A STATISTICAL METHOD FOR DETERMINING ERRORS IN COMPUTING EQUIPMENT USING ANALOG COMPUTERS

A method is described by which computing equipment error distribution may be obtained by using the Monte-Carlo method applied to an analog computer. In addition to the computer proper, the following units are needed to perform this task automatically: a low frequency random noise generator or a random sequence generator, multiplication and storage units for random sequences, and units for computing random process characteristics. The method takes into account the fact that the analog computer errors are of the order of a few percent. The method is illustrated as applied to a potentiometric multiplier having several multiplication stages. The error due to any one stage is computed by selecting a function converter whose characteristic corresponds to the relative voltage transfer function error of a single stage, and by applying a random noise signal for a preset length of time. The output signal will then represent a random process corresponding to the error distribution of a single multiplier stage. The author presents results of 200 such processes using a 3-stage multiplier. The mathematical estimates of relative error and mean error deviation for this series of processes were 5.05 and 1.4% respectively, which compares favorably with exact theoretical calculations that predicted 5.179 and 1.514% for the same quantities. (Author) Edited machine trans. of Izvestiya Vysshikh Uchebnykh Zavedenii. Priborostroenie (USSR) v10 n4 p73-76 1967.