Incipient fault dataset for analog circuits

This dataset contain the pulse responses of the Sallen-Key bandpass filter circuit and the amplifier board circuit. The test excitation is a 10 us pulse signal with an amplitude of 5 V and a frequency of 5 kHZ that exhibits abundant frequency components. By observing the pulse response, the sampling frequency is set to 5 MHz and the number of sampling points for each sample is fixed at 1000 in Case 1. PSPICE is applied for circuit simulation to set up the circuit fault according to the range of fault component parameter values. Through the Monte Carlo analysis, 300 responses are generated for each fault class using PSPICE to construct a dataset in Case 1, which is randomly divided into training and test sample sets in the ratio of 7:3. In Case 2, the sampling frequency is set to 500 KHz. According to the range of fault component parameter values, the components whose parameter values are within this range are selected as fault components, and these fault components are applied to replace the normal components for the fault injection of the real circuit. The actual signals acquired from the real circuit and the simulated signals obtained by PSPICE are integrated as a dataset with 300 responses for each fault class, which is also divided according to the ratio of 7:3.

本数据集包含Sallen-Key带通滤波电路（Sallen-Key bandpass filter circuit）与放大板电路的脉冲响应。本次测试的激励信号为幅度5 V、频率5 kHz的10 μs脉冲信号，该信号具备丰富的频率分量。通过观测脉冲响应，在案例1中采样频率设置为5 MHz，单样本采样点数固定为1000。采用PSPICE开展电路仿真，依据故障元件参数取值范围设置电路故障。通过蒙特卡洛（Monte Carlo）分析，PSPICE为每个故障类别生成300组响应数据，以此构建案例1的数据集，并以7:3的比例随机划分为训练样本集与测试样本集。在案例2中，采样频率设置为500 kHz。依据故障元件参数取值范围，选取参数值处于该范围内的元件作为故障元件，并将这些故障元件替换正常元件，对实物电路实施故障注入。将从实物电路采集的实际信号与PSPICE仿真得到的模拟信号进行整合，构建每个故障类别含300组响应数据的数据集，同样以7:3的比例进行划分。