In order to grasp the environmental data of important places more accurately, an environmental monitoring system based on sensor technology was proposed. In the system, STM32 is used as the main control chip of environmental acquisition, and wind speed sensors and light intensity sensors are used to collect environmental data. The experimental data were collected from six different areas in the smart classroom over a 48-h period, with a sampling interval of 5 min. A total of 1,120 sets of environmental data were obtained, including indicators such as temperature and humidity, light intensity, PM2.5, and PM10. Then, based on the collected air quality data, the air quality prediction model with an improved genetic algorithm is constructed to improve the prediction function of the whole monitoring system. The results show that compared with other optimization methods, the PSO-GA optimization algorithm in this study can reach optimal fitness faster. In terms of prediction model performance, compared with the improved model and other prediction models, the designed SVM air quality prediction model based on the improved genetic algorithm has a smaller prediction error, as measured by MSE and R2, which are 4.6 × 10−5 and 99.15%, respectively. The above results show that the designed environmental data monitoring system can better realize environmental monitoring and can be used in practical scenarios.