Rule-Based Monitoring of Trigger-Rates at ZEUS
Paper: 338
Session: B (talk)
Speaker: Behrens, Ulf, DESY, Hamburg
Keywords: diagnosis, monitoring systems, object-oriented methods, programming languages, expert systems
Rule-Based Monitoring of Trigger-Rates at ZEUS
Ulf Behrens (+), Mariusz Flasinski (*), Lars Hagge (+), Janusz Jurek
(*)
Dirk Kahnert (+), Kars Ohrenberg (+), Pjotr Wojtowicz (*)
(+) Deutsches Elektronen-Synchrotron (DESY), Hamburg, Germany
(*) Dept. Computer Science, Jagiellonan University, Cracow, Poland
A tool for monitoring and surveilling the first-level trigger rates of
the
ZEUS experiment has been developed using a rule-based approach.
Experience
gained from running the experiment has been used to identify nominal,
luminosity dependend values and spreads for the individual trigger rates.
Current trigger rates are monitored and forwarded to a (rule-based)
inference
engine, which issues warnings or alarms whenever individual triggers are
deviating too much and/or too long from their nominal values.
Trigger rates have been modeled using Harel state machines, which have
then
been turned into rules. Compared to conventional programming techniques,
rules
have the advantage that they don't introduce an order (ie sequence) to
the
program. The system was implemented within two months using a commercial
expert system shell. Data transport to/from the inference engine is
implemented
in C++. The system was introduced into the ZEUS environment towards the
end of
the 1996 running period.
The tool is a helpful support for the shift crew running the experiment.
The
talk will discuss the tool's architecture, the rule-based paradigm and
the
technique of deriving rules, and some running experience. Using
rule-based
programming has shown to be an easy and efficient way to create simple
and
helpful tools like this. It is a thing to look at, even more as rule
libraries
for C++ are appearing on the market.