Ocena adekwatności modeli symulacyjnych

Abstract

The main problem encountered while conducting simulation experiments is referring an accepted model and results obtained by means of it to a simulated fragment of reality. All the problems appearing here are called ari evaluation of usefulness or an evaluation of adequacy of simulation models. This stage decides to what extent conclusions concerning a model can be transferred onto an analyzed system, i.e. to what extent a model is adequate in relation to a modelled reality. It is an important and at the same time a difficult problem connected with computer simulation. At the stage of evaluating a model’s usefulness it is necessary to construct a tool allowing to compare a model with an object it reflects. The author has presented a procedure of analyzing the adequacy of simulation models with application of statistical tests and spectrum analysis. This procedure is universal, as it makes allowances for the main problems, which can be encountered while evaluating usefulness of a model (i.e. whether there is or not an autocorrelation between subsequent results of an experiment and a different size of analyzed samples). Consequently, it can be applied to all classes of simulation models. The algorithm of the model's validation was implemented on the microcomputer of IBM PC class in an integrated environment of Turbo Pascal 5.0 language. The article consists of two parts. Its first part ("Evaluation o f adequacy of simulation models by means of statistical tests and spectrum analysis") describes the procedure of analyzing the usefulness of a model proposed by the author by means of statistical tests and spectrum analysis . The second part ("description of programme algorithms implementing an evaluation of adequacy of simulation models") presents programme algorithms used for statistical tests and spectrum analysis.