Zero-crossing Point Detection Dataset - Distorted Sinusoidal Signals

Zero-crossing point detection is necessary to establish a consistent performance in various power system applications. Machine learning models can be used to detect zero-crossing points. A dataset is required to train and test machine learning models in order to detect the zero crossing point. Four datasets are developed for distorted sinusoidal signals. First dataset consists 4936 samples deduced from sinusoidal signals with 10%, 20%, 30%, 40% and 50% noise levels. Second dataset consists 4436 samples deduced from sinusoidal signals with 10%, 20%, 30%, 40% and 50% THD levels. Third dataset consists 3949 samples deduced from sinusoidal signals with 50% THD, and noise levels 10%, 20%, 30% and 40%. Fourth dataset consists 3949 samples deduced from sinusoidal signals with noise levels 5%, 10%, 15% and 20%.These datasets can be helpful to the researchers who are working on zero-crossing point detection problem using machine learning models. All these datasets are created based on MATLAB simulations. Each dataset consists 4 input features called slope, intercept, correlation and RMSE, and one output label with the values either 0 or 1. 0 represents non zero-crossing point class, whereas 1 represents zero-crossing point class.

零交叉点检测对于确保在各种电力系统应用中的一致性能至关重要。生成式人工智能模型可被应用于零交叉点的检测。为训练和测试机器学习模型以实现零交叉点的检测，需构建相应数据集。针对失真正弦信号，已开发出四个数据集。首个数据集由4936个样本构成，源自于含有10%、20%、30%、40%及50%噪声水平的正弦信号。第二个数据集由4436个样本构成，源自于含有10%、20%、30%、40%及50%总谐波失真（THD）水平的正弦信号。第三个数据集由3949个样本构成，源自于含有50%THD和10%、20%、30%及40%噪声水平的正弦信号。第四个数据集由3949个样本构成，源自于含有5%、10%、15%及20%噪声水平的正弦信号。这些数据集对于利用机器学习模型解决零交叉点检测问题的研究人员具有重要参考价值。所有这些数据集均基于MATLAB仿真构建。每个数据集包含4个输入特征，即斜率、截距、相关系数和均方根误差（RMSE），以及一个输出标签，其值或为0或为1。其中，0代表非零交叉点类别，而1代表零交叉点类别。