Tins BRIEF summary of the IFAC Congress provides a broad coverage although the theoretical developments are not emphasized. Note that other more detailed reports of individual sessions and Round Table discussions have been and will be published in Automatica whenever they become available. (G. S. Axelby---editor).
An estimated 1300 persons, representing 36 countries, met in Boston in August for the triennial IFAC Congress held for the first time in this country. ISA served as the administrative society. Attendees heard papers from 32 countries devoted to the theme, Control technology in the service of man.
Computer control. Major subthemes--multiprocessor hierarchies, distributed microprocessors and software for computer control--were explored in great depth in separate sessions, while other computer sessions were devoted to applications in nuclear power generation and in energy control centers.
In the session on computer hierarchies, S. Moriya of Tokyo Gas reported on a similar dispatch and distribution control center, but for a gas rather than electric utility. This very elaborate hierarchy has been operating successfully for about a year and includes large-scale computers (IBM Systems 360 and 370), medium computers (IBM 1800s and 1130, TOS-BAC), and minicomputers (YOD1C, HIDIC and IBM Systems 7).
The highlight of the microcomputer session was brief introductory paper by A. Larson of the U.S. Air Force Academy, putting microprocessors and microcomputers in perspective. He predicted instruction cycle times of about 100 nsec within 3 yr, based upon bipolar IC technology; and large single-chip MOS memories with 64 bits at less than 0.1 cents per bit.
The process control software session included a discussion of operating systems for a microcomputer network by J. D. Schoeffler of Case Western Reserve University and Two tutorial papers. The first, by J. Gertler of the Hungarian Academy of Sciences, was a review of basic programming techniques for the four major functions of process control: mathematical modeling, system identification, digital filtering and algorithmic control. The second, written by J. G. P. Barnes and M. W. Sage of Imperial Chemical Industries, was a survey of available real-time programming languages for process control.
Seven midweek sessions spanned the full spectrum of considerations in control design, encompassing theoretical, social and practical aspects. At the theoretical end, state estimation and optimal and stochastic control were covered; at the opposite end, five papers described testing and reliability in computer control.
Systems reliability was the subject of five papers at one session. Four of them covered computer-aided fault tree analysis in great detail, although the speakers pointed out that in most cases a numerical solution is not sought: the safety and reliability analysis is pursued only to the point of locating bottlenecks. Particularly illuminating was a paper by B. V. Koen of the University of Texas which showed how a list processing technique can organize the computer memory in a way that represents a fault tree, but does not require massive shifts of data to conduct the analysis. After mapping the fault tree into a list structure, this data reduction technique collapses the tree by means of pattern recognition programs. List processing makes possible a fault tree analysis approach which is not feasible with FORTRAN.