基于轨道交通的数据集与模型验证系统设计

Design of Dataset and Model Validation System for Rail Transit Applications

摘要:

针对动车组运营过程中产生的多源异构数据（包括结构化、非结构化及半结构化数据）存在处理复杂、验证冗余及报告格式不统一等问题，传统方法需针对不同数据类型独立开发验证逻辑，导致重复工作，且人工整理报告效率低下，增加了人力成本与验证周期。为此，提出一种基于Java语言与Docker技术相结合的方法，构建统一的数据处理框架与标准化模型验证流程。该方法通过数据规则引擎实现原始数据的规范化解析，利用Docker技术部署实现可复用的验证环境，并以走行部数据为例进行系统验证。结果表明，该方案能有效解决多源数据适配与验证流程标准化问题，显著提升效率，具有较高的实际应用价值。

Abstract:

The operation of electric multiple units (EMUs) generates multi-source heterogeneous data, including structured, unstructured, and semi-structured data, which introduces challenges such as complex processing, redundant validation, and inconsistent report formats. Traditional approaches require developing separate validation logic for different data types, leading to repetitive work, while manual report compilation further reduces efficiency and increases labor costs and validation cycles. To address these issues, this paper proposes a unified data processing framework and standardized model validation approach based on Java and Docker technologies. The method utilizes a data rule engine for normalized parsing of raw data and employs Docker for deploying reusable validation environments. The system is validated using bogie data as a case study. The results demonstrate that the proposed solution effectively resolves multi-source data adaptation and validation process standardization, significantly improving efficiency while exhibiting strong practical applicability.