論文

For the safe and efficient use of lithium-ion batteries, the state of charge (SOC) is a particularly important state variable. In this paper, we propose a method for the online estimation of SOC and model parameters based on a fractional-order equivalent circuit model. Firstly, we constructed a fractional-order battery model that includes pseudo-capacitance and determined the values of the circuit elements offline using the least squares method from actual input–output data based on the driving profile of an automobile. Compared to the integer-order battery model, we confirmed that the proposed fractional-order battery model has higher accuracy. Secondly, we constructed a fractional-order Kreisselmeier-type adaptive observer as an observer that performs state estimation and parameter adjustment simultaneously. Applying the general adaptive law to the battery model results in a redundant design with many adjustable parameters, so we proposed an adaptive law that reduces the number of adjustable parameters without compromising the stability of the observer. The effectiveness of the proposed method was verified through numerical simulations. As a result, the high estimation accuracy and convergence of the proposed adaptive law were confirmed.

Hybrid electric vehicles are positioned as an intermediate form between gasoline and electric vehicles, contributing to lower fuel consumption and emissions. Map control is used for conventional engine control. This method maps optimal values from experimental data, and it has been pointed out that the capacity of the map is increasing and that it is difficult to respond to increasingly sophisticated control objectives. In this paper, we first present a model of a series-parallel hybrid electric vehicle and propose a method using Deep Q-Network, a typical reinforcement learning technique. Through numerical simulation, we verify that the SOC is within an acceptable range throughout the entire run and that energy efficiency can be improved compared to existing map control.

To deal with stringent exhaust emission regulations, an exhaust gas recirculation (EGR) system is widely used in internal combustion engines. However, appropriate control of EGR is difficult because of the gas transport delay and unavailability of the EGR rate. In this paper, a model-based control approach for engine airpath systems is proposed. A physical model of engine airpath considering the transport delay in EGR is constructed. An adaptive observer is introduced to estimate the EGR rate by using available quantities. Feedback error learning control is applied to the airpath system to deal with a characteristic change of the controlled plant. We validate the proposed method by numerical simulations.

Currently, due to the stricter emission legislation, spark-ignition (SI) engine with low pressure exhaust gas recirculation (LP-EGR) has become a research hotspot. However, with the increasing complexity of engine, it is difficult to obtain sufficient control performance by conventional MAP control or PID control. This paper focuses on the airpath system of a SI engine with low pressure EGR, proposes a new model-based control method for the model of the airpath system based on the Feedback Error Learning control using online sequential-extreme learning machine (OS-ELM) as controller algorithm. In this paper, the airpath model has been built and effectiveness of the proposed control method has been confirmed on the model.

This paper deals with the control problem of an engine air path system with low pressure exhaust gas recirculation (EGR). The considered air path system with low EGR can be modelled as a system with time-delays due to the transport delay of gas flow from EGR. It results in over intake fresh air, and thus it causes torque overshoot. An output predictive control system is designed for a time delay air path system in order to obtain the desired fresh air mass flow. In the considered system, the intake fresh air mass flow and the EGR rate are taken as the outputs, and the throttle valve and the EGR valve are considered as inputs of the system in this paper. A detailed air path system model with time-delay will also be derived for confirming the effectiveness of the proposed method through numerical simulations.

Despite its bright prospect to promote affordable energy, one of the main concerns of real-time retail pricing for electricity is price volatility that would create potential bill shocks especially for low-income consumers. This would demotivate consumers to participate in real-time pricing scheme or to act as responsive participants to reduce peak demand. As a motivation to address this concern, this article proposes dual optimization method with fuzzy optimal algorithm to solve the optimization problem presented in the real-time pricing model with the aims to reduce price volatility and to lower the minimum price further. The fuzzy optimal algorithm applies a context-based fuzzy inferencing and a gradient adjustment to arrive at the optimal retail price for every time interval. Context-based fuzzy inferencing allows fuzzy value redefinition so that the desired level of precision is preserved. Gradient adjustment simplifies the derivation of the optimal retail price via a series of iterations. Simulation results reveal that price fluctuations are able to be reduced which would create stability and avoid undesired price spikes. At the same time, the results confirm that further reduction in the minimum retail price can be achieved which would improve welfare benefits while maintaining the desired level of consumer satisfaction.

© 2018 Wiley Periodicals, Inc. Due to the prevalence of Li-ion battery, the importance of the battery management system (BMS) is increasing. BMS is the system for managing the operation of batteries and monitoring their state of health. By using the observer containing parameter estimator for analyzing the equivalent circuit model of batteries, the online diagnosis of the batteries becomes possible. However, the Li-ion battery has complicated dynamics derived from the Warburg impedance, which described the diffusion phenomena of the Li-ion. Applying fractional calculus to the expression of the system, the dynamics of Warburg impedance can be described simply. By designing the adaptive observer which can treat the system described as fractional-order system, it is expected to become possible to identify equivalent circuit containing Warburg element online without approximation. In this paper, we designed the Kreisselmeier-type adaptive observer for the linear fractional-order system. Then, we applied the proposed observer to the half-cell of the lithium-ion battery modeled by Randles circuit under semi-infinite linear diffusion, and confirm the effectiveness of proposed adaptive observer by numerical simulation.

© 2018 Int. J. Elec. & Elecn. Eng. & Telcomm. With the increasing penetration of renewable Distributed Generation (DG), it is important to assess the Maximum Hosting Capacity (MHC) in active distribution networks. Active Network Management (ANM) such as coordinated voltage control, reactive power compensation, DG curtailment, DG power factor control, network reconfiguration and demand response can play an important role in increasing the MHC. The MHC evaluation problem considering all the above elements of ANM can be formulated as a mixed integer nonlinear programming model. However, this original nonconvex model cannot guarantee convergence to optimality. This paper proposes the mixed integer second-order cone programming model for evaluating the MHC, by using exact linearization and second-order cone relaxation. The modified IEEE 33-bus test system is used to demonstrate the effectiveness of the proposed model and analyze the effect of each ANM element on the MHC increase. The results show that when considering all the above elements of ANM, the gain of the MHC is greater than 62%.

© 2007-2012 IEEE. In this paper, we investigate the problems of estimating and tracking the location parameters [i.e., directions-of-arrival (DOAs) and ranges] of multiple near-field (NF) narrowband sources impinging on a symmetric uniform linear array, and a simple subspace-based algorithm for localization of NF sources (SALONS) is presented, where the computationally burdensome eigendecomposition and spectrum peak searching are avoided. In the SALONS, the DOAs and ranges are estimated separately with a one-dimensional subspace-based estimation technique, where the null spaces are obtained through the linear operation of the correlation matrices formed from the antidiagonal elements of the noiseless array covariance matrix, and the estimated DOAs and ranges are automatically paired without any additional procedure. Then the statistical analysis of the presented batch SALONS is studied, and the asymptotic mean-squared-error expressions of the estimated DOAs and ranges are derived. Furthermore, an online algorithm is developed for tracking the multiple moving NF sources with crossover points on their trajectories. The effectiveness and the theoretical analysis of the presented algorithms are verified through numerical examples, and the simulation results show that the proposed algorithms provide good estimation and tracking performance for DOAs and show satisfactory estimation and tracking performance for ranges.

© 2019 SAE Japan and SAE International. In this paper, we propose an onboard calibration method based on the feedback error learning (FEL). This controller is 2 degree of freedom controller. We use a cerebellar model articulation controller (CMAC) as the feedforward controller. By using this, fast learning will be possible. Due to the online learning of this controller, it becomes possible to correspond to changes in the environment such as aging. In this paper, we constructed FEL considering the premixed diesel combustion with dual peak heat release rate under transient operation. To confirm the effectiveness of this control system, experimental results will be shown.

This paper presents a novel perturbation-based extremum seeking (ES) based on the smooth approximation of the sign of the gradient of unknown plant maps, which will improve the stability and the control performance of ES. The stability and the convergence speed of conventional sinusoidally perturbed gradient-based ES are very sensitive to the curvature of an unknown plant map, which has motivated the development of improved ES schemes that are less sensitive to the curvature such as Newton-like ES schemes. In this paper, an alternative ES scheme is proposed by estimating the sign of the gradient instead of the gradient itself. The ES scheme not only frees us from taking care of the curvature of the unknown map but also is able to minimize the increase of implementation complexity by just incorporating the signum function for the conventional ES scheme. Moreover, the signum function is replaced by its smooth approximation such as the softsign or a sigmoid function in order not only to suppress chattering phenomena but also to be able to adjust the behavior in the vicinity of an extremum point. It is also shown that the local-stability of the proposed ES scheme can be proved under some mild assumptions for the approximation function. The effectiveness of the proposed ES is demonstrated by some simulation examples for a simple process model and for a complex nonlinear biological wastewater treatment process model.

This paper is concerned with the applicability of extremum seeking (ES) control for wastewater treatment plants (WWTPs) under several practical constraints such as effluent regulations and/or operational constraints. Considering such constraints to optimize a trade-off between effluent improvement and operational cost reduction is important in real WWTP operation. To meet such practical requirements, this paper proposes a novel ES scheme incorporating multiple output constraints where the priority of each constraint is also considered. Such priority becomes important when one or more of the constraints cannot be satisfied, which will not be the exception in control problems of real WWTP due in part to the considerably large inflow load variation. To make ES be effective even in such situation, the paper proposes a prioritized output constraints ES scheme incorporating switching functions to change the performance index for minimization depending on the active constraints. It is suggested that such switching mechanism is preferably combined with the gradient-sign-based ES that was also proposed by the authors to mitigate the sudden change in the curvature of the performance index caused by the switching functions. Some numerical simulations of the proposed ES with a simple model are performed to illustrate the principle of the proposed idea at first, and then the effectiveness of the proposed ES to WWTPs is demonstrated by using a complex nonlinear biological model.

This study proposes a model predictive controller for concentrating solar power (CSP) plants. Few studies have considered a thermal energy storage system and a power system. The proposed model predictive controller maintains the heat transfer fluid temperature by manipulating its mass flow rate and stabilizes the power frequency by balancing power supply and demand. In the case of the direct normal irradiance is low, the thermal energy storage system makes up for the shortage of thermal. The simulation using actual solar direct normal irradiance data and electricity demand data shows the proposed controller contributes to stable operation of CSP plants.

In this paper, we investigate the problem of estimating the directions-of-arrival (DOAs) and ranges of multiple narrowband near-field sources in unknown spatially nonuniform noise (spatially inhomogeneous temporary white noise) environment, which is usually encountered in many practical applications of sensor array processing. A new subspace-based localization of near-field sources (SLONS) is proposed by exploiting the advantages of a symmetric uniform linear sensor array and using Toeplitzation of the array correlations. Firstly three Toeplitz correlation matrices are constructed by using the anti-diagonal elements of the array covariance matrix, where the nonuniform variances of additive noises are reduced to a uniform one, and then the location parameters (i.e., the DOAs and ranges) of near-field sources can be estimated by using the MUSIC-like method, while a new pair-matching scheme is developed to associate the estimated DOAs and ranges. Additionally, an alternating iterative scheme is considered to improve the estimation accuracy of the location parameters by utilizing the oblique projection operator, where the 'saturation behavior' caused by finite number of snapshots is overcome effectively. Furthermore, the closed-form stochastic Cramér-Rao lower bound (CRB) is also derived explicitly for the near-field sources in the additive unknown nonuniform noises. Finally, the effectiveness of the proposed method and the theoretical analysis are substantiated through numerical examples.