just my ideas to that task: you need a definition of your plateau. I assume a curve maybe like a temperature plateau due to phase changes during heatup or cool down. So in real world of measurements you might never get the same reading for a periode of time. Or you get only a smaller increase or decrease of your values.
look at absolute of the first derivative of your (maybe filtered eg. smoothed) curve. that gives you a zero at a plateau or at least smaller values. now find the minima (zeroes) , decide if the minima are small enough for your plateau definition. starting from these points go 'left and right' on your curve until you reach your 'plateau limits'. Or determin the bounderies of your plateau by a minima threshhold of your (absolute) derivative.
