Online Optimization of Complex Transportation Systems

Abstract

This paper discusses online optimization of real-world transportation systems. We concentrate on transportation problems arising in production and manufacturing processes, in particular in company internal logistics. We describe basic techniques to design online optimization algorithms for such systems, but our main focus is decision support for the planner: which online algorithm is the most appropriate one in a particular setting? We show by means of several examples that traditional methods for the evaluation of online algorithms often do not suffice to judge the strengths and weaknesses of online algorithms. We present modifications of well-known evaluation techniques and some new methods, and we argue that the selection of an online algorithm to be employed in practice should be based on a sound combination of several theoretical and practical evaluation criteria, including simulation. INTRODUCTIONThe strategic planning of complex transportation systems such as public transportation networks, automatically guided vehicles in warehouses, etc. has received a considerable amount of attention in the last decade. Strategic planning is the stage of system design where an object (e.g., a telecommunication network) is designed that will remain static for a certain planning period (the network topology, and edge capacities will not change) such that a few control parameters (e.g., routing and switching) will allow an (almost) optimal handling of all input data (within a certain realistic or predicted range). The system itself is usually not yet operational when this "strategic optimization" takes place. Here, methods of offline optimization apply. The increasing computing power and significant advances in traditional optimization techniques have resulted in substantial savings of resources in this area. Despite many successes of this approach, e.g., for transportation systems, it has turned out that achieving savings also requires an optimized operational control. Such a control involves actions to be executed while a system is running; i.e., input data arise over time, have to be processed, and (irrevocable) decisions have to be made before all input data are known. This means that methods of online optimization have to be employed. In many cases decision making has to satisfy certain real-time requirements: every decision has to be made within strict time limits.In this paper, we survey some new methods (beyond standard competitive analysis) to obtain decision support for the choice of online algorithms in real-world transportation systems. In each case, we are looking for a "good" online control on the basis of online algorithms. The methods discussed are competitive analysis against restricted adversaries (a variant of competitive analysis where the offline adversary is given less power), analysis under A-reasonable load (we compare the

Model Predictive Control of HVAC Systems: Implementation and Testing at the University of California, Merced

“…Modulation of operational set-points and modification of sequences of operation (equipment start/stop logic for example) is fundamentally a dynamic optimization problem whose solution depends on the future evolution of a nonlinear and large scale system. The key enablers are the use of computationally efficient, equation-based models that can be used in conjunction with nonlinear programming algorithms [9,10], thereby allowing the application of optimal control theory for large scale systems [11], and enterprise-scale web-enabled networked control systems. Beyond this demonstration, model-based methods and computational tools must be developed for design and analysis of robust optimal control algorithms, reducing the impact of uncertainty in models and disturbances and enabling the use of optimal control on a routine basis.…”

“…The following chiller calibration follows the procedure outlined in [11]. The temperature-dependent cooling capacity function T CapF and the temperature-dependent energy input ratio function EIRFT must be calibrated using data from the chiller at its maximum cooling capacity.…”

On Minimizing the Maximum Flow Time in the Online Dial-a-Ride Problem

“…The main difference to the previous section is that a server can serve an unlimited number of requests simultaneously if these requests specify all the same node to be visited. It is easy to see that even on the uniform metric space with at least two points the standard unrestricted adversary can construct sequences where it achieves a zero maximum flow time whereas any deterministic online algorithm has a positive flow time for some request, see also [7,8,11]. Consequently, there can not be any strictly competitive algorithm.…”

“…It is well known that for the F max -OlDarp in general metric spaces no strictly competitive algorithms can exist, see e.g. [7,8,11]. For the special case of the F max -OlTsp, where source and destination for each ride coincide (u j = v j for all j), a restriction on the adversary allows for a strictly competitive algorithm on the real line [11].…”

Combinatorial Online Optimization in Real Time

“…An example of a discrete event based simulation system is the library AM SEL [3]. It was used in the investigations in [29].…”