# Fuel Consumption Prediction for Construction Trucks: A Noninvasive Approach Using Dedicated Sensors and Machine Learning

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Methodology and Prediction Models

#### 2.1. Data Acquisition

- Road grade: As one may imagine, a heavy vehicle, given its weight, can rapidly develop speed on a descending road without needing the engine and oppositely requiring a considerable amount of mechanical energy to climb ascending roads. Thus, road grade, which can be monitored by tracking the inclination of the vehicle, together with vehicle moving speed or acceleration, is a promising indicator of the engine need. Thus, since vehicle inclination empirically oscillates between 5 Hz and 15 Hz, it is desired to configure a low-pass filter to attenuate road irregularities and roughness at that frequency range either by postprocessing data or configuring built-in sensor filters;
- Vehicle acceleration: As previously mentioned, a vehicle’s moving acceleration, together with the road grade and cargo weight, can be a promising indicator of the engine’s need (and consequently fuel consumption). Movement acceleration can be obtained by sampling acceleration data with a low pass filter to exclude vibration, road roughness and irregularities. Additionally, as an experimental project, having higher-frequency data also allows perceiving the engine’s rotational speed by carrying through a frequency analysis of the acceleration signal within 13–83 Hz, which corresponds to 800 to 5000 rotations per minute. According to Nyquist’s theorem, the signal must be sampled at, at least, twice the frequency of the original signal, thus at 166 Hz. Moreover, to classify road quality and roughness, which have a substantial impact on fuel consumption, a sampling frequency of around 250 Hz [19,20] is required to measure the vehicle’s frequency vibration. This way, three datasets are defined (vehicle moving speed, road quality/roughness, and motor speed), which can be retrieved by applying three different filters;
- Vehicle global position: Gathering position data allows calculating the total distance traveled, as well as the average speed if the data are timestamped. Distance, together with time, is what the simpler tools often use to estimate fuel consumption, being able to provide a rough estimate of consumption on light vehicles, so one could expect these data to increase the accuracy of the prediction model. Given that the maximum speed allowed on highways in Portugal is 120 km/h and considering the desired 5 m positioning updates, the sampling frequency of this parameter is required to be at least 6.67 Hz. Moreover, with awareness of the road being used together with real-time road quality assessment, one could create a map of roads annotated with the corresponding pavement surface regularity to provide the prediction algorithm with a more accurate and case-specific fuel estimation;
- Cargo weight: As any moving body, the vehicle’s weight influences its inertia and momentum, which in turn dictate the amount of mechanical energy the engine is forced to use to increase or maintain the vehicle’s speed, thus having a strong influence on fuel consumption. Therefore, monitoring the load weight is very important. However, as the load does not vary continuously, it is only required to weigh the load when the truck is loaded or unloaded, allowing these data to be input by a user when a sensor is not present. Measuring the cargo weight, apart from providing data to the fuel prediction model, is also an opportunity to digitize and automate cargo weighing, which is very useful for operations (e.g., material stocking and productivity evaluation), which, at the time, is still mostly a manual process.

#### 2.2. Data Storage and Communication

#### 2.3. Machine Learning

^{2}, comprising the correlation between the observed and the predicted values [26]:

## 3. Experimental Project

#### 3.1. Cyber-Physical Systems

#### 3.2. Data Storage and Communications

#### 3.3. Experimental Setup

## 4. Results and Discussion

#### 4.1. Raw Data and Data Preparation

#### 4.2. Estimation Performance

## 5. Conclusions and Future Work

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Abbreviations

ANN | Artificial Neural network |

API | Application User Interface |

CAN | Controller Area Network |

CSV | Comma-Separated Values |

GNSS | Global Navigation Satellite System |

ML | Machine Learning |

MR | Multiple Regression |

OBD | On-Board Diagnostic |

OCR | Optical Character Recognition |

OTA | Over-The-Air |

RF | Random Forest |

RPM | Rotations Per Minute |

SCP | Secure Copy Protocol |

SSH | Secure Shell |

SVM | Support Vector Machine |

## References

- Caterpillar. Caterpillar Performance Handbook; Caterpillar: Peoria, IL, USA, 2018. [Google Scholar]
- Correia, A.; Cortez, P.; Tinoco, J.; Marques, R. Artificial Intelligence Applications in Transportation Geotechnics. Geotech. Geol. Eng.
**2013**, 31, 861–879. [Google Scholar] [CrossRef] - Parente, M.; Correia, A.; Cortez, P. Use of DM techniques in earthworks management: A case study. In Proceedings of the Geo-Hubei 2014 International Conference on Sustainable Civil Infrastructure, Yichang, China, 20–22 July 2014. [Google Scholar] [CrossRef]
- Hola, B.; Schabowicz, K. Estimation of earthworks execution time cost by means of artificial neural networks. Autom. Constr.
**2010**, 19, 570–579. [Google Scholar] [CrossRef] - Tinoco, J.; Gomes Correia, A.; Cortez, P.; Toll, D.G. Stability condition identification of rock and soil cutting slopes based on soft computing. J. Comput. Civ. Eng.
**2018**, 32, 04017088. [Google Scholar] [CrossRef] [Green Version] - Tinoco, J.; Correia, A.; Cortez, P. Application of data mining techniques in the estimation of the uniaxial compressive strength of jet grouting columns over time. Constr. Build. Mater.
**2011**, 25, 1257–1262. [Google Scholar] [CrossRef] [Green Version] - Ma, X.; Dong, Z.; Chen, F.; Xiang, H.; Cao, C.; Sun, J. Airport asphalt pavement health monitoring system for mechanical model updating and distress evaluation under realistic random aircraft loads. Constr. Build. Mater.
**2019**, 226, 227–237. [Google Scholar] [CrossRef] - Zhao, H.; Wu, D.; Zeng, M.; Tian, Y.; Ling, J. Assessment of concrete pavement support conditions using distributed optical vibration sensing fiber and a neural network. Constr. Build. Mater.
**2019**, 216, 214–226. [Google Scholar] [CrossRef] - Marques, R.; Correia, A.G.; Cortez, P. Data mining applied to compaction of geomaterials. In Proceedings of the 8th International Conference on the Bearing Capacity of Roads, Railways and Airfields, Champaign, IL, USA, 29 June–2 July 2009; pp. 597–605. [Google Scholar]
- Edwards, D.; Griffiths, I. Artificial intelligence approach to calculation of hydraulic excavator cycle time and output. Min. Technol. Trans. Institutions Min. Metall.
**2000**, 109, 23–29. [Google Scholar] [CrossRef] - Tam, C.; Tong, T.K.L.; Tse, S.L. Artificial neural networks model for predicting excavator productivity. Eng. Constr. Archit. Manag.
**2002**, 9, 446–452. [Google Scholar] [CrossRef] - Parente, M.; Correia, A.G.; Cortez, P. Metaheuristics, data mining and geographic information systems for earthworks equipment allocation. Procedia Eng.
**2016**, 143, 506–513. [Google Scholar] [CrossRef] [Green Version] - Alves de Souza, V.; Giusti, R.; Batista, A. Asfault: A low-cost system to evaluate pavement conditions in real-time using smartphones and machine learning. Pervasive Mob. Comput.
**2018**, 51, 27–42. [Google Scholar] [CrossRef] - Nunes, D.; Mota, V. A participatory sensing framework to classify road surface quality. J. Internet Serv. Appl.
**2019**, 10, 1–16. [Google Scholar] [CrossRef] [Green Version] - Majidifard, H.; Adu-Gyamfi, Y.; Buttlar, W.G. Deep machine learning approach to develop a new asphalt pavement condition index. Constr. Build. Mater.
**2020**, 247, 118513. [Google Scholar] [CrossRef] - Svärd, C. Predictive Modelling of Fuel Consumption Using Machine Learning Techniques; Technical Report; Scania CV AB: Stockholm, Sweden, 2014. [Google Scholar]
- Perrotta, F.; Parry, T.; Neves, L.C. Application of machine learning for fuel consumption modelling of trucks. In Proceedings of the 2017 IEEE International Conference on Big Data (Big Data), Boston, MA, USA, 11–14 December 2017; pp. 3810–3815. [Google Scholar] [CrossRef]
- Schoen, A.; Byerly, A.; Hendrix, B.; Bagwe, R.; Santos, E.; Ben-Miled, Z. A Machine Learning Model for Average Fuel Consumption in Heavy Vehicles. IEEE Trans. Veh. Technol.
**2019**, 68, 6343–6351. [Google Scholar] [CrossRef] [Green Version] - Opara, K.R.; Brzezinski, K.; Bukowicki, M.; Kaczmarek-Majer, K. Road roughness estimation through smartphone-measured acceleration. IEEE Intell. Transp. Syst. Mag.
**2021**, in press. [Google Scholar] [CrossRef] - Sattar, S.; Li, S.; Chapman, M. Road Surface Monitoring Using Smartphone Sensors: A Review. Sensors
**2018**, 18, 3845. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Breiman, L. Random forests. Mach. Learn.
**2001**, 45, 5–32. [Google Scholar] [CrossRef] [Green Version] - McCulloch, W.S.; Pitts, W. A logical calculus of the ideas immanent in nervous activity. Bull. Math. Biophys.
**1943**, 5, 115–133. [Google Scholar] [CrossRef] - Boser, B.E.; Guyon, I.M.; Vapnik, V.N. A training algorithm for optimal margin classifiers. In Proceedings of the 5th Annual Workshop on Computational Learning Theory, Pittsburgh, PA, USA, 27–29 July 1992; pp. 144–152. [Google Scholar]
- Chapman, P.; Clinton, J.; Kerber, R.; Khabaza, T.; Reinartz, T.; Shearer, C.; Wirth, R. CRISP-DM 1.0: Step-by-Step Data Mining Guide. The CRISP-DM consortium. SPSS INC
**2020**, 9, 13. [Google Scholar] - Fayyad, U.; Piatetsky-Shapiro, G.; Smyth, P. From data mining to knowledge discovery in databases. AI Mag.
**1996**, 17, 37. [Google Scholar] - Friedman, J.; Hastie, T.; Tibshirani, R. The Elements of Statistical Learning; Springer Series in Statistics; Springer: New York, NY, USA, 2001; Volume 1. [Google Scholar]
- Bi, J.; Bennett, K.P. Regression error characteristic curves. In Proceedings of the 20th International Conference on Machine Learning (ICML-03), Washington, DC, USA, 21–24 August 2003; pp. 43–50. [Google Scholar]
- Simonyan, K.; Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv
**2014**, arXiv:1409.1556. [Google Scholar] - He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- Cortez, P. Data mining with neural networks and support vector machines using the R/rminer tool. In Proceedings of the Industrial Conference on Data Mining, Berlin, Germany, 12–14 July 2010; pp. 572–583. [Google Scholar]
- R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2020. Available online: https://www.R-project.org/ (accessed on 8 February 2021).

**Figure 2.**CRISP-DM methodology [24].

Parameter | Value |
---|---|

Accelerometer data rate | 1 Hz |

Accelerometer, full-scale | 2 g |

Gyroscope data rate | 12.5 Hz |

Gyroscope, full-scale | 250 dps |

Parameter | Value |
---|---|

Data rate | 10 Hz |

GNSS constellations | GPS, Galileo, GLONASS, and Beidou |

Inclination X (Degrees) | Inclination Y (Degrees) | Inclination Z (Degrees) | Latitude (Degrees) | Longitude (Degrees) | Altitude (m) | Speed (m/s) | Clock (yyyy-MM-ddTHH:mm:ssZ in UTC) |
---|---|---|---|---|---|---|---|

0.778198 | −0.29755 | −1.31226 | 39.4447 | −7.47812 | 447 | 0.043 | 2021-07-22T12:42:36.100Z |

0.778198 | −0.29755 | −1.31226 | 39.4447 | −7.47812 | 447 | 0.038 | 2021-07-22T12:42:36.400Z |

0.839233 | −0.30518 | −0.03052 | 39.4447 | −7.47812 | 447 | 0.038 | 2021-07-22T12:42:36.700Z |

0.839233 | −0.30518 | −0.03052 | 39.4447 | −7.47812 | 447 | 0.013 | 2021-07-22T12:42:36.800Z |

0.839233 | −0.30518 | −0.03052 | 39.4447 | −7.47812 | 447 | 0.014 | 2021-07-22T12:42:37.100Z |

0.839233 | −0.30518 | −0.03052 | 39.4447 | −7.47812 | 447 | 0.044 | 2021-07-22T12:42:37.500Z |

0.923157 | −0.28992 | −1.95313 | 39.4447 | −7.47812 | 447 | 0.004 | 2021-07-22T12:42:37.900Z |

0.923157 | −0.28992 | −1.95313 | 39.4447 | −7.47812 | 447 | 0.036 | 2021-07-22T12:42:38.200Z |

Slope Description | Range | Feature Designation |
---|---|---|

Flat surface | −1% < Slope ≤ +1% | AD_0.01n_0.01 |

Light upwards slope | +1% < Slope ≤ +5% | AD_0.01_0.05 |

Moderate upwards slope | +5% < Slope ≤ +10% | AD_0.05_0.1 |

Steep upwards slope | Slope ≥ +10% | AD_0.1 |

Light downwards slope | −5% < Slope ≤ −1% | AD_0.01_0.05n |

Moderate and steep downwards slope | Slope ≤ −5% | AD_0.05n |

AD_0.01n_0.01 (% TDistance) | AD_0.01_0.05 (% TDistance) | AD_0.05_0.1 (% TDistance) | AD_0.1 (% TDistance) | AD_0.01_0.05n (% TDistance) | AD_0.05n (% TDistance) | TDistance (m) | AvSp (m/s) | Cargo (ton) | FConsumption (L) |
---|---|---|---|---|---|---|---|---|---|

0.32 | 0.25 | 0.01 | 0 | 0.36 | 0.06 | 66,341.91 | 18.48 | 29.48 | 30 |

0.4 | 0.33 | 0.05 | 0 | 0.21 | 0.02 | 65,704.09 | 20.52 | 0 | 23 |

0.2 | 0.33 | 0.13 | 0.01 | 0.23 | 0.11 | 52,992.28 | 12.39 | 0 | 20.5 |

0.2 | 0.25 | 0.08 | 0.01 | 0.28 | 0.17 | 35,887.31 | 9.15 | 33 | 25.5 |

0.19 | 0.25 | 0.1 | 0.01 | 0.28 | 0.17 | 36,563.17 | 8.52 | 32.68 | 25 |

0.37 | 0.23 | 0.01 | 0 | 0.35 | 0.05 | 62,786.77 | 17.4 | 33.76 | 26.5 |

0.43 | 0.3 | 0.05 | 0 | 0.21 | 0.01 | 62,282.68 | 20.79 | 0 | 23.5 |

0.38 | 0.22 | 0.01 | 0 | 0.33 | 0.05 | 62,786.77 | 18.91 | 32.46 | 26 |

0.41 | 0.31 | 0.05 | 0 | 0.22 | 0.01 | 62,606.6 | 19.16 | 0 | 23 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 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

**MDPI and ACS Style**

Pereira, G.; Parente, M.; Moutinho, J.; Sampaio, M.
Fuel Consumption Prediction for Construction Trucks: A Noninvasive Approach Using Dedicated Sensors and Machine Learning. *Infrastructures* **2021**, *6*, 157.
https://doi.org/10.3390/infrastructures6110157

**AMA Style**

Pereira G, Parente M, Moutinho J, Sampaio M.
Fuel Consumption Prediction for Construction Trucks: A Noninvasive Approach Using Dedicated Sensors and Machine Learning. *Infrastructures*. 2021; 6(11):157.
https://doi.org/10.3390/infrastructures6110157

**Chicago/Turabian Style**

Pereira, Gonçalo, Manuel Parente, João Moutinho, and Manuel Sampaio.
2021. "Fuel Consumption Prediction for Construction Trucks: A Noninvasive Approach Using Dedicated Sensors and Machine Learning" *Infrastructures* 6, no. 11: 157.
https://doi.org/10.3390/infrastructures6110157