A fault detection method for smart electric meter

Abstract

The present invention relates to the field of electric meter detection, and in particular to a fault detection method for a smart electric meter, the method comprising: constructing an actual data sequence using actual data at multiple moments; inputting the actual data sequence into a preset prediction model to obtain multiple predicted data; calculating the accuracy of the predicted data, and in response to the accuracy being less than a preset accuracy threshold, updating the actual data sequence; inputting the updated actual data sequence into the preset prediction model to obtain optimal predicted data; calculating the error value between the optimal predicted data and the actual data at the current moment, and using the ratio of the error value to the actual data as the abnormality degree of the corresponding parameter; and calculating the priority level of the corresponding parameter based on the abnormality degree of the parameter, for use in detecting electric meter faults. The present invention facilitates the detection of faults in abnormal power grid states and improves fault detection efficiency.

Description

Technical Field

[0001] The present invention relates to the field of electric meter detection, and in particular to a fault detection method for a smart electric meter. Background Art

[0002] Smart meter failures are categorized as device anomalies and power usage anomalies. Power usage anomaly error codes indicate specific anomalies, with different indications for different types of anomalies. For example, E01 indicates voltage anomalies, E02 indicates current anomalies, and E03 indicates power usage anomalies. If the voltage is normal but the current is abnormally high, this may indicate a malfunction in the connected electrical device, such as a short circuit or internal damage causing excessive current flow. If the current is normal but the voltage reading is abnormal (e.g., too high or too low), this may indicate a power grid issue, such as a transformer failure, line short circuit, or unstable power supply. If the voltage and current are normal but power usage is abnormal, this may indicate a malfunction in the smart meter's detection system.

[0003] The Chinese patent application document with publication number CN113420195A discloses a method and system for determining the fault type of a smart meter. The method includes: collecting operating data of the corresponding smart meter, training a local fault detection model based on the operating data and historical data, and transmitting the trained local model parameters to the blockchain network, so that the server obtains multiple local model parameters from the blockchain network, trains a global fault detection model, and transmits the trained global model parameters to the blockchain network; sending a global model parameter request to the blockchain network, so that the blockchain network queries and obtains the latest global model parameters, and sends the parameters to the local device; updating the local fault detection model according to the received global model parameters, and performing fault detection on the corresponding smart meter according to the updated local fault detection model to obtain the fault type.

[0004] During the actual detection process, multiple faults may occur simultaneously in the power grid, and smart meters will use multiple error codes to indicate the fault conditions. However, this representation method will interfere with the technicians' choice of priority judgment direction, that is, under the conditions of multiple faults, it is impossible to quickly find the key fault point, which increases the difficulty of maintenance. Summary of the Invention

[0005] In order to solve the problem in existing detection technologies that a key fault point cannot be found when multiple faults exist, the present invention provides a fault detection method for a smart meter.

[0006] Obtain the actual data of various parameters of the smart meter at each moment, and construct an actual data sequence using the actual data at multiple moments;

[0007] Input the actual data sequence into the preset prediction model to obtain multiple prediction data;

[0008] Calculating the accuracy of the predicted data, and in response to the accuracy being less than a preset accuracy threshold, updating the actual data sequence, and inputting the updated actual data sequence into a preset prediction model to obtain optimal predicted data;

[0009] Calculate the error between the best predicted data at the current moment and the actual data, and use the ratio of the error to the actual data as the abnormality of the corresponding parameter;

[0010] The priority level of the corresponding parameter is calculated according to the abnormality level of the parameter, so as to detect the fault of the electric meter.

[0011] The abnormality degree of the corresponding parameter is obtained through the predicted value of each parameter, and then the priority of the parameter is calculated according to the abnormality degree, so that the meter can update the representation method of the abnormal condition of the power grid at the corresponding moment, which is convenient for detecting faults in the abnormal state of the power grid and improves the efficiency of fault detection.

[0012] Preferably, the method also includes: fitting multiple prediction data to obtain a prediction curve, dividing the prediction curve into a first sub-curve and a second sub-curve, with the first sub-curve being located before the second sub-curve, calculating the slope of each data point in the prediction curve, and further obtaining the average of the slopes of the data points in the first sub-curve and the second sub-curve.

[0013] By segmenting the prediction curve, we can understand the changing trend of the prediction curve and provide a theoretical basis for calculating the accuracy of the prediction results of the prediction model.

[0014] Preferably, the expression for the accuracy of the predicted data is: ;

[0015] Where, Indicates the accuracy of the forecast data at the n+mth moment, represents the mean of the slopes of the data points in the first sub-curve, Represents the mean slope of the data points in the second sub-curve.

[0016] The above formula can accurately quantify the accuracy of the predicted data, thereby improving the accuracy and accuracy of the calculation results.

[0017] Preferably, the prediction curve is divided into a first sub-curve and a second sub-curve with the midpoint of the prediction curve as a dividing point.

[0018] Preferably, the method for updating the actual data sequence is: using the moment corresponding to the last data in the actual data sequence and the current moment to construct a marked time period, and adding the actual data of each moment in the marked time period to the actual data sequence to obtain an updated actual data sequence.

[0019] By updating the actual data series, more actual data is provided to the prediction model, further improving the prediction ability of the model.

[0020] Preferably, the expression of the error value is: ;

[0021] Where, Indicates the error value between the predicted data and the actual data of the parameter at the n+mth moment, Represents the actual data of the parameters at the n+mth moment, Represents the predicted data of the parameters at the n+mth time.

[0022] Preferably, the parameters of the smart meter include current data, voltage data and power consumption data.

[0023] Preferably, the expression of priority screening degree is:

[0024] ;

[0025] ;

[0026] ;

[0027] Where, 、 、 They represent the priority of voltage data, current data and power consumption data at the time n+m respectively. 、 、 They respectively represent the abnormality degree of voltage data, current data, and power consumption data at the n+mth moment.

[0028] By calculating the priority of each parameter, the cause of abnormal power grid conditions can be quickly detected, making it easier to detect power grid faults.

[0029] Preferably, the prediction model is an ARIMA model.

[0030] Preferably, the least squares method is used to fit multiple prediction data to obtain a prediction curve.

[0031] The invention has the following technical effects:

[0032] The abnormality degree of the corresponding parameter is obtained through the predicted value of each parameter, and then the priority of the parameter is calculated according to the abnormality degree, so that the meter can update the representation method of the abnormal condition of the power grid at the corresponding moment, which is convenient for detecting faults in the abnormal state of the power grid and improves the efficiency of fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The present invention is a flowchart of a fault detection method for a smart meter. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0035] The embodiment of the present invention discloses a fault detection method for a smart meter, referring to Figure 1 , including the following steps:

[0036] S1: Obtain the actual data of various parameters of the smart meter at each moment, and use the actual data at multiple moments to construct an actual data sequence.

[0037] The parameters of smart meters include current data, voltage data, and power consumption data. During the operation of the power grid, current data, voltage data, and power consumption data can reflect the operating status of the power grid, that is, they can determine whether the power grid is abnormal. Then, the parameters of the corresponding items are used to construct the actual data sequence.

[0038] For example, get 、…、 、…、 Actual data about current at each moment ( 、…、 、…、 ),use 、…、 The current data at the moment constructs the actual data sequence about the current ( 、…、 ); Similarly, the actual data sequence about voltage is obtained ( 、…、 ) and the actual data series on electricity consumption ( 、…、 ).

[0039] S2: Input the actual data sequence into the preset prediction model to obtain multiple prediction data.

[0040] The prediction model is the ARIMA model. The actual data sequence is input into the ARIMA model to obtain multiple prediction data.

[0041] For example, the actual data sequence about the current ( 、…、 ) is input into the ARIMA model to obtain 、…、 Current prediction data at the moment ( 、…、 );

[0042] The actual data series about voltage ( 、…、 ) is input into the ARIMA model to obtain 、…、 Voltage prediction data at the moment ( 、…、 );

[0043] The actual data series about electricity consumption ( 、…、 ) is input into the ARIMA model to obtain 、…、 Electricity consumption forecast data at the moment ( 、…、 ).

[0044] It can be understood that each moment corresponds to an actual data and a predicted data.

[0045] S3: Calculate the accuracy of the predicted data. In response to the accuracy being less than a preset accuracy threshold, update the actual data sequence. Input the updated actual data sequence into a preset prediction model to obtain the optimal predicted data.

[0046] A prediction curve is obtained by fitting multiple prediction data. The prediction curve is divided into a first sub-curve and a second sub-curve with the midpoint of the prediction curve as the dividing point. The first sub-curve is located before the second sub-curve. The slope of each data point in the prediction curve is calculated, and the mean of the slopes of the data points in the first sub-curve and the second sub-curve is further obtained.

[0047] The expression for the accuracy of the predicted data is:

[0048] ;

[0049] Where, Indicates the accuracy of the forecast data at the n+mth moment, represents the mean of the slopes of the data points in the first sub-curve, Represents the mean slope of the data points in the second sub-curve.

[0050] For example, the actual data sequence about voltage ( 、…、 ) is input into the ARIMA model to obtain 、…、 Voltage prediction data at the moment ( 、…、 ); Use the least square method to fit the predicted data to obtain the prediction curve, and the data points ( 、…、 ) is used as the first sub-curve, and the data point ( 、…、 ) is used as the second sub-curve. At the n+mth moment, if the accuracy is less than the preset accuracy threshold, the accuracy threshold is set manually according to the actual situation, for example, the accuracy threshold is 0.3, indicating that the voltage data at the n+mth moment predicted by the ARIMA model is not accurate enough. Therefore, it is necessary to update the actual data sequence input into the ARIMA model. The specific updating method is to 、…、 The actual data at the moment ( 、…、 ) is added to the actual data sequence to obtain the updated actual data sequence ( 、…、 ), and then the actual data sequence ( 、…、 ) is input into the ARIMA model to obtain the optimal data at the n+mth moment.

[0051] The degree of accuracy indicates the accuracy of the prediction model's prediction results at the corresponding moment. Under normal circumstances, the changing trends of the first sub-curve and the second sub-curve should be approximately the same. Therefore, the mean slope of the first sub-curve and the mean slope of the second sub-curve should be approximately the same. Therefore, when the mean slopes of the first sub-curve and the second sub-curve differ greatly, it indicates that the accuracy of the ARIMA model's prediction results is low at this time, and the length of the actual input sequence needs to be increased to improve the accuracy of the ARIMA model's prediction results.

[0052] S4: Calculate the error between the best predicted data at the current moment and the actual data, and use the ratio of the error to the actual data as the abnormality of the corresponding parameter.

[0053] The expression of the error value is:

[0054] ;

[0055] Where, Indicates the error value between the predicted data and the actual data of the parameter at the n+mth moment, Represents the actual data of the parameters at the n+mth moment, Represents the predicted data of the parameters at the n+mth time.

[0056] The expression of abnormality degree is:

[0057] ;

[0058] Where, Indicates the abnormality of the actual data at the n+mth moment, Indicates the error value between the predicted data and the actual data of the parameter at the n+mth moment, It represents the actual data of the parameter at the n+mth moment, and norm represents the normalization function, which can be the Min-Max normalization algorithm or the Z-Score normalization algorithm.

[0059] It can be understood that at the n+mth moment, the current data, voltage data and power consumption data each correspond to a degree of abnormality. The greater the degree of abnormality, the more abnormal the actual data of the corresponding parameter is. The larger the value, the more serious the degree of abnormality.

[0060] S5: Calculate the priority level of the corresponding parameter according to the abnormality level of the parameter, so as to detect the fault of the electric meter.

[0061] The expression of priority screening is:

[0062] ;

[0063] ;

[0064] ;

[0065] Where, 、 、 They represent the priority of voltage data, current data and power consumption data at the time n+m respectively. 、 、 They respectively represent the abnormality degree of voltage data, current data, and power consumption data at the n+mth moment.

[0066] After obtaining the priority of each parameter, the parameter with the highest priority is displayed on the smart meter first, and the fault of the meter is detected according to the priority. For example, when When the value is maximum, the voltage abnormality is displayed on the smart meter. This indicates that there is a large deviation in the actual voltage data in the power grid. It can detect transformer failure, line short circuit, and unstable power supply, quickly find the fault, and reduce the time used for troubleshooting.

[0067] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A fault detection method for a smart meter, characterized in that: Including steps: Obtain the actual data of various parameters of the smart meter at each moment, and construct an actual data sequence using the actual data at multiple moments; Input the actual data sequence into the preset prediction model to obtain multiple prediction data; Fitting multiple prediction data to obtain a prediction curve, dividing the prediction curve into a first sub-curve and a second sub-curve, with the first sub-curve being located before the second sub-curve, calculating the slope of each data point in the prediction curve, and further obtaining the average of the slopes of the data points in the first sub-curve and the second sub-curve; Calculate the accuracy of the forecast data, specifically: ； Where, Indicates the accuracy of the forecast data at the n+mth moment, represents the mean of the slopes of the data points in the first sub-curve, represents the mean of the slopes of the data points in the second sub-curve; In response to the accuracy being less than a preset accuracy threshold, the actual data sequence is updated. The method for updating the actual data sequence is as follows: using the moment corresponding to the last data in the actual data sequence and the current moment to construct a marked time period, and adding the actual data of each moment in the marked time period to the actual data sequence to obtain an updated actual data sequence; Input the updated actual data sequence into the preset forecast model to obtain the best forecast data; Calculate the error between the best predicted data at the current moment and the actual data, and use the ratio of the error to the actual data as the abnormality of the corresponding parameter; The priority level of the corresponding parameter is calculated according to the abnormality level of the parameter, so as to detect the fault of the electric meter.

2. A fault detection method for a smart meter according to claim 1, characterized in that: The prediction curve is divided into a first sub-curve and a second sub-curve with the midpoint of the prediction curve as the dividing point.

3. The fault detection method for a smart meter according to claim 1, characterized in that: The expression of the error value is: ； Where, Indicates the error value between the predicted data and the actual data of the parameter at the n+mth moment, Represents the actual data of the parameters at the n+mth moment, Represents the predicted data of the parameters at the n+mth time.

4. The fault detection method for a smart meter according to claim 1, characterized in that: The parameters of smart meters include current data, voltage data and power consumption data.

5. The fault detection method for a smart meter according to claim 4, characterized in that: The expression of priority screening is: ； ； ； Where, 、 、 They represent the priority of voltage data, current data and power consumption data at the time n+m respectively. 、 、 They respectively represent the abnormality degree of voltage data, current data, and power consumption data at the n+mth moment.

6. The fault detection method for a smart meter according to claim 1, characterized in that: The forecast model is the ARIMA model.

7. The fault detection method for a smart meter according to claim 1, characterized in that: The least square method is used to fit multiple prediction data to obtain the prediction curve.

Images