Christoph Romaus

Optimal energy management for a hybrid energy storage system combining batteries and double layer capacitors

Many mobile vehicular applications like hybridelectrical cars and autonomous rail vehicles require an onboard energy storage for operation. High demands concerning power and energy density, small volume and weight at the same time cannot be satisfied solely by batteries or double layer capacitors. A suitable approach is to combine storage technologies with complementary characteristics as a hybrid energy storage system. Thus, long term storage like batteries featuring high energy density can be combined with short term storage like double layer capacitors offering high power density and high cycliability. To control the power flows of the system, we propose the use of self-optimization methods involving multi-objective and discrete optimization to design an operating strategy which is able to adapt the system behavior to different conditions not only by adapting its parameters but also its objectives, offering an optimal operation in different situations.

Optimal energy management for a hybrid energy storage system for electric vehicles based on Stochastic Dynamic Programming

For electric and hybrid electric cars, commonly nickel-metal hydride and lithium-ion batteries are used as energy storage. The size of the battery depends not only on the driving range, but also on the power demands for accelerating and braking and life-time considerations. This becomes even more apparent with short driving ranges, e.g. in commuter traffic. By hybridization of the storage, adding double layer capacitors, the battery can be relieved from the stress of peak power and even downsized to the energy demands instead of power demands. The dimensioning of the storage is performed by a parametric study via Deterministic Dynamic Programming. To determine an energy management to control the power flows to the storage online during operation which considers the stochastic influences of traffic and the driver, Stochastic Dynamic Programming is investigated and compared to the optimal strategy found during the dimensioning.

Probabilistic planning integrated in a multi-level dependability concept for mechatronic systems

Self-optimizing mechatronic systems are a new class of technical systems. On the one hand, new challenges regarding dependability arise from their additional complexity and adaptivity. On the other hand, their abilities enable new concepts and methods to improve the dependability of mechatronic systems. This paper introduces a multi-level dependability concept for self-optimizing mechatronic systems and shows how planning can be used to improve the availability and reliability of systems in the operating stages.

Probabilistic Planning for Predictive Condition Monitoring and Adaptation Within the Self-Optimizing Energy Management of an Autonomous Railway Vehicle

Self-optimizing mechatronic systems are a new class of technical systems. On the one hand, new challenges regarding dependability arise from their additional complexity and adaptivity. On the other hand, their abilities enable new concepts and methods to improve the dependability of mechatronic systems. This paper introduces a multi-level dependability concept for selfoptimizing mechatronic systems and shows how probabilistic planning can be used to improve the availability and reliability of systems in the operating phase. The general idea to improve the availability of autonomous systems by applying probabilistic planning methods to avoid energy shortages is exemplified on the example of an innovative railway vehicle.

Discrete-time model of an IPMSM based on variational integrators

Interior permanent-magnet synchronous motors (IPMSM) are widely used as automobile traction drives for electric and hybrid electric vehicles. Since motor control is commonly implemented on a digital platform, a discrete-time motor model is needed. With respect to the calculation effort the discretization is usually done via a first-order explicit Euler approximation within the rotor frame coordinates. For IPMSM traction drives this approach is leading to a systematic modeling error, since the flux trajectory degenerates from a circle to a polygon at higher speeds. Consequently, this approach leads to a speed-depending discretization error, which can have a significant impact on control performance. Hence, this contribution presents a discrete-time model of an IPMSM based on variational integrators (VI). Starting with a variational principle the Euler-Lagrange equations provide a physical continuous-time motor model. By applying a discrete version of the variational principle we receive a symplectic discrete-time IPMSM model without increasing the calculation effort in comparison to the classical Euler discretization. Simulative as well as experimental investigations confirm the benefit of a VI based discrete-time motor model concerning the transient and stationary system model accuracy.

Self-optimization energy management considering stochastic influences for a hybrid energy storage of an electric road vehicle

Electric and hybrid-electric vehicles place high demands for peak power, energy content and efficiency on the energy storage. By hybridization of the storage, adding double layer capacitors, the battery can be relieved from the stress of peak power and even downsized to meet only energy demands instead of power demands. Thus, the storage weight and losses can be significantly reduced. An energy management to distribute the power to both storages can be mathematically optimized applying Stochastic Dynamic Programming (SDP), considering stochastic influences of the driving process. To handle different conditions and driving cycles, we propose self-optimization control strategies involving multi-objective optimization. These strategies are able to autonomously adapt their behavior and relevance of objectives, offering an optimal and secure operation in different situations.

DEFINING PLAN METRICS FOR MULTI-AGENT PLANNING WITHIN MECHATRONIC SYSTEMS

The paradigm of self-optimization introduces flexible systems of objectives to the area of mechatronics. From this flexiblity new coordination problems arise. Such a problem is the coordination of function modules in such a way that the overall system of objectives is met as good as possible. Multi-agent planning offers mechanisms to provide this coordination. In this paper we introduce the application of an established method from the area of decision analysis (value tree anaylsis) to define a common and coherent system of plan metrics for the multi-agent based coordination of function modules. The application of this approach is demonstrated on example of function modules within a railway vehicle.Copyright

Online capable optimized planning of power split in a hybrid energy storage system

A hybrid energy storage system is an energy storage consisting of more than one type of energy storages combining their advantages. The system treated in this paper consists of NiMH batteries and double layer capacitors. The power flow needs to be split up between the storage devices. This defines a multi-objective optimization with the minimization of energy loss and maximization of the abilty to react on unforeseen power demands as objectives. So far this had to be optimized by offline methods on typical situations. This paper presents an optimization method that is capable of planning the power flow split at run time. For this purpose a Pareto Point search algorithm is used which geometrically constructs a function to minimize in the image space of objective functions. This is done by using single-objective optimization to find the endpoints of the Pareto Front. The function to be minimized then defines a trail on which the optimal solution for a given trade-off parameter is searched.

Multiobjective search for the management of a hybrid energy storage system

Self-optimizing mechatronic systems are a new class of technical systems which are able to autonomously reason about their objectives and to select appropriate behavior adaptations with regard to their current environment and changing requirements. Obviously, reasoning and changing the systems behavior during operation is a time critical task. Thus, it is usually not possible to explore all possible behaviors and evaluate them regarding various objectives. To reduce this effort, the set of nondominated solutions, the Pareto set, can be computed in advance and thus the effort of adapting the system behavior is reduced to a decision making process. This paper introduces a search based approach to approximate the Pareto set with a multi objective search algorithm. The algorithm is designed for the management of a hybrid energy storage module in a rail-bound vehicle where the two contradicting objectives of maximizing the power reserve to maintain robust against unexpected power peaks and minimizing the deterioration of the storages have to be optimized simultaneously.

Examples of Self-optimizing Systems

In this chapter, the benefits resulting from self-optimization will be described based on application examples from the Collaborative Research Center 614 “Self-optimizing Concepts and Structures in Mechanical Engineering”. First, the autonomous rail vehicle RailCab developed at the University of Paderborn is introduced. Then, the RailCab subsystems Self-Optimizing Operating Point Control, Intelligent Drive Module, Active Suspension Module, Active Guidance Module and Hybrid Energy Storage System and their test rigs are described in detail as well as an overall approach for Energy Management. The chapter concludes with the presentation of other development platforms such as the BeBot, an intelligent miniature robot acting optimally in groups, and the X-by-wire vehicle Chameleon with independent single-wheel chassis actuators. All the above mentioned demonstrators are used to validate the methods and procedures developed in the Collaborative Research Center. The experiences gained, provide direct input into further development and optimization of the design as well as the self-optimization process.