A fuzzy neuron inference of physiological process (FNI-PP) based virtual plant growth model is proposed. Using machine learning theory, the model can automatically learn and fit the plant growth function according to measured data and extract the plant growth rules. During plant growth, the source and sink organs respond the surrounding virtual environment according to its inbuilt growth function, and produce, allocate and consume assimilates as well as update the L-grammar representing the plant structure. The model can automatically adjust parameters of the growth function and the L-grammar to respond the environmental heterogeneity. Cayenne-based simulations show that the model can accurately extract the growth function and the structural pattern of the plant, and vividly demonstrate the response to environment.