Search Results

Sensor selection for the control of modern powertrains is a recognised technical challenge. The key question is which set of sensors is best suited for an effective control strategy? This paper addresses the question through probabilistic modelling and Bayesian analysis. By quantifying uncertainties in the model, the propagation of sensor information throughout the model can be observed. The specific example is an abstract model of the slip behaviour of Selective Catalytic Reduction (SCR) DeNOx aftertreatment systems. Due to the ambiguity of the sensor reading, linearization-based approaches including the Extended Kalman Filter, or the Unscented Kalman Filter are not successful in resolving this problem. The stochastic literature suggests approximating these nonlinear distributions using methods such as Markov Chain Monte Carlo (MCMC), which is able in principle to resolve bimodal or multimodal results.

The idea of grid friendly charging is to use electricity from the grid to charge batteries when electricity is available in surplus and cheap. There are several ways of achieving this, for example using droop control, using night time electricity tariffs, or using smart metering. The goal is twofold: to avoid putting additional load on the electricity grid and power generation, and to reduce the cost to the consumer. This paper looks at the saving potential when charging an electric car using real time tariffs provided by a smart meter, using the Ameren tariffs in Illinois as an example. If prices are known in advance (day-ahead pricing), the optimization only requires picking the cheapest time slots for charging the battery. Further savings can be made by using real time prices that are not known in advance, but the optimization problem then depends on price prediction models, and it becomes much more difficult to solve.

The TC48 project is developing a state-of-the-art, exceptionally low cost, 48V Plug-in hybrid electric (PHEV) demonstration drivetrain suitable for electrically powered urban driving, hybrid operation, and internal combustion engine powered high speed motoring. This paper explains the motivation for the project, and presents the layout options considered and the rationale by which these were reduced. The vehicle simulation model used to evaluate the layout options is described and discussed. The modelling work was used in order to support and justify the design choices made. The design of the vehicle's control systems is discussed, presenting simulation results. The physical embodiment of the design is not reported in this paper. The paper describes analysis of small vehicles in the marketplace, including aspects of range and cost, leading to the justification for the specification of the TC48 system.

The goal of grid friendly charging is to avoid putting additional load on the electricity grid when it is heavily loaded already, and to reduce the cost of charging to the consumer. In a smart metering system, Day Ahead tariff (DA) prices are announced in advance for the next day. This information can be used for a simple optimization control, to select to charge at cheapest times. However, the balance of supply and demand is not fully known in advance and the Real-Time Prices (RTP) are therefore likely to be different at times. There is always a risk of a sudden price change, hence adding a stochastic element to the optimization in turn requiring dynamic control to achieve optimal time selection. A stochastic dynamic program (SDP) controller which takes this problem into account has been made and proven by simulation in a previous paper.

This literature review considers the problem of finding a suitable configuration of sensors and actuators for the control of an internal combustion engine. It takes a look at the methods, algorithms, processes, metrics, applications, research groups and patents relevant for this topic. Several formal metric have been proposed, but practical use remains limited. Maximal information criteria are theoretically optimal for selecting sensors, but hard to apply to a system as complex and nonlinear as an engine. Thus, we reviewed methods applied to neighboring fields including nonlinear systems and non-minimal phase systems. Furthermore, the closed loop nature of control means that information is not the only consideration, and speed, stability and robustness have to be considered. The optimal use of sensor information also requires the use of models, observers, state estimators or virtual sensors, and practical acceptance of these remains limited.

This paper presents a method to achieve a low order system model of the urea-based SCR catalyst coated filter (SCR-in-DPF or SCRF or SDPF), while preserving a high degree of fidelity. Proper orthogonal decomposition (POD), also known as principal component analysis (PCA), or Karhunen-Loéve decomposition (KLD), is a statistical method which achieves model order reduction by extracting the dominant characteristic modes of the system and devises a low-dimensional approximation on that basis. The motivation for using the POD approach is that the low-order model directly derives from the high-fidelity model (or experimental data) thereby retains the physics of the system. POD, with Galerkin projection, is applied to the 1D + 1D SCR-in-DPF model using ammonia surface coverage and wall temperature as the dominant system states to achieve model order reduction.

When a diesel engine is running at steady state, the diesel combustion process still has some level of variation from cycle to cycle, even if engine load and all control inputs are fixed. This variation is a disturbance for the speed governor, and it could lead to less than optimal engine performance in terms of fuel economy, exhaust gas emission and noise emission. The most effective way to reduce this steady state combustion variation is by applying fuel path feedback control. The control action can be performed at a fixed frequency, or at a defined cycle event time. Intra-cycle control has the highest capacity to suppress the combustion deviation, as it measures the current cycle combustion performance and compensates for it within the same cycle using a very fast control response. Correct knowledge and a model of the disturbance sources and combustion variation patterns are essential in the design process of this intra-cycle control strategy.

Models of hybrid powertrains are used to establish the best combination of conventional engine power and electric motor power for the current driving situation. The model is characteristic for having two control inputs and one output constraint: the total torque should be equal to the torque requested by the driver. To eliminate the constraint, several alternative formulations are used, considering engine power or motor power or even the ratio between them as a single control input. From this input and the constraint, both power levels can be deduced. There are different popular choices for this one control input. This paper presents a novel model based on an input linearizing transformation. It is demonstrably superior to alternative model forms, in that the core dynamics of the model (battery state of energy) are linear, and the non-linearities of the model are pushed into the inputs and outputs in a Wiener/Hammerstein form.

Hybrid technologies enable the reduction of noxious tailpipe emissions and conformance with ever-decreasing allowable homologation limits. The complexity of the hybrid powertrain technology leads to an energy management problem with multiple energy sinks and sources comprising the system resulting in a high-dimensional time dependent problem for which many solutions have been proposed. Methods that rely on accurate predictions of potential vehicle operations are demonstrably more optimal when compared to rule-based methodology [1]. In this paper, a previously proposed energy management strategy based on an offline optimization using dynamic programming is investigated. This is then coupled with an online model predictive control strategy to follow the predetermined optimal battery state of charge trajectory prescribed by the dynamic program.

Vehicle emissions and fuel economy in real-world driving conditions are currently under considerable scrutiny. Key to achieving optimum performance for a given hardware set and control scheme is a calibration that optimizes controller gains such that inputs are scheduled over the operating space to minimize emissions and maximize fuel economy. Generating a suitable calibration requires data that is both precise and accurate, this data is used to generate models that are deployed as part of the calibration optimization process. This paper evaluates the repeatability of typical steady-state measurements used for calibration of engine controllers that will ultimately determine vehicle level emissions for homologation include Real Driving Emissions (RDE). Stabilization requirements as indicated by three different measurements are evaluated and shown to be different within the same experiment, depending on the metric used.

The energy consumption of a vehicle is typically determined either by testing or in simulation. While both approaches are valid, they only work for a specific drive cycle, they are time intensive, and they do not directly result in a closed-form relationship between key parameters and consumption. This paper presents an alternative approach that determines the consumption based on a simple analytical model of the vehicle and statistical parameters of the drive cycle, specifically the moments of the velocity. This results in a closed-form solution that can be used for analysis or synthesis. The drive cycle is quantified via its moments, specifically the average speed, the standard deviation of the speed as well as the higher order moments skewness, and the kurtosis. A mixed quadratic term is added to account for acceleration or aggressiveness, but it is noticeably distinct from the conventional metric of positive kinetic energy (PKE).

This paper presents a novel approach to augment existing engine calibrations to deliver improved engine performance during a transient, through the application of multi-objective optimization techniques to the calibration of the Variable Valve Timing (VVT) system of a 1.0 litre gasoline engine. Current mature calibration approaches for the VVT system are predominantly based on steady state techniques which fail to consider the engine dynamic behaviour in real world driving, which is heavily transient. In this study the total integrated fuel consumption and engine-out NOx emissions over a 2-minute segment of the transient Worldwide Light-duty Test Cycle are minimised in a constrained multi-objective optimisation framework to achieve an updated calibration for the VVT control. The cycle segment was identified as an area with high NOx emissions.

Hybrid electric vehicles offer significant fuel economy benefits, because battery and fuel can be used as complementing energy sources. This paper presents the use of dynamic programming to find the optimal blend of power sources, leading to the lowest fuel consumption and the lowest level of harmful emissions. It is found that the optimal engine behavior differs substantially to an on-line adaptive control system previously designed for the Lotus Evora 414E. When analyzing the trade-off between emission and fuel consumption, CO and HC emissions show a traditional Pareto curve, whereas NOx emissions show a near linear relationship with a high penalty. These global optimization results are not directly applicable for online control, but they can guide the design of a more efficient hybrid control system.

This paper presents the implementation of a vehicle and powertrain model of the parallel hybrid electric vehicle which can be used for several purposes: as a model for estimating fuel consumption, as a model for estimating performance, and as a control model for the hybrid powertrain optimisation. The model is specified as a multi-domain physical model in MATLAB Simscape, which captures the key electrical, mechanical and thermal energy flows in the vehicles. By applying hand crafted boundary conditions, this model can be simulated either in the forwards or backwards direction, and it can easily be simplified as required to address specific control problems. Modelling in the forwards direction, the driver inputs are specified, and the vehicle response is the model output. In the backwards direction, the vehicle velocity as a function of time is the specified input, and the engine torque, and fuel consumption are the model outputs.

In order to identify predictive models for a diesel engine combustion process, combustion cylinder pressure together with other fuel path variables such as rail pressure, injector current and sleeve pressure of 1000 continuous cycles were sampled and collected at high resolution. Using these engine steady state test data, three types of modeling approach have been studied. The first is the Auto-Regressive-Moving-Average (ARMA) model which had limited prediction ability for both peak combustion pressure (Pmax) and Indicated Mean Effective Pressure (IMEP). By applying correlation analysis, proper inputs were found for a linear predictive model of Pmax and IMEP respectively. The prediction performance of this linear model is excellent with a 30% fit number for both Pmax and IMEP. Further nonlinear modeling work shows that even a nonlinear Neural Network (NN) model does not have improved prediction performance compared to the linear predictive model.