Low-voltage substation area is an essential part of the power system, and the secure operation and maintenance of it is emerged as a critical problem. High-frequency data acquisition provide accurate monitoring for low-voltage substation area, while these collected data need to be efficiently aggregated for processing and analysis. Data compression is an advanced method to realize low-latency data aggregation. Therefore, we designed a dynamic scale data compression model and the data aggregation delay model. Then, an optimization problem is formulated to minimize the average data aggregation delay by jointly optimizing data compression strategy and routing selection. A low-latency DQN-based data compression and aggregation algorithm is proposed to solve it. Simulation results show that the proposed algorithm has superior performances in delay performance.