Passive sensing networks can maintain the operation of the network by capturing energy from the environment, thereby solving the energy limitation problem of network nodes. Therefore, passive sensing networks are widely used in data collection in complex environments. However, the complexity of the network deployment environment makes passive sensing nodes unable to obtain stable energy from the surroundings. Therefore, better routing strategies are needed to save network energy consumption. In response to this problem, this paper proposes an IPv6 passive-aware network routing algorithm for the Internet of Things. This method is based on the characteristics of passive sensing networks. By analyzing the successful transmission rate of the network node transmission link, transmission energy consumption, end-to-end transmission delay, and waiting delay of IPv6 packets, the utility evaluation function of the route is obtained. After the utility evaluation function is obtained, the network routing is selected through the utility evaluation function. Then, the utility value and the deep neural network method are combined to train the classification model. The classification model assigns the best routing strategy according to the characteristics of the current network, thereby improving the energy consumption and delay performance of the network.