Research on Smart City Traffic Flow Optimization Based on Hidden Markov Models

With the continuous advancement of smart transportation development and the "Transportation Powerhouse" strategy, traffic flow forecasting plays a crucial role in alleviating traffic congestion and enhancing road resource utilization efficiency. Among existing traffic flow prediction methods, traditional statistical models struggle to capture nonlinear characteristics, while machine learning models face challenges such as high data requirements and parameter sensitivity. Furthermore, Furthermore, existing studies lack systematic model integration and comprehensive comparative analysis. This paper focuses on Hidden Markov Models (HMMs), employing literature review and comparative analysis to systematically examine the application progress and technical characteristics of traditional statistical models, machine learning models, and HMMs in traffic flow forecasting since 1990. Findings indicate that HMMs effectively capture both the randomness and temporal patterns of traffic flows, demonstrating superior stability and adaptability in short-term traffic flow prediction and congestion state identification.Improved HMMs effectively mitigate overfitting through state number optimization and algorithmic enhancements, further elevating prediction accuracy in complex traffic scenarios. The conclusions of this study provide guidance for the rational selection of traffic flow prediction models and lay a theoretical foundation for subsequent research on multi-factor integrated HMM optimization.