For guiding precise irrigation in greenhouse, the environment factors such as air temperature, humidity and light were taken to establish regression models with water evaporation amount in cultivate substrates. The evaporation amount of water in greenhouse was used to forecast amount of evaporation of other cultivated substrates. Water consumption of tomato plants was simulated using BP neural network for its non-linear. Tap water evaporation can be used to instruct irrigation and BP neural network is suitable for plant water consumption simulation, and farmers can make use of all of them simply and cheaply.