PURPOSE: To detect the abnormality with high accuracy even with a program having its large executing time fluctuation by considering not only the next estimated executing time decided from the shift average of the past program executing times but the executing time fluctuation caused by the increase of load.
CONSTITUTION: A time monitoring system 101 starts the time measurement via a counter 114 by a monitor start instruction given from a CPU 102 and with rises of a SEL signal line 105 and a start signal line 106. A limit value calculator 125 starts at the fall of a signal line 116 to obtain the average program executing time and the standard deviation from the information on the past program executing times stored in the registers R0WR9 and multiplies these a values by K to obtain the limit monitor time value and to estimate the next program executing time. While the program executing time is compared with the limit value by a comparator 128. When the abnormality is detected, the signal is sent to the CPU 102 from the system 101 selected by a decoder 103 via a signal line 109.
UENO HITOSHI
SAKURABA TATSUTOSHI
SOMA MASATO