A novel approach for pipeline generation using a mixed GA and tree strategy

A municipal pipeline is a typical and vital component of the urban lifeline in the urban infrastructure system. The rational layout of the municipal pipeline lays the foundation for the safe and resilient operation of the urban infrastructure system. Unfortunately, current municipal pipeline route layout relies heavily on engineers' knowledge and abundant data formed in practice, which is time-consuming and error-prone. With the fast development of urban infrastructure systems, a more efficient approach for municipal pipeline generation is in high demand when the topology of the municipal pipeline becomes more complex. However, artificial intelligence-based pipeline generation algorithms require being trained with a vast amount of data in the database, which will result in a significant amount of computing power. This paper proposes a topology generation algorithm for municipal pipelines based on the genetic algorithm (GA), which incorporates a priori experience using a tree generation strategy. In addition, the algorithm does not require any data training. The proposed algorithm is validated using a real case from Jiulonghu Campus, Southeast University, China. Results show that using two main pipeline expansions, choosing the correct number of branching points, and imposing a penalty term can effectively improve the topology generation. Practitioners can use this approach for municipal pipeline layout design to enhance its efficiency and effectiveness. Ultimately, the method proposed in this paper can inspire subsequent research by integrating a priori knowledge into the topology and objective function of the proximity algorithm.