Abstract: Massive real-time monitoring data of a concrete lifting cable crane provides as a direct, reliable research basis for analysis of its practical operation efficiency. How to identify accurately and quickly its working cycle and extract useful information from these data is a key issue for efficiency analysis. Therefore, we develop a pattern recognition method. The working state of a cable crane is defined by the rates of changes in the lifting weight and its average; its concrete lifting modes defined by a time series of its characteristic changes. Then, we extract the statistical law of weight lifting parameters and define a set of pattern recognition standard templates; a pattern recognition model of lifting concrete featured with state similarity recognition plus state penalty function is developed. Especially, in view of the false fluctuation in the monitoring data of the load caused by the wind or cable acceleration, and the false stability of the load caused by accidental pauses in weight lifting or unloading, we construct a method of discrimination of isolated points plus multiple iterations to identify and correct a working state. Application of this model and measured data analysis show that it identifies accurately and quickly the working modes of a cable crane, the start and end points of each working state, and its instant and average lifting weights, thus providing important support for practical operation efficiency analysis and scheduling optimization of cable cranes.

Key words: construction for hydraulic works, massive monitoring data, cable crane, lifting concrete, pattern recognition