سیستم های پشتیبانی تصمیم گیری و هماهنگی شرکای کنسرسیوم تامین

We propose to enhance decision support systems (DSS) by coordination capabilities to align the decision making of consortium partners for improving the process quality in volatile process environments. In computer simulation experiments, we reveal conceptional shortcomings of traditional DSS if unforeseen process-threatening events like unexpected workloads have to be handled by a consortium. It is demonstrated for a transportation process planning scenario that a process quality increase is achieved if a DSS is extended by components that align the process-related decision making of legally independent supply consortium partners. The central idea proposed to add coordination capabilities to DSS is to temporarily adjust decision models of subordinate consortium partners to the fulfillment degree of consortium objectives in order to establish a coordinated decision making.

A decision support system (DSS) is an information processing system especially dedicated to derive or support the derivation of goal-oriented decisions in complex decision situations. Such a system combines process-related data with analytical decision models in order to enable a computer-based control of value creation processes [1]. DSS are set up according to some commonly agreed design paradigms. Within this contribution we report a research about the evaluation of existing DSS-concepts for the management of transport processes in a volatile process environment. In such an environment, the planning conditions and requirements vary significantly within a small time span. Processes running in a volatile environment need a frequent update in order to maintain their efficiency. The management of processes in a volatile surrounding becomes even more challenging if two or more decision makers are involved (often on different decision levels). Here, the needs and objectives of all participating acteurs have to be considered simultaneously during the process planning. A supply chain consortium is an often found example for such a complicated decision making scenario. The superior coordinator instructs a subordinate service partner to fulfill certain orders. In reaction to this call, the service partner deploys its resources to fulfill the orders by executing processes. In this report, we deal with such a two-decision-maker-situation arising from a supply chain scenario, in which the subordinate service partner offers transportation. By means of the two-decision-maker situation from supply chain planning, we reveal shortcomings of DSS design paradigms caused by extraordinary events whose processing is not defined in advance. Although previous research [2] and [3] has addressed new ideas for handling those events DSS architectures are unable to handle these events efficiently. The main contribution of this article is the proposal of an extension of the three-layer event-handling-concept [4] by a fourth layer that controls even the handling of extraordinary events. We define and evaluate a four layer event handling system for the management of extraordinary situations in the above-mentioned two-decision-maker-situation. We state the following research hypotheses guiding the research reported here: (i) two (or more) concurrent process decisions makers cannot be sufficiently supported by a DSS set up according to existing design paradigms. (ii) A consideration of several decision makers in the layout of a DSS contributes to overcome some deficiencies arising from conflicting or even contradicting planning goals of interacting process decision makers. The organization of this paper is as follows. We start with the description of the investigated process decision situation (Section 2). Then, we compile DSS-design principles (Section 3), propose a multi-agent-system-based DSS for the aforementioned decision situation (Section 4), and demonstrate its shortcomings. A conceptual extension of design guidelines for DSS is proposed in Section 5 and applied to extend the DSS from Section 4. Performing computational simulation experiments, we evaluate the extended system (Section 6).

We have presented recent research about DSS-improvements for processes running in a volatile environment that are managed by two decision making units (forming a supply consortium): a superior coordinator and a subordinate fleet managing agent. The research hypotheses stated in the report’s introduction have been verified. Within computational simulation experiments we have first approached the limits of model-based DSS applied in a volatile process environment: a collapse of the process punctuality caused by a short but severe workload peak is provoked. These simulation results have revealed that the interests of the superior supply chain coordinator cannot be sufficiently integrated into the DSS. The reasons for this bad performance of the DSS have been analyzed and it has turned out that coordination between the acting agents is not possible using a state-of-the-art DSS. It has been revealed that significant changes of the planning situation (the temporal scarceness of resources) have not been reflected into the formalized problem representation (process model) maintained by the DSS because the superior coordinator that knows about the varied planning situation has not been part of the DSS. To remedy this shortcoming of a DSS we have proposed to improve the MBMS-component of the DSS by adding a component that emulates the behavior of the superior consortium coordinator. This component adjusts the maintained formal process model to a varied process environment so that unexpected planning situation variations can now be reflected into the process model. We have defined and evaluated the proposed extension for the investigated dynamic vehicle routing problem. Strategies to adjust the objective function and/or the constraint set of the maintained decision model have been set up and tested. We have been able to show that the enhancement of the DSS by coordination issues leads to a significant increase of the process quality even after the appearance of a workload peak. Thereby, we have verified our second research hypothesis. Future research regarding enhanced DSS will follow two directions. On the one hand, we will analyze the applicability of multi-dimensional performance metrics for the transport scenario. Here, the major challenge is the consolidation of concurring or even contracting feedback signals into an applicable intervention directive. On the other hand, we will transfer the idea of integrating coordination-issues into DSS to other application areas in order to evaluate the general applicability of the proposed concept.