This paper studied the effects of steel slag, slag, desulfurization gypsum and fly ash on properties of mortar. The results showed that the compressive strength and flexural strength of steel slag-cement composite material declined with the increasing of steel slag.  The 28d compressive strength and flexural strength of steel slag-cement composite material with 20 wt.% slag is 49.2MPa and 6.8MPa respectively, which increased by 3.3% and 16.2% compared with that without slag. Mechanical properties of steel slag-cement-slag composite material could be improved when the content of desulfurization gypsum was 3wt.%. When the content of plasticizer was 0.4wt.%, the workability of cement mortar was good. In addition, the 28d compressive strength of steel cement mortar reached to13.5MPa（meeting the requirements of M10）when the sand ratio was 1:4. Besides, the 28d compressive strength of steel cement mortar reached to 7.5MPa（meeting the requirements of M5） when the sand ratio was 1:5.

The Back-Propagation Neural Network is utilized to research rock fragmentation prediction in cement mine. The paper establishes a three-layer feed forward BP Neural Network model by Matlab software. After the model is trained it is used to simulate and predict measured sample .The prediction result proves that using the model to predicate rock fragmentation distribution is completely feasible and has better precision. The model also can be used to auxiliary verify optimal design of mine blasting parameters .

This paper analyzed the operation difference in setting time test of cement paste. It was found that forming sample’s operation difference was the main reason why inspection result of setting time was scattered. In addition, water consumption of normal consistency using measuring water usage was a bit more than that of using weighing water usage. However，the test results of normal consistency were not affected by the differences of water consumption with these two methods,  and the cement paste setting time was almost not affected.