Precise estimation of state of health (SOH) are of great importance for proper operation of lithium-ion batteries equipped in electric vehicles. For real applications, it is however difficult to estimate battery SOH due to stochastic operation, which in turn speeds up aging process of the battery. To attain the precise SOH estimation, an efficient estimation manner based on machine learning is proposed in this study. Firstly, the voltage profile during charging and discharging process and incremental capacity variation are acquired through the cycle life test, and the healthy features correlating to battery degradation are extracted. Secondly, the grey relation analysis and entropy weight method are employed to analyze the healthy features. Finally, the long short-term memory is established to achieve the SOH estimation of battery. The experimental results highlight that the proposed method can effectively predict the battery SOH with preferable accuracy, stability and robustness. INDEX TERMS Healthy features (HFs), grey relational analysis (GRA), entropy weight method (EWM), long short-term memory (LSTM), state of health (SOH). NOMENCLATURE A. ABBREVIATIONS

This paper proposes an energy management strategy for a power-split plug-in hybrid electric vehicle (PHEV) based on reinforcement learning (RL). Firstly, a control-oriented power-split PHEV model is built, and then the RL method is employed based on the Markov Decision Process (MDP) to find the optimal solution according to the built model. During the strategy search, several different standard driving schedules are chosen, and the transfer probability of the power demand is derived based on the Markov chain. Accordingly, the optimal control strategy is found by the Q-learning (QL) algorithm, which can decide suitable energy allocation between the gasoline engine and the battery pack. Simulation results indicate that the RL-based control strategy could not only lessen fuel consumption under different driving cycles, but also limit the maximum discharge power of battery, compared with the charging depletion/charging sustaining (CD/CS) method and the equivalent consumption minimization strategy (ECMS).

In this paper, a stochastic model predictive control (MPC) method based on reinforcement learning is proposed for energy management of plug-in hybrid electric vehicles (PHEVs). Firstly, the power transfer of each component in a power-split PHEV is described in detail. Then an effective and convergent reinforcement learning controller is trained by the Q-learning algorithm according to the driving power distribution under multiple driving cycles. By constructing a multi-step Markov velocity prediction model, the reinforcement learning controller is embedded into the stochastic MPC controller to determine the optimal battery power in predicted time domain. Numerical simulation results verify that the proposed method achieves superior fuel economy that is close to that by stochastic dynamic programming method. In addition, the effective state of charge tracking in terms of different reference trajectories highlight that the proposed method is effective for online application requiring a fast calculation speed.

In this article, a multi-objective optimization-oriented energy management strategy is investigated for fuel cell hybrid vehicles on the basis of rule learning. The degradation of fuel cells and lithium-ion batteries are considered as the objective function and translated into the equivalent hydrogen consumption. The optimal fuel cell power sequence and state of charge trajectory, considered as the energy management input, are solved offline via the Pontryagin's minimum principle. The K-means algorithm is employed to hierarchically cluster the optimal data set for preparation of rules extraction, and then the rules are excavated by the improved repeated incremental pruning to production error reduction algorithm and fitted by the quasi-Newton method. The simulation results highlight that the proposed rule learning-based energy management strategy can effectively save hydrogen consumption and prolong fuel cell life with real-time application potential.

Online optimal energy management of plug-in hybrid electric vehicles has been continually investigated for better fuel economy. This paper proposed a predictive energy management strategy based on multi neural networks for a multi-mode plug-in hybrid electric vehicle. To attain it, firstly, the offline optimal results prepared for knowledge learning are derived by dynamic programming and Pontryagin's minimum principle. Then, the mode recognition neural network is trained based on the optimal results of dynamic programming and the recurrent neural network is firstly exploited to realize online co-state estimation application. Consequently, the velocity prediction-based online model predictive control framework is established with the co-state correction and slacked constraints to solve the real-time optimal control sequence.A series of numerical simulation results validate that the optimal performance yielded from global optimal strategy can be exploited online to attain the satisfied cost reduction, compared with equivalent consumption minimum strategy, with the assistance of estimated real time co-state and slacked reference. In addition, the computation duration of proposed algorithm decreases by 23.40%, compared with conventional Pontryagin's minimum principle-based model predictive control scheme, thereby proving its online application potential.

No abstract

Capacity prediction of lithium-ion batteries represents an important function of battery management systems. Conventional machine learning-based methods for capacity prediction are inefficient to learn long-term dependencies during capacity degradations. This paper investigates the deep learning method for lithium-ion battery's capacity prediction based on long short-term memory recurrent neural network, which is employed to capture the latent long-term dependence of degraded capacity. The neural network is adaptively optimized by the Adam optimization algorithm, and the dropout technique is exploited to prevent overfitting. Based on the offline cycling aging data of batteries, the capacity prediction performance is validated and evaluated. The experimental results demonstrate that the proposed algorithm can accurately track the nonlinear degradation trend of capacity within the whole lifespan with a maximum error of only 2.84%. INDEX TERMS Lithium-ion batteries, capacity prediction, aging factors, long short-term memory (LSTM).

Intersections have attracted wide attention owing to their complexity and high rate of traffic accidents. In the process of developing L3-and-above autonomous-driving techniques, it is necessary to solve problems in autonomous driving decisions and control at intersections. In this article, a decision-and-control method based on reinforcement learning and speed prediction is proposed to manage the conjunction of straight and turning vehicles at two-way single-lane unsignalized intersections. The key position of collision avoidance in the process of confluence is determined by establishing a road-geometry model, and on this basis, the expected speed of the straight vehicle that ensures passing safety is calculated. Then, a reinforcement-learning algorithm is employed to solve the decision-control problem of the straight vehicle, and the expected speed is optimized to direct the agent to learn and converge to the planned decision. Simulations were conducted to verify the performance of the proposed method, and the results show that the proposed method can generate proper decisions for the straight vehicle to pass the intersection while guaranteeing preferable safety and traffic efficiency.