Exploring intelligent full-chain reconfiguration of analysis and testing in universities: Taking environmental solid waste analysis as an example

University analytical testing platforms are an important component of the national science and technology infrastructure. With the advancement of organized scientific research and the implementation of the national strategy for scientific and technological self-reliance, research demands have shifted from single, routine testing to complex analyses involving multiple technologies and interdisciplinary integration. A complete characterization task in materials, environmental, biological, or chemical sciences often requires simultaneous acquisition of multi-dimensional data, including elemental composition, phase structure, molecular information, and morphological features—no single instrument can independently fulfill such requirements. However, the traditional “demand–plan–execution–report” process in most university testing platforms faces four major challenges: difficulty in demand communication, slow plan formulation, blind execution, and scattered report data. Its service efficiency has failed to meet the stringent national requirements for strengthening organized research and serving strategic demands. This study aims to address these challenges by proposing an intelligent full-chain reconfiguration path for university analytical testing that simultaneously enhances process operational efficiency and research support effectiveness.

This study proposes a “four-dimensional integration and dual-effect promotion” approach with the DeepSeek large language model serving as the core technological engine. The four dimensions are guided demand parsing, holistic plan formulation, visualized process tracking, and in-depth report generation. An integrated intelligent platform is constructed that combines functions such as testing application, plan matching, progress tracking, and report generation. The demand parsing module converts natural language descriptions into structured test instructions through intelligent guidance, transforming multi-round manual communication into human–machine collaborative interaction. The plan formulation module identifies the core scientific question common to multiple testing tasks, evaluates temporal dependencies to provide warnings of sample state degradation resulting from pretreatment steps, recommends parallel sample configurations, and verifies equipment compatibility. The process tracking module decomposes the workflow into standardized nodes, enables real-time progress monitoring, and provides automatic notifications and anomaly warnings. The report generation module aggregates multi-source data, performs cross-instrument correlation analysis, and generates comprehensive reports containing scientific judgments and follow-up recommendations. Taking the analysis of the occurrence states of harmful elements in environmental solid waste as a case study, this study elaborates on how the traditional manual coordination model is reconstructed into a data-driven intelligent process through the proposed path.

The proposed intelligent full-chain reconfiguration path has been officially implemented on an analytical testing platform. Practical application demonstrates that the proposed model significantly shortens the plan formulation cycle, reduces communication costs, and improves report timeliness. The demand parsing process is completed in approximately 10 min under the new model, compared to several days of communication under the traditional approach. The system successfully identifies temporal conflicts in pretreatment steps and recommends appropriate parallel sample configurations, preventing sample failure and reducing material waste. The comprehensive report integrates results from multiple instruments, provides cross-instrument correlation analysis and scientific interpretation, and offers specific follow-up research recommendations. Users receive not only data reports but also integrated conclusions that combine multi-technology results with scientific judgments and research suggestions.

The traditional university analytical testing service faces process pain points, including difficult demand communication, slow plan formulation, blind execution, and scattered report data, with the core problem being overreliance on manual coordination and empirical judgment. Using the DeepSeek large language model, this study constructed an intelligent testing platform with four functional modules to validate the “four-dimensional integration and dual-effect promotion” intelligent full-chain reconfiguration path. Practical data demonstrate that this path significantly improves operational efficiency while enhancing the effectiveness of research support, enabling the systematic transformation of platform functions from instrument management to in-depth research support. Through deep integration and comprehensive analysis of multi-source data, the model provides strong support for generating more forward-looking and insightful scientific research conclusions. This study offers a replicable practical example for university analytical testing platforms to empower scientific research service transformation through artificial intelligence technology. Future work will extend the proposed approach to more complex testing scenarios, providing universities with an actionable practical path for leveraging artificial intelligence to enhance analytical testing efficiency and support organized research.