A novel fuzzy support vector machine based on unbalanced samples(FSVM-US)is proposed to solve the high false positive rate problem since the gyroscope output is susceptible to unmanned aerial vehicle(UAV)airborne electromagnetic environment and the gyroscope abnormal signal sample is rather rare.Firstly,the standard deviation of samples projection to normal vector for SVM classifier hyper plane is analyzed.The imbalance feature expression reflecting the hyper plane shift for the number imbalance between samples and the dispersion imbalance within samples is derived.At the same time,the denoising factor is designed as the exponential decay function based on the Euclidean distance between each sample and the class center.Secondly,the imbalance feature expression and denoising factor are configured into the membership function.Each sample has its own weight denoted the importance to the classifier.Finally,the classification simulation experiments on the gyroscope fault diagnosis system are conducted and FSVM-US is compared with the standard SVM,FSVM,and the four typical class imbalance learning(CIL)methods.The results show that FSVM-US classifier accuracy is 12% higher than that of the standard SVM.Generally,FSVM-US is superior to the four CIL methods in total performance.Moreover,the FSVMUS noise tolerance is also 17% higher than that of the standard SVM. A novel fuzzy support vector machine based on unbalanced samples (FSVM-US) is proposed to solve the high false positive rate problem since the gyroscope output is susceptible to unmanned aerial vehicle (UAV) airborne electromagnetic environment and the gyroscope abnormal signal sample is rather rare . Firstly, the standard deviation of samples projection to normal vector for SVM classifier hyperplane is analyzed. The imbalance feature expression reflects the hyper plane shift for the number imbalance between samples and the dispersion imbalance within samples is derived. At the same time, the denoising factor is designed as the exponential decay function based on the Euclidean distance between each sample and the class center. Secondarily, the imbalance feature expression and denoising factor are configured into the membership function. Since the sample has its own weight pointed to the classifier .Finally, the classification simulation experiments on the gyroscope fault diagnosis system ar e conducted and FSVM-US is compared with the standard SVM, FSVM, and the four typical class imbalance learning (CIL) methods. Results show that FSVM-US classifier accuracy is 12% higher than that of the standard SVM. Normal, FSVM -US is superior to the four CIL methods in total performance. Moreover, the FSVMUS noise tolerance is also 17% higher than that of the standard SVM.