变电站升压变压器异常声压预测数据

根据搜集到的频率、距离、幅值、介质密度、声速等基础信息，计算出变电站升压变压器声压值，根据升压变压器的声压异常判别区间图表，判断出升压变压器的声压是否存在异常，从而为企业作业的安全提供了保障。1.数据搜集：搜集到升压变压器的声学振动信号数据，并将其作为特征变量。同时，收集对应的声压数据，作为目标变量。2.预处理：利用归一化公式x=（xi-min）/（max-min）；其中xi是样本字段中第i条数据，x是归一化后的值；3.模型训练：用Python语言环境运行，方法为XGBoost算法，预测变电站变压器的声压。XGBoost的目标函数由两部分组成：损失函数和正则化项。损失函数用于衡量模型预测值与实际值之间的差异，而正则化项则用于控制模型的复杂度，防止过拟合。公式如下：F(x）=求和L（yi，y^i）+求和（m(f(k))），其中L（yi，y^i）表示第i个样本的损失函数，yi是实际值，y^i是预测值；m(f(k))表示第 k 棵树的复杂度。当损失函数F（x）最小化时则停止迭代过程，在此过程中y^i预测值是通过特征函数进行求解的，具体计算方式如下：通过对于目标函数的求导，紧接着并令导数等于0，从而解出叶子节点的最优分数。通过不断地选择最优分裂点并构建树结构，XGBoost最终可以得到一棵最优的树模型。然后，将多棵树模型的预测结果进行累加，即可得到最终的预测值。4.模型迭代和更新，利用交叉验证来评估模型的稳定性和性能，其中交叉验证公式如下：m=求和（yi-y^i）/N；其中m代表交叉验证误差、N代表数据样本的数量、yi代表第 i 个样本的实际观测值、y^i代表第 i 个样本的模型预测值。若满足交叉验证的误差范围则退出迭代过程，输出模型参数值。

Based on collected basic information such as frequency, distance, amplitude, medium density and sound velocity, the sound pressure value of the step-up transformer in substations is calculated. Combined with the sound pressure anomaly discrimination interval chart of the step-up transformer, we determine whether the sound pressure of the transformer is abnormal, thereby ensuring the operational safety of enterprises.
1. Data Collection: Collect acoustic vibration signal data of the step-up transformer as feature variables, and collect corresponding sound pressure data as target variables.
2. Preprocessing: Use the normalization formula $x=(x_i - ext{min})/( ext{max} - ext{min})$, where $x_i$ is the i-th data in the sample field, and $x$ is the normalized value.
3. Model Training: Run in the Python environment, using the XGBoost algorithm to predict the sound pressure of substation transformers. The objective function of XGBoost consists of two parts: the loss function and the regularization term. The loss function measures the difference between the model's predicted values and actual values, while the regularization term controls the complexity of the model to prevent overfitting. The formula is as follows: $F(x) = sum L(y_i, hat{y}_i) + sum m(f^{(k)})$, where $L(y_i, hat{y}_i)$ represents the loss function of the i-th sample, $y_i$ is the actual value, $hat{y}_i$ is the predicted value, and $m(f^{(k)})$ represents the complexity of the k-th tree. The iteration stops when the objective function $F(x)$ is minimized. During this process, the predicted value $hat{y}_i$ is solved via the feature function. The specific calculation method is as follows: first, take the derivative of the objective function, then set the derivative to zero to solve for the optimal score of the leaf nodes. By continuously selecting the optimal split points and constructing tree structures, XGBoost finally obtains an optimal tree model. Then, the final predicted value is obtained by accumulating the prediction results of multiple tree models.
4. Model Iteration and Update: Use cross-validation to evaluate the stability and performance of the model. The cross-validation formula is as follows: $m = sum (y_i - hat{y}_i)/N$, where $m$ represents the cross-validation error, $N$ represents the number of data samples, $y_i$ represents the actual observed value of the i-th sample, and $hat{y}_i$ represents the model's predicted value of the i-th sample. The iteration exits when the cross-validation error falls within the specified range, and the model parameters are output.

搜集汇总

数据集介绍

特点

该数据集包含501条记录，每季度更新，采用CSV格式存储，记录了变电站升压变压器的频率、距离、幅值等基础信息及声压值，用于预测和判断声压异常情况。应用XGBoost算法进行模型训练和预测，旨在保障企业作业安全。