With the accelerated intelligent transformation of the automotive industry, the wheel hubs, as the core safety components of vehicles, have increasingly higher requirements for detection accuracy and efficiency. However, the existing wheel hub detection methods are longer able to meet the demands for efficient and high-precision automatic detection. To address the shortcomings of high cost, low efficiency, and insufficient accuracy in existing wheel hub detection methods, a hybrid deep learning model (CNN-BiLSTM-AM) combining convolutional neural network (CNN), bidirectional long short-term memory (BiLSTM), and attention mechanism (AM) was proposed. A vehicle longitudinal control model was constructed to realize speed tracking, thereby achieving efficient and high-precision automatic wheel hub detection. When using BiLSTM for data modeling, this model employed CNN to suppress noise to solve the noise sensitivity defect of BiLSTM, and introduced AM to focus on critical information to avoid the issue of losing key features in simple models, thereby improving prediction accuracy. Simulation results showed that compared with traditional RNN (recurrent neural network), LSTM, GRU (gated recurrent unit), the CNN-BiLSTM-AM model achieved an average improvement of 3.17% in determination coefficient R² and an average reduction of 29.76% in MAE (mean absolute error) in the overall prediction task. Bench test results indicated that the proposed method could effectively complete wheel hub performance detection, achieving high-precision speed tracking while improving detection efficiency. This method provides an efficient solution for the automatic detection of vehicle wheel hubs.