Anomaly Detection and Diagnosis Method for Internal Logs of Communication Systems Based on Historical Log Analysis

With the deep and extensive application of automated testing frameworks in the research and development of communication systems, the number of test cases has grown exponentially. The resulting massive and heterogeneous internal log files make the analysis of test results exceptionally complex and extremely time-consuming. This paper proposes an innovative method for anomaly detection and diagnosis of internal logs in communication systems based on historical log analysis. By introducing an integrated Robot-assisted testing analysis tool and combining advanced machine learning and natural language processing models, the method constructs an automated process from log collection, parsing, feature extraction to anomaly identification and root cause diagnosis. The core lies in using historical log data to train models, enabling intelligent automatic classification and precise localisation of the causes of test case failures. Experimental and application results demonstrate that this method can significantly improve the efficiency of communication system testing and the accuracy of analysis results, reducing the testing cycle from the traditional one person-month required for manual analysis to 0.25 person-months, providing strong automated support for ensuring the quality of large-scale, highly complex communication systems.