The invention relates to a health state evaluation method, system and device for a numerical control machine tool and a medium, and the method comprises the steps: training model parameters of a health state analysis model based on each pre-obtained training sample data set, and calculating a corresponding likelihood probability, forming a preset model parameter set representing different preset working condition states of the machine tool by each group of trained model parameters, and forming a preset likelihood probability set by each likelihood probability; and based on the obtained preset model parameter set and the preset likelihood probability set, evaluating the health state of the current numerical control machine tool. According to the method, the historical monitoring data of the numerical control machine tool is utilized to construct the multi-dimensional data vector, the health state of the numerical control machine tool is evaluated by calculating the health degree, the number of machine tool sensors does not need to be increased, the model has wider practicability, the calculation complexity is reduced, and accurate, rapid and real-time health degree evaluation is achieved; the method can be widely applied to the numerical control machine tool health state monitoring field.