Abstract: For the problem of the large quantity of characteristics needed in the process of engine performance evaluation and prediction, and the non-uniform standard of optimal selection and evaluation of characteristic parameters, this paper presents an optimal selection method of engine state characteristic parameters based on multi-index comprehensive evaluation. A certain engine is taken as the research object, and 12 original characteristics parameters of this engine are analyzed to establish a series of evaluation indicators from the aspects of correlation, monotonicity, discreteness, etc. to realize object evaluation of the state characteristics. The entropy weight is proposed to determine the weight of multiple indexes objectively in the perspective of information content. And the optimal ranking of the state characteristic parameters is realized through ideal points and similarity. By weighted fusion of the selected feature parameters by similarity index, the progress fuzzy rule model is established and the engine life prediction is realized

Key words: engine, life prediction, entropy, ideal point, feature evaluation, PFR