There is no doubt that logistics is an area, that has already been important in the past and will become even more important in the future. The standard (model-) problems in transportation logistics, such as routing, assignment, dispatching etc. are usually combinatorial optimization problems. Therefore, they are considered to be hard to solve (some have been shown to be NP-hard problems). Very often, additional constraints have to be considered in practice, and many of the constraints, cost coefficients and relations are in reality not crisp but fuzzy. Also, decision support systems in logistics are expected to operate online.
Fuzzy sets have been and are being used in different ways to help to solve these problems. For a decade or more, problems of traffic management have been tackled by using fuzzy approaches, and also the use of fuzzy sets in the area of fleet scheduling is well known. Eventually, one of the recent successive developments seems to be the use of approximate reasoning in container transport and storage. In the presentation, the first two areas will be surveyed and visualized by presenting one exemplary application each.
The last area will be discussed in more detail; the problem described is the simultaneous control of 44 cranes, belonging to a facility under construction that will have storage space for 30 000 TEU's (20 foot container). The facility is structured in 22 blocks over which there move the 44 double rail mounted gentry cranes. There will be (on the waterside) 60 AGV's (automatic guided vehicles), 4 quays for very large container ships (altogether 1400 m long) and 14 bridges that unload the ships. On the landside, there will be 6 parallel rail connections and gates for as many trucks arriving simultaneously. In the final stage, it is planned to have 1.9 million containers move through this terminal per year.
The cranes move the containers from and to either the land or the waterside into or from the storage area, and they also do the relocation of containers in the storage area. Sometimes the information available is incomplete. Stochastic modeling, however, would not be appropriate because it would take too much time and the decisions needed have to refer to single containers and not be statistical measures. Therefore, the decision making has been cast into an approximate reasoning knowledge based system, in which the knowledge is stored in the form of (fuzzy) if-then statements and the inference engine is a multi-stage meaning preserving engine, in which degrees of compensation between the different conditions are modeled by using various fuzzy operators.