关键词: 温室温度, 小波神经网络, 主成分分析法, 黏菌算法

Abstract: The temperature of greenhouse environment is a key factor affecting the growth of greenhouse crops, and its evolution is comprehensively influenced by multiple factors due to obvious nonlinear and complex characteristics, and its prediction is very difficult. Based on this, an improved slime mold algorithm and optimized wavelet neural network model CASMA-WNN for crop greenhouse environmental temperature prediction was proposed. The improved slime mold algorithm was used to optimize the initial values of key parameters of the wavelet neural network, PWLCM chaotic mapping was introduced into the traditional slime mold algorithm to optimize the initial population diversity, and an adaptive hybrid mutation mechanism was designed to improve the search accuracy of the algorithm. At the same time, the principal component analysis method was used to extract the principal component characteristic factors that had the highest correlation with crop greenhouse temperature to achieve data dimensionality reduction. The principal component factors were input into the optimized and tuned wavelet neural network model to predict the environmental temperature of crop greenhouse. The results show that compared with BP neural network and standard wavelet neural network, the average absolute error of CASMA-WNN model can be reduced by 34.23% and 26.56% respectively, and its data fit is higher and the prediction error is smaller, which can provide decision support for the appropriate control of crop growth temperature in the greenhouse.

Key words: greenhouse temperature, wavelet neural network, principal component analysis, slime mold algorithm