The automotive industry is one of the pivotal sectors of the national economy, and its quality and safety issues have attracted significant attention. Due to the high latency of automotive quality feedback, online complaints from users have become a crucial source of information for manufacturers to identify quality defects. Existing research primarily focuses on the static extraction of automotive quality defects, such as defect identification and classification, with limited attention to the dynamic evolution of quality issues. Therefore, based on the characteristics of online automotive quality complaint texts, this paper applies the latent Dirichlet allocation model to extract quality defect topics, categorizes complaint texts by topics, and calculates the time intervals and sentiment scores of complaint texts under each defect topic, simultaneously depicting the trend of complaints in terms of both frequency and intensity. Since the joint distribution of time intervals and sentiment scores is difficult to determine, we propose a robust bivariate control chart (denoted as BEWMA-LC) based on rank tests and empirical Copula. This method enables simultaneous online monitoring of time intervals, sentiment scores, and their correlations for different complaint topics, while further identifying the root causes of anomalies after triggering an alarm. Using the Volkswagen Sagitar as a case study, based on the online complaint texts from 2010 to 2021 sourced from the “Chezhi” website, we demonstrate the effectiveness of the proposed method in real-time monitoring and diagnosis of dynamic automotive quality changes by comparative analysis with the existing nonparametric control charts.