Event clustering and event series characterization based on expected frequency

A non-transitory program storage device readable by a computer, tangibly embodying a program of instructions executed by the computer to perform the method steps for clustering time stamps in time series data, the method comprising the steps of:

• receiving, from an electrically connected network, a one-dimensional array t of ordered timestamps and an expected frequency dT, wherein N is a number of timestamps in the one-dimensional array t;
• determining a set of time intervals, wherein a time interval is determined between each pair of adjacent timestamps in the one-dimensional array t of ordered timestamps;
• determining a first binary array that indicates whether each time interval in the set of time intervals is greater than or less than the expected frequency dT;
• determining a second binary array, wherein each element of the second binary array is a difference between a corresponding pair of adjacent elements of the first binary array;
• appending an ith timestamp to a set of opening interval bounds τ− if a corresponding second binary array is equal to −1;
• appending an ith timestamp to a set of closing interval bounds τ+ if a corresponding second binary array is equal to 1;
• appending an ith timestamp to a set of isolated points x if a next corresponding first binary array value is equal to 1;
• merging the set of set-e-opening interval bounds τ− and the set of closing interval bounds τ+ into a set of cluster intervals τ;
• and
• outputting the set of cluster intervals τ and the set of isolated points x;
• and using said set of cluster intervals and said set of isolated points to determine a reliability of said electrically connected network and generate messages indicative of the reliability of said electrically connected network.