Coefficient of road adhesion is an important parameter affecting vehicle handling stability and safety. This paper designs a method for estimation of coefficient of road adhesion based on stacked generalisation and bayesian optimisation. Firstly, by changing the road adhesion coefficient and vehicle speed, the estimation simulation test is designed. Secondly, the long and short-term memory network and convolutional neural network model are selected as the base learner to initially estimate the coefficient of road adhesion, and the preliminary estimation results are later utilized to train the meta-learner to reduce the estimation error. Finally, optimizing meta-learner hyperparameter models using Bayesian algorithms. The validity of the model proposed in this paper is verified by simulation tests.