Fuzzy c-means (FCM) algorithm optimizes the objective function to obtain the membership degree of each sample point to all class centers, thereby determining the class membership of the sample points to achieve automatic classification of sample data The FCM algorithm is a type of soft clustering method, which has mature theory and wide application, and is a very excellent clustering method. However, the clustering accuracy of FCM algorithm is severely affected by noise and outliers, resulting in poor generalization and robustness when facing different datasets. In order to improve the robustness of the FCM algorithm, this paper proposes an improved fuzzy c-means clustering algorithm based on grid with unknown number of clusters (G-FCM). In the G-FCM algorithm, the difficulty of clustering high-dimensional data is first reduced through grid division. Then, by continuously iterating the cluster centers, the parameter adaptation of the cluster centers is achieved, reducing the impact of noise on the accuracy of the fuzzy clustering algorithm and improving the robustness of fuzzy clustering in high-dimensional data.