关键词: 最小二乘支持向量机, 模糊隶属度, 先验知识, 噪声, 故障诊断

Abstract: To address the drawback that the Least Squares Support Vector Machines (LSSVM) is sensitive to noises or outliers, a LSSVM model incorporating with prior knowledge on data was proposed based on noise distribution and sample affinity. Information of noise distribution for samples was introduced in the training process. A strategy based on the sample affinity was presented to discriminate data with noises. A fuzzy membership was automatically generated and assigned to each corresponding data point in the sample set by using the strategy and the noise model. The ability of FLSSVM was improved to resist noises. The flexibility was increased to treat data points with noises or outliers. The proposed method was applied to fault diagnosis for the lubricating oil refining process. The experimental result shows that the proposed method has better robustness.

Key words: Least Squares Support Vector Machines （LSSVM）, fuzzy membership, prior knowledge, noise, fault diagnosis