Transformer monitoring data is an important basis for analyzing its operating status

but affected by external interference

short-term failure of communication ports and other factors; meanwhile

there are often missing values and noise values in the data

affecting the accuracy of status analysis. In order to improve data quality

this paper proposes a data cleaning method based on autoregressive integrated moving average model (ARIMA) and hippopotamus optimization algorithm-bidirectional long short-term memory (HO-BiLSTM). Firstly

anomaly detection is performed on the transformer monitoring data by using an improved anomaly detection method based on ARIMA. Then

HO is employed to optimize the hyperparameters of the BiLSTM model. Based on the optimal hyperparameter combination

a BiLSTM model is constructed

which is subsequently utilized to repair the abnormal data within the transformer monitoring data. Finally

the method is applied to the historical monitoring data of a transformer in the East China Power Grid. The example analysis shows that the proposed method can be adopted to effectively identify abnormal data in the monitoring data

and its performance indexes of root mean square error

average absolute error and coefficient of determination are 0.303 1

0.240 0 and 0.943 9

respectively. Compared with the traditional data cleaning method

the method performs better in the optimization of data quality and provides more accurate and reliable data supports for transformer condition analysis.