Power transmission line tower arrangement method and system based on deep reinforcement learning

By constructing a simulation environment and policy network based on deep reinforcement learning for pole ranking, and designing a multi-objective reward function, the computational complexity and constraint handling problems of traditional methods in complex terrain are solved, and the automation and efficient optimization of pole ranking are realized.

CN122452382AActive Publication Date: 2026-07-24四川电力设计咨询有限责任公司
3 Cites 0 Cited by

Patent Information

Application Number
CN202610921273.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-25
Publication Date
2026-07-24
Estimated Expiration
2046-06-25

AI Technical Summary

Technical Problem

Traditional tower placement methods suffer from high computational complexity in complex terrain, weak ability to handle multiple constraints, and lack of strategy flexibility. As a result, the generated schemes require extensive manual correction. Furthermore, existing reinforcement learning algorithms are unstable during training in tower placement tasks and are difficult to meet engineering safety constraints.

Method used

A deep reinforcement learning-based approach is adopted to construct a simulation environment and policy network, design a multi-objective composite reward function, train the policy network through course learning policies, generate continuous actions to meet complex constraints such as electrical insulation and mechanical strength, and realize global terrain perception and optimization through the Transformer architecture.

Benefits of technology

It has achieved automation and intelligence in tower placement, and the generated schemes meet the practical requirements in terms of engineering feasibility and economy. The design efficiency and quality have been greatly improved, and multiple candidate schemes can be generated and compared quickly.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a power transmission line tower arrangement method and system based on deep reinforcement learning. The method comprises the following steps: constructing a simulation environment; designing a strategy network based on Transform, and outputting span and tower height continuous actions; a multi-target composite reward function is constructed, and a compliance evaluator is integrated; a course learning strategy is adopted, reward parameters are adjusted in stages, and the strategy is guided to be transited from security priority to economic optimization; and post-processing and evaluating the reasoning result. According to the method, the global perception capability of Transformer is utilized to process complex constraints, a single tower decision is at a millisecond level, on the premise that the tower arrangement success rate is larger than or equal to 98%, the number of towers is reduced by 10%-15%, the engineering cost is reduced by 5%-12%, the generalization capability is high, different voltage levels and terrain types can be rapidly adapted, and automation, intelligence and cost optimization of transmission line tower arrangement are achieved.
Need to check novelty before this filing date? Find Prior Art