In most CHF experiments, the CHF detection is usually accomplished by measurement of temperature using thermocouples, resistance temperature device(RTD), and etc. There is some ambiguity of human subjectivity in the experimental decision of CHF occurrence. This judgment can cause lack of consistency and objectivity in experiments. In this regard, we investigate the CHF condition, especially the LPLF condition. From the investigation of the CHF condition and conventional definition of the CHF, we develop the temperature pattern recognition systems, which are able to detect the CHF occurrence. The CHF patterns are recognized using spatiotemporal neural network (STN) and wavelet transform. Each CHF detection method shows good agreement with human decision. (C) 2000 Elsevier Science Ltd.