PURPOSE: To realize online decision of stability by employing a neural network, receiving a state amount of generator calculated based on current system conditions and outputting a stability decision result, as a stability deciding means thereby deciding the stability of power system at high speed with high accuracy.

CONSTITUTION: System information collected at a system information collecting section 1 is stored in an information storing section 2 and taken into a power flow state operating section 3 and then current power flow state of system is calculated based on the system information and system facility data fed from a system facility storing section 4. A transient stability calculating section 5 performs transient stability calculation upto an instance of fault removal for multiple predicted faults stored at a predictive fault storing section 6 based on current power flow state. A step-out mode deciding section 7 for selecting a possible step-out mode based on the state amount of each generator calculated at the transient stability calculating section 5 and a stability deciding section 8 receiving the results from the step-out mode deciding section 7 to decide the stability are constituted of neural networks and display decision results on an output unit 9.