Singular point-containing electronic signal fault hybrid prediction method and verification device

The invention discloses a singular point-containing electronic signal fault hybrid prediction method, and the method comprises the steps: taking a cubic non-diversified spline function as a predictionunit of OS-ELM output, and carrying out the calculation of a second derivative of a spline model, so as to construct a series of observation frames. According to the method, an improved particle swarm algorithm is adopted to optimize a hidden layer of OS-ELM to help predict the trend of an observation sequence and reconstruct a spline function. In a subsequent verification part, two numerical simulation cases and a practical application exist, time series data of the numerical simulation cases and the practical application have singular points, and the effectiveness of the method is proved through the cases. 本发明公开了一种含有奇异点的电子信号故障混合预测方法，这种混合方法将三次非多样式样条函数作为OS-ELM输出的预测单元，对样条模型的二次导数进行计算以构建一系列的观测帧。同时，本方法采用改进的粒子群算法优化OS-ELM的隐藏层，来帮助预测观测序列的趋势并重建样条函数。在后续验证部分，有两个数值模拟案例和一个实际应用，其时间序列数据都具有奇异点，本发明分别以这些案例来证明所提方法的有效性。