In order to improve precision and efficiency of temperature prediction

propose a new temperature prediction method for greenhouse

which uses particle swarm optimization（ PSO） algorithm to optimize the least squares support vector machine（ LSSVM） under the condition of principal component analysis（ PCA）. The model parameters of the LSSVM are optimized by PSO algorithm

and the optimal parameters combination is used to construct the nonlinear prediction model for temperature and its impact factors. The six environmental parameters in the artificial greenhouse are collected by using the greenhouse intelligent monitoring system

and the above model is validated by the measured data. The experimental results show that

compared with the prediction models of PCA-LSSVM

and PSO-LSSVM

the proposed prediction model is the best under the same condition. The three models evaluation indexs are superior to other prediction methods. The temperature prediction model based on the PCA-PSO-LSSVM has great advantages in global optimization

convergence speed

self-learning ability

selfadaptability

and has high prediction precision.