In order to solve the problem of the health status of single sample equipment, a new method of health management and forecasting based on hidden Markov is proposed. The method uses the equipment monitoring data to construct the hidden data. The Markovian health assessment model is used to simulate the prediction accuracy of different observation sequences and different observation times to determine the optimal model parameters. Then, the real-time data is substituted into the model, and the results are calculated from the model to determine the health of the equipment. Finally, the current data and historical data are fitted to predict the safe and reliable life of the system. The method can effectively solve the health assessment of single-sample multi-state equipment.