This　paper　presents　an　algorithm　for　automated　fault　tree　synthesis　using　information　from　Operability Study．The　algorlthm　is　implemented　in　computer　programs　written　in　Prolog、Modified　Opembility　Study provides　input output　type　of　lnformatoion　for　a　plant，The　information　is　expressed　in　logic　formats（viz、，cause or　consequence　e（luations），and　stored　in　databases　in　computers．Search　is　carτied　out　through　the　databases，and fault　trees　are　generated　sequentially　by　developing　the　fault　events　resulting　from　the　search．Then，fault　tree synthesis　is　demonstrated　for　a　sodium　bicarbonate　pilot　plant　using　the　proposed　methodology，

As　occurrence　of　leak　from　fuel　tank　etc。combines　with　a￠cident，it　is　urgent　and　necessary　to　develop　air －tightness　testing　device　which　has　high　sensitivity　for　detection、The　air－tightness　testing　device　reported　before could　respond　to　that　request．For　making　that　device　practicable，it　is　important　to　investigate　the　effect　of surrounding　temperature　and　to　establish　the　method　of　correction　for　temperature，In　the　range　between15 ～40℃，the　effect　of　temperature　was　investigated　experimentally，then，the　method　of　coτrection　was　discussed、 As　Kalman　filter　has　state　equation　of　system　in　itself，it　is　one　of　the　realistic　methods　to　correct　the　parameter in　this　state　equation　according　to　surrounding　temperature、By　experiment，it　was　made　clear　that　the　relation between　the　parameter　and　surrounding　temperature　is　denoted　by　linear　equation（see fig．5），then　this　equation was　bui1Unto　Kalman filter　for　the　correction　of　parameter．As　the　result，it　became　possible　to　diagnose　in　above range　of　temperature　with　high　accuracy（see　tabel4）．It　is　also　confirmed　that　this　method　is　able　to　use　to diagnose　even　in　a　case　conventional　method　which　judges　by　only　comparing　the　pressure　level　can't diagnose．

This　paper　deals　with　detection　of　faults　in　sensor／actuator　area　of　digital　control　systems．Fault　detection systems　using　Kalman filter　has　been　proposed，but　it　camot　apply　to　real　systems　in　many　cases　when　even　a permissible　change　of　parameters　occurs．Then，this　paper　proposes　a　new　fault　detection　system　using　an extended　Kalman　filter．New　one　can　apply　in　above　cases　which　old　one　can't（see　fig，4 and 5）．Computer simulation　indicates　good　performance　and　abilities　of　the　new　method（see　fig．8，9and　lO）.