A Convolutional Neural Network-based Method for Tracing Anomalies in Power Grid Project Cost Data

The research on the power grid project cost data anomaly traceability method based on convolutional neural network is proposed. After comprehensively analyzing the composition of power grid project cost data from three perspectives of basic data, engineering quantity data and cost data, the convolutional neural network algorithm is introduced, and the mechanism of staggered fusion grouping is used to realize full coverage traceability of data anomalies. In the test results, F1_ score of the test results under the abnormal traceability method of the design method engineering cost data not only shows high stability, but also always above 0.90. The maximum F1_score reaches 0.99, the minimum F1_score also reached 0.91, and traceability has a good full coverage effect.