Hi Marco,

Very innovative approach !! Did u baseline the results against statistical approach like survival function ? If yes, which performed better ? Also while the accuracy is high, it does not answer the question how much ‘Remaining useful life’ is left.

For e.g. The algorithm might predict my turbine A is gonna fail, but the real question is when ? If i know for sure say my turbine will last 3 more weeks or some x number of more cycles, then i can plan for predictive maintenance more effectively. The algorithm simply does not answer the ‘RUL’ part.

Looking forward to your thoughts on this.