.R. The Rising Threat of Atmospheric CO2: A Review on the Causes, Impacts, and Mitigation Strategies. Environments 2023, 11, 39751–39775. [Google Scholar] [CrossRef]
- Bronte-Moreno, O.; González-Barcala, F.J.; Muñoz-Gall, X.; Pueyo-Bastida, A.; Ramos-González, J.; Urrutia-Landa, I. Impact of air pollution on asthma: A scoping review. Open Respir. Arch. 2023, 5, 100229. [Google Scholar] [CrossRef]
- Lee, H.; Calvin, K.; Dasgupta, D.; Krinner, G.; Mukherji, A.; Thorne, P.; Trisos, C.; Romero, J.; Aldunce, P.; Barret, K.; et al. IPCC, 2023: Climate Change 2023: Synthesis Report, Summary for Policymakers. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change; Core Writing Team, Lee, H., Romero, J., Eds.; IPCC: Geneva, Switzerland, 2023. [Google Scholar]
- Zhang, L.; Wei, J.; Tu, R. Temporal-spatial analysis of transportation CO2 emissions in China: Clustering and policy recommendations. Heliyon 2024, 10, e24648. [Google Scholar] [CrossRef]
- Su, J. Research on the Impact of Automobile Exhaust on Air Pollution. In Proceedings of the 2022 International Conference on Urban Planning and Regional Economy (UPRE 2022), Online, 22–24 April 2022; pp. 497–501. [Google Scholar] [CrossRef]
- Wurtsbaugh, W.; Paerl, H.; Dodds, W. Nutrients, eutrophication and harmful algal blooms along the freshwater to marine continuum. Wiley Interdiscip. Rev. Water 2019, 6, e1373. [Google Scholar] [CrossRef]
- MALMQVIST, E. A major step toward cleaner air in the EU. ACID NEWS, June 2024; No. 2. [Google Scholar]
- Jan, D.; Felipe, R. Euro 7: The New Emission Standard for Light- and Heavy-Duty Vehicles in the European Union. 2024. Available online: https://theicct.org/wp-content/uploads/2024/03/ID-116-%E2%80%93-Euro-7-standard_final.pdf (accessed on 21 September 2024).
- NHTSA. Corporate Average Fuel Economy | NHTSA — nhtsa.gov. 2024. Available online: https://www.nhtsa.gov/laws-regulations/corporate-average-fuel-economy, (accessed on 21 September 2024).
- Agarwal, A.; Mustafi, N. Real-world automotive emissions: Monitoring methodologies, and control measures. Renew. Sustain. Energy Rev. 2021, 137, 110624. [Google Scholar] [CrossRef]
- Dimaratos, A.; Toumasatos, Z.; Doulgeris, S.; Triantafyllopoulos, G.; Kontses, A.; Samaras, Z. Assessment of CO2 and NOx Emissions of One Diesel and One Bi-Fuel Gasoline/CNG Euro 6 Vehicles During Real-World Driving and Laboratory Testing. Front. Mech. Eng. 2019, 5, 62. [Google Scholar] [CrossRef]
- Ghaffarpasand, O.; Pope, F.D. Telematics data for geospatial and temporal mapping of urban mobility: Fuel consumption, and air pollutant and climate-forcing emissions of passenger cars. Sci. Total Environ. 2023, 894, 164940. [Google Scholar] [CrossRef]
- He, Z.; Ye, G.; Jiang, H.; Fu, Y. Vehicle Emission Detection in Data-Driven Methods. Math. Probl. Eng. 2020, 2020, 4875310. [Google Scholar] [CrossRef]
- Wong, P.; Vong, C.; Ip, W.; Wong, H.C. Flexibility study on telemetric vehicle emission examination. Int. J. Satell. Commun. Policy Manag. 2012, 1, 220. [Google Scholar] [CrossRef]
- Koupal, J.; DenBleyker, A.; Manne, G.K.; Batista, M.H.; Schmitt, T. Capabilities and Limitations of Telematics for Vehicle Emissions Inventories. Transp. Res. Rec. 2021, 2676, 49–57. [Google Scholar] [CrossRef]
- Hind, G.W.; Ballantyne, E.E.; Stincescu, T.; Zhao, R.; Stone, D.A. Extracting dashcam telemetry data for predicting energy use of electric vehicles. Transp. Res. Interdiscip. Perspect. 2024, 27, 101189. [Google Scholar] [CrossRef]
- Hao, L.; Yin, H.; Wang, J.; Wang, X.; Ge, Y. Potential of big data approach for remote sensing of vehicle exhaust emissions. Sci. Rep. 2021, 11, 5472. [Google Scholar] [CrossRef]
- Mondal, S.; Gupta, A. Evaluation of driver Acceleration/Deceleration behavior at signalized intersections using vehicle trajectory data. Transp. Lett. 2022, 15, 350–362. [Google Scholar] [CrossRef]
- Zheng, F.; Li, J.; Van Zuylen, H.J.; Lu, C. Influence of driver characteristics on emissions and fuel consumption. IET Intell. Transp. Syst. 2019, 13, 1770–1779. [Google Scholar] [CrossRef]
- Hamidi, B.; Lajqi, N.; Hamidi, L. Modelling and sensitive analysis of the impact on telematics system in vehicles. IFAC-PapersOnLine 2016, 49, 232–236. [Google Scholar] [CrossRef]
- Zhou, M.; Jin, H.; Wang, W. A review of vehicle fuel consumption models to evaluate eco-driving and eco-routing. Transp. Res. Part D Transp. Environ. 2016, 49, 203–218. [Google Scholar] [CrossRef]
- Robinson, M.K.; Holmén, B. Hybrid-electric passenger car energy utilization and emissions: Relationships for real-world driving conditions that account for road grade. Sci. Total Environ. 2020, 738, 139692. [Google Scholar] [CrossRef] [PubMed]
- Azevedo, J.A.H.; Cassiano, D.R.; Bertoncini, B.V. Real driving cycles and emissions for urban freight transport. Front. Big Data 2024, 7, 1375455. [Google Scholar] [CrossRef]
- Zang, J.; Song, G.; Wu, Y.; Yu, L. Method for Evaluating Eco-Driving Behaviors Based on Vehicle Specific Power Distributions. Transp. Res. Rec. 2019, 2673, 409–419. [Google Scholar] [CrossRef]
- Shakib, M.N.; Shamim, M.; Shawon, M.N.H.; Isha, M.K.F.; Hashem, M.; Kamal, M. An adaptive system for detecting driving abnormality of individual drivers using Gaussian mixture model. In Proceedings of the 2021 5th International Conference on Electrical Engineering and Information Communication Technology (ICEEICT), Dhaka, Bangladesh, 18–20 November 2021; pp. 1–6. [Google Scholar]
- Ye, W.; Xu, Y.; Zhou, F.; Shi, X.; Ye, Z. Investigation of bus drivers’ reaction to ADAS warning system: Application of the Gaussian mixed model. Sustainability 2021, 13, 8759. [Google Scholar] [CrossRef]
- Yang, C.H.; Chang, C.C.; Liang, D. A novel GMM-based behavioral modeling approach for smartwatch-based driver authentication. Sensors 2018, 18, 1007. [Google Scholar] [CrossRef]
- Song, X.; Aryal, S.; Ting, K.; Liu, Z.; He, B. Spectral–Spatial Anomaly Detection of Hyperspectral Data Based on Improved Isolation Forest. IEEE Trans. Geosci. Remote Sens. 2021, 60, 5516016. [Google Scholar] [CrossRef]
- Moosavi, S.; Ramnath, R. Context-aware driver risk prediction with telematics data. Accid. Anal. Prev. 2023, 192, 107269. [Google Scholar] [CrossRef]
- Yao, R.; Sun, L.; Long, M. VSP-based emission factor calibration and signal timing optimisation for arterial streets. IET Intell. Transp. Syst. 2019, 13, 228–241. [Google Scholar] [CrossRef]
- Wang, J. Analysis of Shared Bicycle Usage based on K-Means and GMM Clustering Algorithm. In Proceedings of the 2021 2nd International Seminar on Artificial Intelligence, Networking and Information Technology (AINIT), Shanghai, China, 15–27 October 2021; pp. 92–96. [Google Scholar] [CrossRef]
- Lewandowski, S.; Ullrich, A. Measures to reduce corporate GHG emissions: A review-based taxonomy and survey-based cluster analysis of their application and perceived effectiveness. J. Environ. Manag. 2023, 325, 116437. [Google Scholar] [CrossRef]
Dataset | Features |
---|---|
Telemetric Dataset | trip_id, timestamp, speed, acceleration, latitude, longitude, bearing, driver_id, date, duration_seconds, distance_meters, city |
Road Segment Dataset | geometry, Road_Segment_ID, City, Start_Coordinates, End_Coordinates, Length, Speed_Limit, Number_of_Lanes, Road_Type, POI_Density, Distance_to_School, Distance_to_Hospital, Distance_to_Segment |
Weather Dataset | date, city, humidity, pressure, temperature, wind_direction, wind_speed, PRCP |
VSP Range (kW/t) | CO (mg/s/veh) | HC (mg/s/veh) | NOx (mg/s/veh) |
---|---|---|---|
VSP < −10 | 1.9025 | 0.0673 | 0.3437 |
−10 ≤ VSP < −2 | 2.0918 | 0.1030 | 0.5046 |
−2 ≤ VSP < 0 | 2.5419 | 0.1593 | 0.5562 |
0 ≤ VSP < 2 | 1.8237 | 0.2323 | 0.5855 |
2 ≤ VSP < 5 | 2.3533 | 0.1896 | 0.6916 |
5 ≤ VSP < 9 | 2.2451 | 0.2592 | 0.8216 |
9 ≤ VSP < 13 | 2.6964 | 0.3180 | 1.0906 |
13 ≤ VSP < 17 | 4.0725 | 0.4383 | 1.1764 |
17 ≤ VSP < 20 | 3.9979 | 0.5472 | 1.3588 |
VSP ≥ 20 | 4.5135 | 0.5174 | 1.4514 |
Task | Silhouette Score | DBI | CHI |
---|---|---|---|
Driving pattern clustering | 0.729 | 0.888 | 59,367,387.3 |
Class | Accuracy | Precision | Recall | F1-Score |
---|---|---|---|---|
Class 0 | 0.87 | 0.88 | 0.85 | 0.86 |
Class 1 | 0.86 | 0.89 | 0.87 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Muhammad, A.S.; Wang, C.; Chen, L. A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning. Vehicles 2024, 6, 2170-2194. https://doi.org/10.3390/vehicles6040106
Muhammad AS, Wang C, Chen L. A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning. Vehicles. 2024; 6(4):2170-2194. https://doi.org/10.3390/vehicles6040106
Chicago/Turabian StyleMuhammad, Auwal Sagir, Cheng Wang, and Longbiao Chen. 2024. "A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning" Vehicles 6, no. 4: 2170-2194. https://doi.org/10.3390/vehicles6040106
APA StyleMuhammad, A. S., Wang, C., & Chen, L. (2024). A Telemetric Framework for Assessing Vehicle Emissions Based on Driving Behavior Using Unsupervised Learning. Vehicles, 6(4), 2170-2194. https://doi.org/10.3390/vehicles6040106