ֺ뻷
JOURNAL OF GEOLOGICAL HAZARDS AND ENVIRONMENT PRESERVATION
1999 10 1 VOL.10 No.1 1999



֪ʶĻʩо* 

ժҪĿǰʩٵ ܹCLIPSԽһ򵥵Ļ֪ʶĻʩҪڽ ݵԶԤ⡢ԶԼѧϰܵʩӦʩ
ؼʡӹ̡ʩ⡡ܿ
ͼšTU712.2TU473.2
RESEARCH ON A MONITOR BASED ON KNOWLEDGE OF EXCAVATION CONSTRUCTION
Wang Xiaohong
(Department of Geotechnical Engineering,Tongji University,Shang hai200092)
AbstractWith artificial intelligent tool CLIPS,this pape r builded a construction monitor of excavation engineering,which can mainly be u sed to solve some current problems,such as self-correction of monitor data,forma tion forecast,formation self-warning and self-learning construction measure.
Key WordsExcavation engineeringConstruction monitor  Intelligent control
ڱճлӹʩڵʡʩصӰ죬ѻ߲ܵԤӰĳ̶ȡˣбҪʩиʩԿܵزλƺ͸¹ܵȱıμ⣬ȽϼݡԤֵֵܿɿطӳʩɵӰ죬׼ȷʽӳӰĳ̶ȣԱ֤ʩΧİȫͬʱӰصģԡȷԣƼʩԱľҲΪҪˣһ֪ʶļϵͳʮֱҪġ
1994ͬôѧϺѧϺ˾ͬƿĶܹʩĻ֪ʶϵͳڵʩȡϲ˵ĳɼڻʩϺĳʹв˼⡢ݲɼ䡢ȫԶķԽĿõģ ȴĿǰһ׷λܳܣңĿǰڵĴ󲿷ּĿȻרҵ ԱɼݣȻΪ󣬵ǳɱϵͣȽʺҹ顣 һּ򵥵ġܹʵʩļϵͳнϴʵ塣
ԭ򣬱˹ܹCLIPS (the C Language Integrated Production System)ԣҪڽݵԶԤ⡢ԶԼ ѧϰܵĽʩʩ
1֪ʶı
(1) CLIPS Լ飺CLIPS պ(NASA)Լѷռģ80ڿһּɻϵͳCLIPSԹΪҪ֪ʶ﷽ʽͬʱҲṩʵ顱롰ʵģ塱ʵǰģʽԣRETEģʽƥ㷨(Rete matching algorithm)ָЧķCLIPSʮʺϽʵʱϵͳʵ֡
(2) ֪ʶı֪ʶʾɰͼ1趨а֪ʶб롣


ͼ1Ŀ봫ϵʾͼ
Fig.1Diagram showing relations of monitor item with sensors
Ŀ壺
(item <name> <status>)
У<name> Ŀʶi1i2i3ȣʾһʩļݡҪѡ Ρˮƽָ֧ꡣ<status>˵Ŀǡokwa rningdangerous
ĿӦĴĹϵɣ
ʵ(sensor <name> <associated-item>)ʾУ<name>Ǵʶ<asso ciated-device>Ǵ豸ıʶ磺
(sensor si1 i1)
(sensor si2 i1)

(sensor sij i1)
磬ڻ i1si1si2sijֱʾλڲͬȵĴ
ʵ(ranges <name> <RW> <RD> <SW> <SD>)ʾľֵ<name>ʾ  ʶ<RW>(RATE-WARNING)仯ʵľֵ<RD>(RATE-DANGEROUS)仯ʵΣֵ <SW>(SUM-WARNING)仯ۻľֵ<SD>(SUM-DANGEROUS)仯ۻΣֵЩֵһԱĸݡӹֲᡷ֪ʶдһΧ-ʽֵҲ湤ӦĿʵʵΣֵ뾯ֵơʩ
(condition <name> measure <answer>);(condition <name> decis ion <question> <ok> <cancel>)ʹ״̬µʩʩѧϰУ<name ><answer>ǸʱĴ𰸣У<question>ʾ⣬<ok><cancel>ǶԸп϶񶨻شʱҪ
2ʵ
2.1ݿɿȵб
ȷɿļݣǽмصǰͱҪüרȨֵ(salience) 10Ĺ򣬷ֱбݵĿԡ
(declare (salience 10))
(phrase input)
(data-source user)


*(printout t what is the value for sensor? id ?)
(bind? value(read))
(if (not (numberp?value)) then (halt))
(if (eq?item no-direction) then
(if(CONT?value) then (assert (original-value?id?value?time)))
else ((printout t Your original-date is probably wrong!)
(halt))
else (if (CONT1 abs (?value)) then
(assert (original-value?id?value?time)))
else ((printout t Your original-date is probably wrong!)
(halt))
ùȨֵ10ʹÿֻҪ׶ΣҪݵȻڶ֣бֵǷΪֵ(numberp)ǣ˳(halt)ʵ(no-direction)ݴԵطΪࣺһûз(ֻһ仯)һзģӵλƣȵȡCONTCONT1ɼרƾȻҲɱһ-ʽãⷽ٣в߱ܡ 
2.2Ԥ
ԤʩҪɲ֣Ŀǰ÷ԭԽΪӣҲΪʺֳʩԱʵʱơ
ͨԻʩʵʱӦϵͳʶķƳѧģͣԻӵıԤ⡣
ٶʩӰ廷λơŶӦ͸ѹ֮໥ϵ໥Ӱ죬ԣѧģͿΪ
Xt =A1Xt-1+A2Xt-2+ +AnXt-n+t
ʽУXt=(12m12m12 m)ti,i,i(i=1,2,m)ֱΪmõĻλơŶӦ͸ѹAi(i=12n)Ϊģϵ(3m 3m ׾)nΪģͽtΪģʸһסͳΪ
E(t)=0E(ttT )=Qn
ʽУQnΪ3m ά󣬳ΪģЭ
ģͿԽжԤ⣬ʹͱ߽ʵʱеУ˳ֵѺͲ㡣Ҫǣģʹü򵥡㣬Һ١
CLIPSṩһϵзûƵʵSYSTEMֱӵģͱдĳKALMAM.EXE
(openorig.datorigr)

(systemkalman.exe)

(openresu.datresuw)
2.3׶εı
ԤֵȽϣжǰһʩպʩǷԤҪȷŻһʩأԤֵԶжδOKWARNINGDANGE ROUS֪ʶļ򣬷ִOKWARNINGDANGEROUS
(1)(phrase trends)
 (cycle ?time)
 (pro-sensor-value ?sensor ?pro-value ?time)
 (pro-sensor-svalue ?pro-svalue ?time)
 (ranges ?sensor ?rw ? ?sw?)
 (test (and (> ?pro-value ?rw)
 (> ?pro-svalue ?sw))
ݡ
 (assert(sensor-state ?sensor ok ?pro-value?time)))
(2)
 (ranges ?sensor ?rw ?rd ?sw ?sd)
 (test (or (and (> ?pro-value ?rw)
  (< ?pro-value ?rd))
(and (<?pro-svalue ?sw)
ݡ  (<?pro-svalue ?sd))))
 (assert (sensor-state ?sensor warning ?pro-value ?time)))
(3)
 (ranges?sensor ? ?rd ?sd)
 (test (or (> ?pro-value ?rd)
 (>?pro-svalue ?sd)))
ݡ
 (assert (sensor-state ?sensor dangerous ?pro-value ?time)))
2.4׶εĴʩ
WARNINGDANGEROUSʵǸԱļ뾭ȷģʾγ̶Ȳͬѡˣÿͬ͹Ĵʩ֮ԲͬĲȷΪߵʹͬԤ֪ʶ֯ṹԴЧʡ⣬Ĵʩ֣Ҳṩѧϰܣûġŵص㣬Ӧ֪ʶµصֲ֪ʶĲ㡣һҪǿǣЩʩֱֻܴۡ㡢Чķʽ⣬꾡ѧͼ㡣
ⲿֿνƵһҪĿģҲֳ֪ʶҪԺͱҪԡ
ͼ2Ǽԭʼ֪ʶṹģ͡ͷ߱ʾѧϰǰȡĴʩ

ͼ2֪ʶṹģ
Fig.2Knowledge structure model of the monitor
ʵܣִоҪĶڴʩģ۵ʩȷ򲻺ʱһµľԱʾѧϰܡ
ͶӦļʵ⣬½(αʾ)
clips > (reset)
clips > (run)
Now are we thinking about the construction environment (yes or no)? yes
Is building (yes or no)? no
Is road (yes or no)? no
Is subway (yes or no)? no
Is sub-pipeline (yes or no)? yes
Now we can deal with the problems with !
Am I correct? no
What do we take measure? Or do you add some condition (w or d) d
Rigid of flexible
What do we take measure if
