Suitable soil moisture content (SMC) can not only increase the ability of tea tree roots to absorb and utilize nutrients but also improve the utilization rate of soil nutrients, which can ensure a continuous and stable yield and tea leaf quality. Traditional methods for predicting soil water content generally have low accuracy and efficiency problems. A real-time soil information collection system based on a wireless sensor network was built, and a new predicting SMC Model (AO-SVM) for tea plantations using support vector machine optimized (SVM) by Aquila Optimizer (AO) was constructed and evaluated. The SMC prediction model was established using weather data, soil temperature (ST), soil electrical conductivity (SEC), and PH value (pH), and soil water potential (SWP), and so on. First, the correlation between individual SMC, ST, SEC, pH, SWP was analyzed and parameters with high correlation with soil water content were subsequently identified. The AO-SVM model was utilized to predict the soil moisture content. The experiments showed that the R2 of AO-SVM model proposed in this paper is 0.925. It indicates that the AO-SVM model is effective and feasible and achieves advantageous performance over long short term memory (LSTM), generalized regression neural network (GRNN), the opposition-based chaotic salp swarm algorithm optimized SVM (OCSSA-SVM), sparrow search algorithm optimized SVM (SSA-SVM), particle swarm optimization SVM (PSO-SVM), and the whale algorithm optimized SVM (WOA-SVM) model, which can help guide the irrigation and fertilization management of tea plantations.