Control Map Generation Strategy for Hybrid Electric Vehicles Based on Machine Learning With Energy Optimization

Abstract

In this study, a control map generation strategy for hybrid electric vehicles based on machine learning (ML) with optimization data was studied using a multimode hybrid electric vehicle. The optimization data from dynamic programming were used to produce the control maps by employing different ML methods, including Gaussian naïve Bayes, linear discriminant analysis, decision tree, k-nearest neighbors, and support vector machine. Since control map domains separated into several domains can exhibit unrealistic control behavior during engine on-off and hybrid mode shift processes, control maps separated into the same number of domains were used for the simulation study among the different ML methods. The demand torque and power maps by ML training were used for simulations of representative driving test cycles from the Environmental Protection Agency. The results with ML methods indicate that operating domains in the torque and power maps were separated for different driving modes, while they were not clearly separated in the results of DP optimization. Further, the results with the control maps for demand power exhibited slightly improved fuel efficiency compared to the maps for demand torque. This study is meaningful because a control map generation strategy based on ML was not only studied in order to observe the possibility of utilizing DP optimization results for real vehicles, but different types of ML methods were also analyzed and discussed to find appropriate methods for vehicle control map generation in terms of demand torque and power operating points.

Efficient Management of Energy Consumption of Electric Vehicles Using Machine Learning—A Systematic and Comprehensive Survey

“…Offline [23,32,35,38,45,47,49,51,55,57,59,73,[79][80][81][82][83]86,88,91,94,103,107,109,110,116,122,128,131,133,134,136,141,149,156,157 The building models perform the training offline, which means that the prediction model is constructed once in the training phase and cannot be updated with new data. Therefore, over time, when the model gets out-of-date and does not work perfectly in the way that it should, it becomes necessary to re-train the model with more or newer data and then update the system which includes the new model.…”

“…Highway [19][20][21][22]24,25,27,28,30,[32][33][34][35]38,44,45,49,[53][54][55]57,60,64,66,69,70,73,84,86,[88][89][90][91][93][94][95][96][97][98][99]101,102,105,107,108,112,113,[115][116][117][118][119]121,…”

Determination of Energy Savings via Fuel Consumption Estimation with Machine Learning Methods and Rule-Based Control Methods Developed for Experimental Data of Hybrid Electric Vehicles

“…Hua et al proposed an estimation model for electric vehicle energy consumption based on both vehicle parameters, such as speed and environmental data (e.g., GPS position and temperature), with a machine learning algorithm [34]. In addition, ML has been widely used to control the fuel economy of hybrid electric vehicles [35][36][37][38]. Harold et al developed a framework that would allow supervised machine learning to automatically retrain the supervisory powertrain control approach for hybrid electric vehicles [35].…”

“…The Gaussian process regression method provided the best fit for the road input data used to create the CO 2 emission model. Kwon investigated a control map generation technique based on ML and optimization data for a multimode hybrid electric car [38].…”