The use of data analytics and machine learning to identify unusual spikes or patterns in energy consumption that deviate from expected baselines. In thermal systems, this process is vital for detecting insulation failures, steam leaks, or equipment malfunctions in real time.