BiGRU-based intelligent electric meter metering module fault prediction and diagnosis method

Abstract

The invention provides a BiGRU-based intelligent electric meter fault diagnosis method. The method comprises the following steps of: respectively constructing data sets from voltage, current and electricity consumption information acquired from an intelligent electric meter, and preprocessing the data sets; inputting the data sets into a BiGRU prediction model to predict data information of the intelligent electric meter in a future period of time; and inputting the predicted data into a BiGRU diagnosis model, and obtaining the working state label of the intelligent electric meter through the analysis of the diagnosis model. According to the method, a brand-new intelligent electric meter data set construction mode is defined, the BiGRU model is applied to the field of intelligent electric meter fault prediction and diagnosis, the detection precision when a metering module fails is improved, and actual requirements can be met.

Description

technical field [0001] The invention relates to the technical field of intelligent control, in particular to a method for predicting and diagnosing a failure of a smart meter metering module based on BiGRU. Background technique [0002] With the increasing scale of the electricity consumption information collection system, the fault detection and troubleshooting of a large number of smart meters for low-voltage users requires a lot of human resources, and the manual on-site verification of the fault causes of smart meters is complex and time-sensitive; 2) In the electricity consumption information collection system, the failure of the smart meter is often discovered after a period of time, which not only brings troubles to the user, but also causes economic losses to the power supply company. An abnormal smart meter will even cause a series of security problems. The smart meter is an important terminal of the electricity consumption information collection system. The data c...

AI Technical Summary

Problems solved by technology

[0002] With the increasing scale of the electricity consumption information collection system, the fault detection and troubleshooting of a large number of smart meters for low-voltage users requires a lot of human resources, and the manual on-site verification of the fault causes of smart meters is complex and time-sensitive; 2) In the electricity consumption information collection system, the failure of the smart meter is often discovered after a period of time, which not only brings troubles to the user, but also causes economic losses to the power supply company. Abnormal smart meters can even cause a series of security issues

The smart meter is an important terminal of the electricity consumption information collection system. The data collected by it is widely used in various application services of the smart grid. The lack of reliability of the data will inevitably affect the normal operation of the entire system.

3) The failure of the metering module of the smart meter accounts for a large proportion of all the failures of the smart meter. The failure of the metering module is a long-term change process, and it is difficult to be detected without a large impact. The existing detection The accuracy of the method cannot meet the actual requirements when diagnosing the fault of the metering module

[0004] RNN is a model for modeling and analyzing time series. Its characteristic is that it considers the influence of historical state on the current situation. Compared with other intelligent algorithms, RNN has a better grasp of global data and is often used for text data classification and prediction. , but when the input data is too large, the RNN network has the problem of gradient disappearance and gradient explosion; Long Short Term Memory (LSTM) is an optimization model based on RNN network, which solves the problem by introducing gating units and cell states. The gradient problem of the traditional RNN network, but there are many parameters to be considered in the LSTM network, the model is complex, and it takes a long time to analyze the data; Cho et al. merged the gated unit of the LSTM and gave birth to the Gated Recurrent Unit (Gated RecurrentUnit , GRU) network, the performance of the GRU model for data processing has been significantly improved; inspired by the bidirectional RNN model, the GRU-based bidirectional gated recurrent unit BiGRU has taken into account the influence of the two directions of the input data context, and has greatly improved the prediction accuracy. big improvement

At present, deep learning algorithms are widely used in the field of fault diagnosis, but there are few related researches on smart meter fault diagnosis.

Images