Mahmoud Abou-Nasr

Modeling of vehicle driving conditions using transition probability models

This paper considers modeling of vehicle driving conditions using transition probability models (TPMs) for applications of dynamic optimization. The properties of transition probabilities for vehicle speed, vehicle acceleration, and road grade are discussed based on the analysis and experimental vehicle data. The KL-divergence is shown to provide an effective metric that can differentiate similar driving conditions from dissimilar ones.

Recurrent neural network training for energy management of a mild Hybrid Electric Vehicle with an ultra-capacitor

The paper exemplifies the design flow of a neural network controller for energy management of a parallel Hybrid Electric Vehicle (HEV) with an ultra-capacitor. As the energy storage capacity of the ultra-capacitor is limited, the energy management in such a powertrain can be particularly challenging, and charging/discharging of the ultra-capacitor has to be performed optimally to reduce fuel consumption and avoid drivability degradation. In this paper, a neural network model of the powertrain is first trained from the input/output data generated by a powertrain model; with this neural network model a neural network controller, which prescribes the power split between the engine and the electric motor, is trained using a multi-stream Extended Kalman Filter (EKF)-based training method. The weights of the neural network controller are then further refined on the original plant model by a stochastic perturbation approach. The choices of the neural network architecture and of the cost function to model the plant and the controller are rationalized. The fuel consumption improvement is compared against the optimal solution obtained using dynamic programming.

Empirical modeling of vehicle fuel economy based on historical data

This paper addresses modeling and predicting vehicle fuel economy based on simple vehicle characteristics. The models are identified using a historical vehicle fuel economy data set. First, the use of least squares regression analysis is pursued, and a mathematical model is created that is capable of predicting highway fuel economy based on six vehicle characteristics: engine displacement volume, vehicle maximum power, vehicle maximum torque, vehicle weight, vehicle wheelbase, and vehicle cross sectional area. Then neural network models are developed and shown to achieve higher accuracy as compared to the regression models, with 70 percent of the data in the validation data set predicted within 2 mpg. Furthermore, we demonstrate that by employing a hybrid architecture, where vehicles are first clustered and then separate models are developed for vehicle clusters, the model accuracy can be improved further.

Ensembles of neural networks with generalization capabilities for vehicle fault diagnostics

This paper presents a two-step ensemble approach for vehicle fault diagnostics, an ensemble selection algorithm, BFES, and an analog Bayesian ensemble decision function, A-Bayesian-Entropy. We show through experiments that a neural network ensemble designed and trained by the proposed methodology, and selected by BFES with A-Bayesian-Entropy as the ensemble decision function can generalize well to vehicle models that are different from the vehicles used to generate training data.

Exploiting Terrain Information for Enhancing Fuel Economy of Cruising Vehicles by Supervised Training of Recurrent Neural Optimizers

In this chapter, we present a novel data driven approach based on supervised training of feed forward neural networks for solving nonlinear optimization problems. Then we extend the approach to approximate the solution of deterministic, discrete dynamic programming problems by using recurrent networks. We apply this data driven methodology on a real-world fuel economy application in which we train a neural optimizer to prescribe the optimum cruise speed that minimizes fuel consumption, based on the instantaneous and a limited history of the vehicle speeds and road grades, with no a priori knowledge of the future path. The optimizer is tested in simulation on novel road segments. In simulation tests, the optimizer prescribed grade based modulated speed, has achieved about 8–10.6 % fuel savings over driving with constant cruise speed on the same roads, out of which 3.7–10.6 % were due to exploiting the road grades.

Terrain identification in grayscale images with recurrent neural networks

This paper presents an approach for terrain identification in grayscale images based on recurrent neural networks. The network in this work has 16 inputs that represent 16, horizontally contiguous pixels from the grayscale image. The network is trained as a binary classifier that classifies the input pixels while being scanned from the top to the bottom of the image. Experiments were performed on grayscale images of a road in natural surroundings of grass, some trees and falling tree leaves. The trained network classifier in generalization testing experiments has managed to classify pixels representing the road as they are being scanned with accuracy of ∼ 89 % and pixels representing falling tree leaves with accuracy of ∼ 88 %.

Dynamic adaptation of a vehicle's cruising speed with recurrent neural networks for enhancing fuel economy

This paper presents an architecture for optimally modulating the cruise speed around its set point. Our objective is to minimize the overall fuel consumption over a trip without impacting the overall trip time, by exploiting the vehicle dynamics and the terrain, specifically in this paper, the road grades. The overall trip time is defined as the time to complete the trip while driving at the constant cruise speed, which was set by the driver at the beginning of the trip. We test this architecture with data acquired by an instrumented vehicle driven on city and highway roads in Southeast Michigan. Our testing was very promising and showed that we can achieve up to 11% of overall fuel economy of which 10.8% are from exploiting the road grades.