Automatic data labeling method and system for multi-source heterogeneous power data

Abstract

The invention discloses an automatic data annotation method and system for multi-source heterogeneous power data, relates to the field of power, and solves the problem that an existing automatic data annotation method is lack of accuracy during power data anomaly monitoring. Comprising the steps of S1, respectively obtaining a power monitoring time period, first to a-th power devices, a first device nameplate image and a-th device nameplate image, and creating a device nameplate image recognition model to obtain device power data, S2, performing current monitoring on a plurality of power devices according to the device power data to obtain device current monitoring data, s3, voltage monitoring is conducted on the multiple pieces of power equipment according to the equipment power data to obtain equipment voltage monitoring data, and S4, abnormity marking is conducted on the power equipment in the power monitoring time period according to the equipment current monitoring data and the equipment voltage data, and the accuracy of power abnormity marking can be improved.

Description

Technical Field

[0001] The present invention belongs to the electric power field and relates to artificial intelligence technology, and specifically to an automatic data labeling method and system for multi-source heterogeneous electric power data. Background Art

[0002] The existing automatic data annotation methods have the following defects when performing power anomaly monitoring: Existing power data annotation cannot complete the acquisition and analysis of complex power data, which makes it difficult to accurately identify complex power anomalies, resulting in a lack of accuracy in the results of power data annotation; Existing power data annotation methods usually only perform data annotation and anomaly monitoring on power data at a certain time point or within a certain period of time, which results in the lack of real-time performance of the acquired power data and the lack of application value of the automatic data annotation results. To this end, we propose an automatic data annotation method and system for multi-source heterogeneous power data. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention aims to provide an automatic data labeling method and system for multi-source heterogeneous power data. The present invention obtains equipment power data based on the acquisition of a power monitoring time period, a first power device to an ath power device, a first device nameplate image to an ath device nameplate image, and an equipment nameplate image recognition model, performs current monitoring on multiple power devices in the power monitoring time period according to the equipment power data, and obtains a current anomaly assessment coefficient corresponding to each power device to obtain equipment current monitoring data, performs voltage monitoring on multiple power devices in the power monitoring time period according to the equipment power data, and obtains a voltage anomaly assessment coefficient corresponding to each power device to obtain equipment voltage monitoring data, obtains an abnormal power monitoring coefficient corresponding to each power device by analyzing the equipment current monitoring data and the equipment voltage data, obtains an abnormal power monitoring coefficient threshold of the power equipment, performs numerical comparison on the abnormal power monitoring coefficient of the power equipment with the abnormal power monitoring coefficient threshold, and marks the power equipment in the power monitoring time period as abnormal according to the numerical comparison.

[0004] In order to achieve the above object, the present invention adopts the following technical solution: A method for automatic data annotation of multi-source heterogeneous power data comprises the following specific steps: Step S1: respectively acquiring a power monitoring time period, a first power device to an ath power device, a first device nameplate image to an ath device nameplate image, and a device nameplate image recognition model to obtain device power data; Step S2: performing current monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a current abnormality assessment coefficient corresponding to each power device to obtain device current monitoring data; Step S3: performing voltage monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a voltage anomaly assessment coefficient corresponding to each power device to obtain device voltage monitoring data; Step S4: Obtain the power equipment abnormal monitoring coefficient corresponding to each power equipment by analyzing the equipment current monitoring data and the equipment voltage data, obtain the power equipment abnormal monitoring coefficient threshold, perform numerical comparison between the power equipment abnormal monitoring coefficient and the power equipment abnormal monitoring coefficient threshold, and mark the power equipment in the power monitoring time period as abnormal according to the numerical comparison.

[0005] Furthermore, the step S1 further includes the following specific steps: Step S11: In the process of power monitoring of the target power area, the time value corresponding to the current moment is marked as the first power monitoring time point, the time point corresponding to a characteristic power monitoring duration before the first power monitoring time point is marked to obtain the second power monitoring time point, and the time period between the first power monitoring time point and the second power monitoring time point is marked as the power monitoring time period; Step S12: acquiring the power equipment existing in the target power area to obtain a plurality of power equipment, and naming the obtained plurality of power equipment as the first power equipment to the ath power equipment; Step S13: respectively acquiring the nameplate images corresponding to the first power device to the ath power device, and obtaining the nameplate images of the first device to the ath device; Step S14: Establishing a device nameplate image recognition model; Step S15: define the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model as equipment power data.

[0006] Furthermore, the step S14 further includes the following specific steps: Step S141: obtaining a plurality of power equipment nameplate images by using data crawler technology, marking the rated current value and the rated voltage value in each power equipment nameplate image by manual marking, naming the marked power equipment nameplate image as an equipment nameplate marked image, and obtaining a plurality of equipment nameplate marked images; Step S142: dividing the plurality of equipment nameplate label images into a label image training set and a label image test set according to the training test feature ratio; Step S143: creating an image initial training model through an existing convolutional neural network platform, and training the image initial training model using the labeled image training set until each device nameplate labeled image in the image initial training model has completed one training; Step S144: testing the image initial training model using the labeled image test set, and obtaining the model test success rate; Step S145: Obtain the model test calibration success rate, and perform a numerical comparison between the model test success rate and the model test calibration success rate. When the model test success rate is greater than or equal to the model test calibration success rate, the training of the image initial training model is completed, and the equipment nameplate image recognition model is obtained. When the model test success rate is less than the model test calibration success rate, continue to use the labeled image training set to train the image initial training model until the model test success rate is greater than or equal to the model test calibration success rate.

[0007] Furthermore, the step S144 further includes the following specific steps: Step S1441: respectively obtaining the rated current value and the rated voltage value in each device nameplate mark image recognized by the image initial training model, and obtaining a plurality of recognized current values ​​and a plurality of recognized voltage values; Step S1442: numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; Step S1443: if the identified current value is consistent with the rated current value and the identified voltage value is consistent with the rated voltage value, then the corresponding device nameplate mark image is marked as a first test type image; Step S1444: if the identified current value is inconsistent with the rated current value or the identified voltage value is inconsistent with the rated voltage value, the corresponding device nameplate mark image is marked as a second test type image; Step S1445: Acquire the number of first test type images in the labeled image test set to obtain the number value of the first test type images, acquire the number of images in the labeled image test set to obtain the total value of the test set images, calculate the ratio of the number value of the first test type images to the total value of the test set images, and obtain the model test success rate.

[0008] Furthermore, the step S2 further includes the following specific steps: Step S21: acquiring equipment power data, and acquiring the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S22: performing current monitoring on a first power device in a power monitoring cycle to obtain a current anomaly assessment coefficient corresponding to the first power device; Step S23: respectively monitoring the current of the second power equipment to the a-th power equipment to obtain the current abnormality assessment coefficients respectively corresponding to the second power equipment to the a-th power equipment; Step S24: defining the current anomaly assessment coefficients respectively corresponding to the first power device to the ath power device as device current monitoring data.

[0009] Furthermore, the step S22 further includes the following specific steps: Step S221: randomly selecting a number of current monitoring time points within a power monitoring cycle, and naming the selected multiple current monitoring time points as the first current monitoring time point to the cth current monitoring time point in chronological order; Step S222: respectively obtaining current values ​​of the first power device corresponding to the first current monitoring time point to the cth current monitoring time point, to obtain the first monitoring current value to the cth monitoring current value; Step S223: Calculate the variance of the first monitoring current value to the cth monitoring current value to obtain the monitoring current variance; Step S224: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated current value corresponding to the first power equipment, and comparing the first monitoring current value to the cth monitoring current value with the rated current value respectively; when the monitoring current value is less than or equal to the rated current value, marking the corresponding monitoring current value as a first type of monitoring current value; when the monitoring current value is greater than the rated current value, marking the corresponding monitoring current value as a second type of monitoring current value; Step S225: acquiring the second type current value in quantity, obtaining the abnormal current point quantity value, calculating the ratio of the abnormal current point quantity value to c, and obtaining the abnormal current quantity ratio; Step S226: averaging multiple second-type current values ​​to obtain an average of abnormal current values, calculating a difference between the average of the abnormal current values ​​and the rated current value, and taking an absolute value of the obtained difference to obtain an abnormal current value deviation; Step S227: obtaining a current abnormality assessment coefficient corresponding to the first power equipment by calculating the monitored current variance, the abnormal current quantity ratio and the abnormal current numerical deviation; The current anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows:

[0010] Among them, Ixs is the current anomaly assessment coefficient corresponding to the first power equipment, Ifc is the monitoring current variance, Icb is the abnormal current quantity ratio, and Ipc is the abnormal current numerical deviation.

[0011] Furthermore, the step S3 further includes the following specific steps: Step S31: Acquire equipment power data, and acquire the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S32: monitoring the voltage of the first power equipment in the power monitoring cycle to obtain a voltage anomaly assessment coefficient corresponding to the first power equipment; Step S33: monitoring the voltages of the second power equipment to the a-th power equipment respectively, and obtaining the voltage anomaly assessment coefficients corresponding to the second power equipment to the a-th power equipment respectively; Step S34: defining the voltage anomaly assessment coefficients respectively corresponding to the first power equipment to the ath power equipment as equipment voltage monitoring data.

[0012] Furthermore, the step S32 further includes the following specific steps: Step S321: randomly selecting a number of voltage monitoring time points within a power monitoring cycle, and naming the selected multiple voltage monitoring time points as the first voltage monitoring time point to the dth voltage monitoring time point in chronological order; Step S322: respectively obtaining voltage values ​​of the first power equipment corresponding to the first voltage monitoring time point to the dth voltage monitoring time point, and obtaining the first monitoring voltage value to the dth monitoring voltage value; Step S323: Calculate the variance of the first monitoring voltage value to the dth monitoring voltage value to obtain the monitoring voltage variance; Step S324: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated voltage value corresponding to the first power equipment, and numerically comparing the first monitoring voltage value to the dth monitoring voltage value with the rated voltage values ​​respectively; when the monitoring voltage value is less than or equal to the rated voltage value, marking the corresponding monitoring voltage value as a first type of monitoring voltage value; when the monitoring voltage value is greater than the rated voltage value, marking the corresponding monitoring voltage value as a second type of monitoring voltage value; Step S325: acquiring the quantity of the second type voltage values ​​to obtain the quantity of abnormal voltage points, calculating the ratio of the quantity of abnormal voltage points to d to obtain the abnormal voltage quantity ratio; Step S326: averaging multiple second-type voltage values ​​to obtain an average of abnormal voltage values, calculating a difference between the average of abnormal voltage values ​​and a rated voltage value, and taking an absolute value of the obtained difference to obtain an abnormal voltage value deviation; Step S327: obtaining a voltage anomaly assessment coefficient corresponding to the first power equipment by calculating the monitored voltage variance, the abnormal voltage quantity ratio, and the abnormal voltage numerical deviation; The voltage anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows:

[0013] Among them, Vxs is the voltage anomaly assessment coefficient corresponding to the first power equipment, Vfc is the monitoring voltage variance, Vcb is the abnormal voltage quantity ratio, and Vpc is the abnormal voltage numerical deviation.

[0014] Furthermore, the step S4 further includes the following specific steps: Step S41: acquiring device current monitoring data, and acquiring current abnormality assessment coefficients corresponding to the first power device to the ath power device respectively according to the device current monitoring data; Step S42: acquiring device voltage monitoring data, and acquiring voltage anomaly assessment coefficients corresponding to the first power device to the ath power device respectively according to the device voltage monitoring data; Step S43: Calculating the current abnormality assessment coefficient and the voltage abnormality assessment coefficient corresponding to each power device to obtain the power device abnormality monitoring coefficient; The abnormal monitoring coefficient of power equipment is calculated, and the specific formula is as follows:

[0015] Among them, Dyc is the power equipment abnormality monitoring coefficient, Ixs is the current abnormality assessment coefficient, and Vxs is the voltage abnormality assessment coefficient; Step S44: obtaining a threshold value of an abnormal monitoring coefficient of electric power equipment, comparing the abnormal monitoring coefficient of electric power equipment with the threshold value of an abnormal monitoring coefficient of electric power equipment, and marking an abnormality on electric power equipment in a power monitoring time period according to the numerical comparison; The step S44 further includes the following specific steps: Step S441: respectively obtaining a current anomaly assessment coefficient threshold and a voltage anomaly assessment coefficient threshold, and calculating the current anomaly assessment coefficient threshold and the voltage anomaly assessment coefficient threshold to obtain a power equipment abnormality monitoring coefficient threshold; Step S442: When the abnormal monitoring coefficient of the power equipment is greater than or equal to the abnormal monitoring coefficient threshold of the power equipment, the corresponding power equipment is marked as abnormal, and an abnormal power data warning is issued; Step S443: When the power equipment abnormality monitoring coefficient is less than the power equipment abnormality monitoring coefficient threshold, continue to monitor the power data of the corresponding power equipment.

[0016] An automatic data annotation system for multi-source heterogeneous power data. The specific working process of each module is as follows: Data acquisition module: respectively acquires the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model to obtain equipment power data; Current monitoring module: performs current monitoring on multiple power devices in the power monitoring time period according to the power data of the device, and obtains the current abnormality assessment coefficient corresponding to each power device to obtain the device current monitoring data; Voltage monitoring module: performs voltage monitoring on multiple power equipment in the power monitoring time period according to the equipment power data, and obtains the voltage anomaly assessment coefficient corresponding to each power equipment to obtain equipment voltage monitoring data; Data labeling module: by analyzing the equipment current monitoring data and equipment voltage data, the power equipment abnormal monitoring coefficient corresponding to each power equipment is obtained, the power equipment abnormal monitoring coefficient threshold is obtained, the power equipment abnormal monitoring coefficient is numerically compared with the power equipment abnormal monitoring coefficient threshold, and the power equipment in the power monitoring time period is marked as abnormal according to the numerical comparison.

[0017] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. The present invention monitors power data by respectively acquiring current data and voltage data of power equipment within a power monitoring time period, and dynamically updates the data through the power monitoring time period to ensure the real-time nature of the power data monitoring results, thereby improving the application value of power data annotation data.

[0018] 2. The present invention obtains the rated current and voltage values ​​of different power equipment by training the equipment nameplate image recognition model, which can further improve the convenience of the data acquisition process, save the cost of acquiring power labeling data, and improve the sustainability of abnormal monitoring of power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0020] Figure 1 It is a diagram of the implementation steps of the present invention; Figure 2 It is the overall system block diagram of the present invention. DETAILED DESCRIPTION

[0021] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0022] See also Figure 1 The present invention provides a technical solution: an automatic data annotation method and system for multi-source heterogeneous power data, including the following specific steps: Step S1: respectively acquiring a power monitoring time period, a first power device to an ath power device, a first device nameplate image to an ath device nameplate image, and a device nameplate image recognition model to obtain device power data; Step S11: In the process of power monitoring of the target power area, the time value corresponding to the current moment is marked as the first power monitoring time point, the time point corresponding to a characteristic power monitoring duration before the first power monitoring time point is marked to obtain the second power monitoring time point, and the time period between the first power monitoring time point and the second power monitoring time point is marked as the power monitoring time period; Step S12: acquiring the power equipment existing in the target power area to obtain a plurality of power equipment, and naming the obtained plurality of power equipment as the first power equipment to the ath power equipment; Step S13: respectively acquiring the nameplate images corresponding to the first power device to the ath power device, and obtaining the nameplate images of the first device to the ath device; Step S14: Establishing a device nameplate image recognition model; The step S14 further includes the following specific steps: Step S141: obtaining a plurality of power equipment nameplate images by using data crawler technology, marking the rated current value and the rated voltage value in each power equipment nameplate image by manual marking, naming the marked power equipment nameplate image as an equipment nameplate marked image, and obtaining a plurality of equipment nameplate marked images; Step S142: dividing the plurality of equipment nameplate label images into a label image training set and a label image test set according to the training test feature ratio; Step S143: creating an image initial training model through an existing convolutional neural network platform, and training the image initial training model using the labeled image training set until each device nameplate labeled image in the image initial training model has completed one training; Step S144: testing the image initial training model using the labeled image test set, and obtaining the model test success rate; The step S144 further includes the following specific steps: Step S1441: respectively obtaining the rated current value and the rated voltage value in each device nameplate mark image recognized by the image initial training model, and obtaining a plurality of recognized current values ​​and a plurality of recognized voltage values; Step S1442: numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; Step S1443: if the identified current value is consistent with the rated current value and the identified voltage value is consistent with the rated voltage value, then the corresponding device nameplate mark image is marked as a first test type image; Step S1444: if the identified current value is inconsistent with the rated current value or the identified voltage value is inconsistent with the rated voltage value, the corresponding device nameplate mark image is marked as a second test type image; Step S1445: acquiring the number of first test type images in the labeled image test set to obtain the number value of the first test type images, acquiring the number of images in the labeled image test set to obtain the total value of the test set images, calculating the ratio of the number value of the first test type images to the total value of the test set images, and obtaining the model test success rate; Step S145: Obtain the model test calibration success rate, and compare the model test success rate with the model test calibration success rate. When the model test success rate is greater than or equal to the model test calibration success rate, the training of the image initial training model is completed, and the equipment nameplate image recognition model is obtained. When the model test success rate is less than the model test calibration success rate, continue to use the labeled image training set to train the image initial training model until the model test success rate is greater than or equal to the model test calibration success rate. Step S15: defining the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model as equipment power data; Step S2: performing current monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a current abnormality assessment coefficient corresponding to each power device to obtain device current monitoring data; Step S21: acquiring equipment power data, and acquiring the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S22: performing current monitoring on a first power device in a power monitoring cycle to obtain a current anomaly assessment coefficient corresponding to the first power device; The step S22 further includes the following specific steps: Step S221: randomly selecting a number of current monitoring time points within a power monitoring cycle, and naming the selected multiple current monitoring time points as the first current monitoring time point to the cth current monitoring time point in chronological order; Step S222: respectively obtaining current values ​​of the first power device corresponding to the first current monitoring time point to the cth current monitoring time point, to obtain the first monitoring current value to the cth monitoring current value; Step S223: Calculate the variance of the first monitoring current value to the cth monitoring current value to obtain the monitoring current variance; Step S224: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated current value corresponding to the first power equipment, and comparing the first monitoring current value to the cth monitoring current value with the rated current value respectively; when the monitoring current value is less than or equal to the rated current value, marking the corresponding monitoring current value as a first type of monitoring current value; when the monitoring current value is greater than the rated current value, marking the corresponding monitoring current value as a second type of monitoring current value; Step S225: acquiring the second type current value in quantity, obtaining the abnormal current point quantity value, calculating the ratio of the abnormal current point quantity value to c, and obtaining the abnormal current quantity ratio; Step S226: averaging multiple second-type current values ​​to obtain an average of abnormal current values, calculating a difference between the average of the abnormal current values ​​and the rated current value, and taking an absolute value of the obtained difference to obtain an abnormal current value deviation; Step S227: obtaining a current abnormality assessment coefficient corresponding to the first power equipment by calculating the monitored current variance, the abnormal current quantity ratio and the abnormal current numerical deviation; The current anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows:

[0023] Wherein, Ixs is the current anomaly assessment coefficient corresponding to the first power equipment, Ifc is the monitoring current variance, Icb is the abnormal current quantity ratio, and Ipc is the abnormal current numerical deviation; Step S23: respectively monitoring the current of the second power equipment to the a-th power equipment to obtain the current abnormality assessment coefficients respectively corresponding to the second power equipment to the a-th power equipment; Step S24: defining the current anomaly assessment coefficients respectively corresponding to the first power device to the ath power device as device current monitoring data; Step S3: performing voltage monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a voltage anomaly assessment coefficient corresponding to each power device to obtain device voltage monitoring data; Step S31: Acquire equipment power data, and acquire the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S32: monitoring the voltage of the first power equipment in the power monitoring cycle to obtain a voltage anomaly assessment coefficient corresponding to the first power equipment; The step S32 further includes the following specific steps: Step S321: randomly selecting a number of voltage monitoring time points within a power monitoring cycle, and naming the selected multiple voltage monitoring time points as the first voltage monitoring time point to the dth voltage monitoring time point in chronological order; Step S322: respectively obtaining voltage values ​​of the first power equipment corresponding to the first voltage monitoring time point to the dth voltage monitoring time point, and obtaining the first monitoring voltage value to the dth monitoring voltage value; Step S323: Calculate the variance of the first monitoring voltage value to the dth monitoring voltage value to obtain the monitoring voltage variance; Step S324: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated voltage value corresponding to the first power equipment, and numerically comparing the first monitoring voltage value to the dth monitoring voltage value with the rated voltage values ​​respectively; when the monitoring voltage value is less than or equal to the rated voltage value, marking the corresponding monitoring voltage value as a first type of monitoring voltage value; when the monitoring voltage value is greater than the rated voltage value, marking the corresponding monitoring voltage value as a second type of monitoring voltage value; Step S325: acquiring the quantity of the second type voltage values ​​to obtain the quantity of abnormal voltage points, calculating the ratio of the quantity of abnormal voltage points to d to obtain the abnormal voltage quantity ratio; Step S326: averaging multiple second-type voltage values ​​to obtain an average of abnormal voltage values, calculating a difference between the average of abnormal voltage values ​​and a rated voltage value, and taking an absolute value of the obtained difference to obtain an abnormal voltage value deviation; Step S327: obtaining a voltage anomaly assessment coefficient corresponding to the first power equipment by calculating the monitored voltage variance, the abnormal voltage quantity ratio, and the abnormal voltage numerical deviation; The voltage anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows:

[0024] Wherein, Vxs is the voltage anomaly assessment coefficient corresponding to the first power equipment, Vfc is the monitoring voltage variance, Vcb is the abnormal voltage quantity ratio, and Vpc is the abnormal voltage numerical deviation; Step S33: monitoring the voltages of the second power equipment to the a-th power equipment respectively, and obtaining the voltage anomaly assessment coefficients corresponding to the second power equipment to the a-th power equipment respectively; Step S34: defining the voltage anomaly assessment coefficients corresponding to the first power equipment to the ath power equipment respectively as equipment voltage monitoring data; Step S4: marking abnormalities on power equipment in the power monitoring time period according to the equipment current monitoring data and the equipment voltage data respectively; Step S41: acquiring device current monitoring data, and acquiring current abnormality assessment coefficients corresponding to the first power device to the ath power device respectively according to the device current monitoring data; Step S42: acquiring device voltage monitoring data, and acquiring voltage anomaly assessment coefficients corresponding to the first power device to the ath power device respectively according to the device voltage monitoring data; Step S43: Calculating the current abnormality assessment coefficient and the voltage abnormality assessment coefficient corresponding to each power device to obtain the power device abnormality monitoring coefficient; The abnormal monitoring coefficient of power equipment is calculated, and the specific formula is as follows:

[0025] Among them, Dyc is the power equipment abnormality monitoring coefficient, Ixs is the current abnormality assessment coefficient, and Vxs is the voltage abnormality assessment coefficient; Step S44: obtaining a threshold value of an abnormal monitoring coefficient of electric power equipment, comparing the abnormal monitoring coefficient of electric power equipment with the threshold value of an abnormal monitoring coefficient of electric power equipment, and marking an abnormality on electric power equipment in a power monitoring time period according to the numerical comparison; The step S44 further includes the following specific steps: Step S441: respectively obtaining a current anomaly assessment coefficient threshold and a voltage anomaly assessment coefficient threshold, and calculating the current anomaly assessment coefficient threshold and the voltage anomaly assessment coefficient threshold to obtain a power equipment abnormality monitoring coefficient threshold; Step S442: When the abnormal monitoring coefficient of the power equipment is greater than or equal to the abnormal monitoring coefficient threshold of the power equipment, the corresponding power equipment is marked as abnormal, and an abnormal power data warning is issued; Step S443: When the power equipment abnormality monitoring coefficient is less than the power equipment abnormality monitoring coefficient threshold, continue to monitor the power data of the corresponding power equipment.

[0026] In this application, if corresponding calculation formulas appear, the above calculation formulas are all dimensionless and take their numerical calculations. The weight coefficients, proportional coefficients and other coefficients in the formulas are set to a result value obtained by quantifying each parameter. The size of the weight coefficient and the proportional coefficient can be determined as long as it does not affect the proportional relationship between the parameter and the result value.

[0027] Embodiment 2 See also Figure 2 , based on another concept of the same invention, an automatic data annotation system for multi-source heterogeneous power data is now proposed, including a data acquisition module, a current monitoring module, a voltage monitoring module, a data annotation module and a server, wherein the data acquisition module, the current monitoring module, the voltage monitoring module, the data annotation module and the server are respectively connected to the server, and the server controls the data acquisition module, the current monitoring module and the data annotation module respectively; The data acquisition module respectively acquires the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model to obtain equipment power data; In the process of power monitoring of the target power area, the time value corresponding to the current moment is marked as the first power monitoring time point, the time point corresponding to a characteristic power monitoring duration before the first power monitoring time point is marked to obtain the second power monitoring time point, and the time period between the first power monitoring time point and the second power monitoring time point is marked as the power monitoring time period; It should be noted here that: In the present application, as the time value corresponding to the current moment changes, the first power monitoring time point and the second power monitoring time point also change accordingly, thereby achieving dynamic update of the power monitoring time period; The target power area involved here is the research object corresponding to the power data in this application; Acquire the power equipment existing in the target power area to obtain a plurality of power equipment, and name the obtained plurality of power equipment as the first power equipment to the ath power equipment; It should be noted here that: In this application, a is the quantity value corresponding to the power equipment, and a is an integer greater than 0; Respectively obtain the nameplate images corresponding to the first power device to the ath power device to obtain the first device nameplate image to the ath device nameplate image; Establish equipment nameplate image recognition model; The details are as follows: A plurality of power equipment nameplate images are obtained by using data crawler technology, and the rated current value and the rated voltage value in each power equipment nameplate image are marked by manual marking, and the marked power equipment nameplate image is named as an equipment nameplate marked image, thereby obtaining a plurality of equipment nameplate marked images; Dividing a plurality of equipment nameplate labeled images into a labeled image training set and a labeled image test set according to a training test feature ratio; It should be noted here that: In the present application, the training-test feature ratio is specifically 2:8, that is, the ratio of the number of images corresponding to the labeled image training set and the labeled image test set is 2:8; Create an image initial training model through the existing convolutional neural network platform, and use the labeled image training set to train the image initial training model until each device nameplate label image in the image initial training model has completed one training; Use the labeled image test set to test the image initial training model and obtain the model test success rate; The details are as follows: Respectively obtain the rated current value and the rated voltage value in each device nameplate mark image recognized by the image initial training model to obtain multiple recognized current values ​​and multiple recognized voltage values; Compare each identified current value with the rated current value in the corresponding device nameplate mark image, and compare each identified current value with the rated current value in the corresponding device nameplate mark image; If the identified current value is consistent with the rated current value and the identified voltage value is consistent with the rated voltage value, the corresponding device nameplate mark image is marked as a first test type image; If the identified current value is inconsistent with the rated current value or the identified voltage value is inconsistent with the rated voltage value, the corresponding device nameplate mark image is marked as a second test type image; The number of first test type images in the labeled image test set is obtained to obtain the number value of the first test type images, the number of images in the labeled image test set is obtained to obtain the total value of the test set images, and the ratio of the number value of the first test type images to the total value of the test set images is calculated to obtain the model test success rate; Obtain the model test calibration success rate, and compare the model test success rate with the model test calibration success rate. When the model test success rate is greater than or equal to the model test calibration success rate, the training of the image initial training model is completed, and the equipment nameplate image recognition model is obtained. When the model test success rate is less than the model test calibration success rate, continue to use the labeled image training set to train the image initial training model until the model test success rate is greater than or equal to the model test calibration success rate. The power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model are defined as equipment power data; The data acquisition module acquires the equipment power data and transmits it to the current monitoring module, the voltage monitoring module and the data annotation module; The current monitoring module performs current monitoring on multiple power devices in the power monitoring time period according to the power data of the device, and obtains the current abnormality assessment coefficient corresponding to each power device to obtain the device current monitoring data; Acquire equipment power data, and acquire a power monitoring time period, a first power device to an ath power device, a first equipment nameplate image to an ath equipment nameplate image, and an equipment nameplate image recognition model according to the equipment power data; Performing current monitoring on a first power device in a power monitoring cycle to obtain a current anomaly assessment coefficient corresponding to the first power device; The details are as follows: In the power monitoring cycle, a number of current monitoring time points are randomly selected, and the selected multiple current monitoring time points are named as the first current monitoring time point to the cth current monitoring time point in chronological order; It should be noted here that: In the present application, c is the quantity value corresponding to the current monitoring time point, and c is an integer greater than 0; Respectively obtain current values ​​of the first power equipment corresponding to the first current monitoring time point to the cth current monitoring time point, and obtain the first monitoring current value to the cth monitoring current value; Calculating the variance of the first monitoring current value to the cth monitoring current value to obtain the monitoring current variance; Use the equipment nameplate image recognition model to identify the first equipment nameplate image, obtain the rated current value corresponding to the first power equipment, compare the first monitoring current value to the cth monitoring current value with the rated current value respectively, when the monitoring current value is less than or equal to the rated current value, mark the corresponding monitoring current value as a first type of monitoring current value, when the monitoring current value is greater than the rated current value, mark the corresponding monitoring current value as a second type of monitoring current value; And the second type of current value is quantitatively acquired to obtain the abnormal current point quantity value, and the ratio of the abnormal current point quantity value to c is calculated to obtain the abnormal current quantity ratio; Calculating the average of multiple second-type current values ​​to obtain an average of abnormal current values, calculating the difference between the average of the abnormal current values ​​and the rated current value, and taking the absolute value of the obtained difference to obtain an abnormal current value deviation; The current anomaly assessment coefficient corresponding to the first power equipment is obtained by calculating the monitored current variance, the abnormal current quantity ratio and the abnormal current numerical deviation; The current anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows: ; Wherein, Ixs is the current anomaly assessment coefficient corresponding to the first power equipment, Ifc is the monitoring current variance, Icb is the abnormal current quantity ratio, and Ipc is the abnormal current numerical deviation; Repeat the process of monitoring the current of the first power device, monitor the current of the second power device to the a-th power device respectively, and obtain the current anomaly assessment coefficients corresponding to the second power device to the a-th power device respectively; The current anomaly assessment coefficients respectively corresponding to the first power equipment to the ath power equipment are defined as equipment current monitoring data; The current monitoring module acquires the equipment current monitoring data and transmits it to the data annotation module; The voltage monitoring module performs voltage monitoring on multiple power devices in the power monitoring time period according to the power data of the equipment, and obtains the voltage anomaly assessment coefficient corresponding to each power device to obtain the equipment voltage monitoring data; Acquire equipment power data, and acquire a power monitoring time period, a first power device to an ath power device, a first equipment nameplate image to an ath equipment nameplate image, and an equipment nameplate image recognition model according to the equipment power data; Performing voltage monitoring on a first power device in a power monitoring cycle to obtain a voltage anomaly assessment coefficient corresponding to the first power device; The details are as follows: In the power monitoring cycle, a number of voltage monitoring time points are randomly selected, and the selected multiple voltage monitoring time points are named as the first voltage monitoring time point to the dth voltage monitoring time point in chronological order; It should be noted here that: In this application, d is the quantity value corresponding to the voltage monitoring time point, and d is an integer greater than 0; Respectively obtain voltage values ​​of the first power equipment corresponding to the first voltage monitoring time point to the dth voltage monitoring time point, and obtain the first monitoring voltage value to the dth monitoring voltage value; Calculating the variance of the first monitoring voltage value to the dth monitoring voltage value to obtain the monitoring voltage variance; Use the equipment nameplate image recognition model to identify the first equipment nameplate image, obtain the rated voltage value corresponding to the first power equipment, and compare the first monitoring voltage value to the dth monitoring voltage value with the rated voltage value respectively; when the monitoring voltage value is less than or equal to the rated voltage value, mark the corresponding monitoring voltage value as a first type of monitoring voltage value; when the monitoring voltage value is greater than the rated voltage value, mark the corresponding monitoring voltage value as a second type of monitoring voltage value; And the second type voltage value is quantitatively acquired to obtain the number value of abnormal voltage points, and the ratio of the number value of abnormal voltage points to d is calculated to obtain the abnormal voltage number ratio; Calculating the average of multiple second-type voltage values ​​to obtain an average of abnormal voltage values, calculating the difference between the average of abnormal voltage values ​​and the rated voltage value, and taking the absolute value of the obtained difference to obtain an abnormal voltage value deviation; The voltage anomaly assessment coefficient corresponding to the first power equipment is obtained by calculating the monitored voltage variance, the abnormal voltage quantity ratio and the abnormal voltage numerical deviation; The voltage anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows:

[0028] Wherein, Vxs is the voltage anomaly assessment coefficient corresponding to the first power equipment, Vfc is the monitoring voltage variance, Vcb is the abnormal voltage quantity ratio, and Vpc is the abnormal voltage numerical deviation; Repeat the process of monitoring the voltage of the first power equipment, monitor the voltage of the second power equipment to the a-th power equipment respectively, and obtain the voltage anomaly assessment coefficients corresponding to the second power equipment to the a-th power equipment respectively; The voltage anomaly assessment coefficients respectively corresponding to the first power equipment to the a-th power equipment are defined as equipment voltage monitoring data; The voltage monitoring module acquires the equipment voltage monitoring data and transmits it to the data annotation module; The data annotation module marks abnormalities of power equipment in the power monitoring time period according to the equipment current monitoring data and the equipment voltage data; Acquire device current monitoring data, and acquire current anomaly assessment coefficients corresponding to the first power device to the ath power device respectively according to the device current monitoring data; Acquire device voltage monitoring data, and acquire voltage anomaly assessment coefficients corresponding to the first power device to the ath power device respectively according to the device voltage monitoring data; The abnormal current assessment coefficient and the abnormal voltage assessment coefficient corresponding to each electric power equipment are calculated to obtain the abnormal monitoring coefficient of the electric power equipment; The abnormal monitoring coefficient of power equipment is calculated, and the specific formula is as follows: ; Among them, Dyc is the power equipment abnormality monitoring coefficient, Ixs is the current abnormality assessment coefficient, and Vxs is the voltage abnormality assessment coefficient; Obtaining a threshold value of an abnormal monitoring coefficient of electric power equipment, comparing the abnormal monitoring coefficient of electric power equipment with the threshold value of an abnormal monitoring coefficient of electric power equipment, and marking an abnormality on electric power equipment in a power monitoring time period according to the numerical comparison; The details are as follows: Obtaining a current anomaly assessment coefficient threshold and a voltage anomaly assessment coefficient threshold respectively, and calculating the current anomaly assessment coefficient threshold and the voltage anomaly assessment coefficient threshold to obtain a power equipment anomaly monitoring coefficient threshold; It should be noted here that: The current anomaly assessment coefficient threshold and voltage anomaly assessment coefficient threshold involved here are respectively the maximum current anomaly assessment coefficient and the maximum voltage anomaly assessment coefficient corresponding to the power equipment under normal working conditions; The threshold value of the abnormal monitoring coefficient of power equipment is calculated, and the specific formula is as follows: ; Wherein, Dycy is the power equipment abnormality monitoring coefficient threshold, Ixsy is the current abnormality assessment coefficient threshold, and Vxsy is the voltage abnormality assessment coefficient threshold; When the abnormal monitoring coefficient of the power equipment is greater than or equal to the abnormal monitoring coefficient threshold of the power equipment, the corresponding power equipment is marked as abnormal and an abnormal power data warning is issued; When the power equipment abnormality monitoring coefficient is less than the power equipment abnormality monitoring coefficient threshold, the corresponding power equipment power data monitoring continues.

[0029] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well.

Claims

1. An automatic data annotation method for multi-source heterogeneous power data, characterized in that: include: Step S1: respectively obtaining a power monitoring time period, a first power device to an ath power device, and a first device nameplate image to an ath device nameplate image, and creating a device nameplate image recognition model to obtain device power data; Step S2: performing current monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a current abnormality assessment coefficient corresponding to each power device to obtain device current monitoring data; Step S3: performing voltage monitoring on multiple power devices in a power monitoring time period according to the device power data, and obtaining a voltage anomaly assessment coefficient corresponding to each power device to obtain device voltage monitoring data; Step S4: Obtain the power equipment abnormal monitoring coefficient corresponding to each power equipment by analyzing the equipment current monitoring data and the equipment voltage data, obtain the power equipment abnormal monitoring coefficient threshold, perform numerical comparison between the power equipment abnormal monitoring coefficient and the power equipment abnormal monitoring coefficient threshold, and mark the power equipment in the power monitoring time period as abnormal according to the numerical comparison.

2. The automatic data annotation method for multi-source heterogeneous power data according to claim 1 is characterized in that: The step S1 further includes the following specific steps: Step S11: In the process of performing power monitoring on the target power area, the time value corresponding to the current moment is marked as the first power monitoring time point, the time point corresponding to a characteristic power monitoring duration before the first power monitoring time point is marked as the second power monitoring time point, and the time period between the first power monitoring time point and the second power monitoring time point is marked as the power monitoring time period; Step S12: acquiring the power equipment existing in the target power area to obtain a plurality of power equipment, and naming the obtained plurality of power equipment as the first power equipment to the ath power equipment; Step S13: respectively acquiring the nameplate images corresponding to the first power device to the ath power device, and obtaining the nameplate images of the first device to the ath device; Step S14: Establishing a device nameplate image recognition model; Step S15: define the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model as equipment power data.

3. The automatic data annotation method for multi-source heterogeneous power data according to claim 2 is characterized in that: The step S14 further includes the following specific steps: Step S141: acquiring a plurality of power equipment nameplate images, marking the rated current value and the rated voltage value in each power equipment nameplate image by manual marking, naming the marked power equipment nameplate image as a device nameplate marked image, and obtaining a plurality of device nameplate marked images; Step S142: dividing the plurality of equipment nameplate label images into a label image training set and a label image test set according to the training test feature ratio; Step S143: creating an image initial training model through an existing convolutional neural network platform, and training the image initial training model using the labeled image training set until each device nameplate labeled image in the image initial training model has completed one training; Step S144: testing the image initial training model using the labeled image test set, and obtaining the model test success rate; Step S145: Obtain the model test calibration success rate, and perform a numerical comparison between the model test success rate and the model test calibration success rate. When the model test success rate is greater than or equal to the model test calibration success rate, the training of the image initial training model is completed, and the equipment nameplate image recognition model is obtained. When the model test success rate is less than the model test calibration success rate, continue to use the labeled image training set to train the image initial training model until the model test success rate is greater than or equal to the model test calibration success rate.

4. The automatic data annotation method for multi-source heterogeneous power data according to claim 3 is characterized in that: The step S144 further includes the following specific steps: Step S1441: respectively obtaining the rated current value and the rated voltage value in each device nameplate mark image recognized by the image initial training model, and obtaining a plurality of recognized current values ​​and a plurality of recognized voltage values; Step S1442: numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; numerically comparing each identified current value with the rated current value in the corresponding device nameplate mark image; Step S1443: if the identified current value is consistent with the rated current value and the identified voltage value is consistent with the rated voltage value, then the corresponding device nameplate mark image is marked as a first test type image; Step S1444: if the identified current value is inconsistent with the rated current value or the identified voltage value is inconsistent with the rated voltage value, the corresponding device nameplate mark image is marked as a second test type image; Step S1445: Acquire the number of first test type images in the labeled image test set to obtain the number value of the first test type images, acquire the number of images in the labeled image test set to obtain the total value of the test set images, calculate the ratio of the number value of the first test type images to the total value of the test set images, and obtain the model test success rate.

5. The automatic data annotation method for multi-source heterogeneous power data according to claim 1 is characterized in that: The step S2 further includes the following specific steps: Step S21: acquiring equipment power data, and acquiring the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S22: performing current monitoring on a first power device in a power monitoring cycle to obtain a current anomaly assessment coefficient corresponding to the first power device; Step S23: respectively monitoring the current of the second power equipment to the a-th power equipment to obtain the current abnormality assessment coefficients respectively corresponding to the second power equipment to the a-th power equipment; Step S24: defining the current anomaly assessment coefficients respectively corresponding to the first power device to the ath power device as device current monitoring data.

6. The automatic data annotation method for multi-source heterogeneous power data according to claim 5 is characterized in that: The step S22 further includes the following specific steps: Step S221: randomly selecting a number of current monitoring time points within a power monitoring cycle, and naming the selected multiple current monitoring time points as the first current monitoring time point to the cth current monitoring time point in chronological order; Step S222: respectively obtaining current values ​​of the first power device corresponding to the first current monitoring time point to the cth current monitoring time point, to obtain the first monitoring current value to the cth monitoring current value; Step S223: Calculate the variance of the first monitoring current value to the cth monitoring current value to obtain the monitoring current variance; Step S224: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated current value corresponding to the first power equipment, and comparing the first monitoring current value to the cth monitoring current value with the rated current value respectively; when the monitoring current value is less than or equal to the rated current value, marking the corresponding monitoring current value as a first type of monitoring current value; when the monitoring current value is greater than the rated current value, marking the corresponding monitoring current value as a second type of monitoring current value; Step S225: acquiring the second type current value in quantity, obtaining the abnormal current point quantity value, calculating the ratio of the abnormal current point quantity value to c, and obtaining the abnormal current quantity ratio; Step S226: averaging multiple second-type current values ​​to obtain an average of abnormal current values, calculating a difference between the average of the abnormal current values ​​and the rated current value, and taking an absolute value of the obtained difference to obtain an abnormal current value deviation; Step S227: obtaining a current abnormality assessment coefficient corresponding to the first power equipment by calculating the monitored current variance, the abnormal current quantity ratio and the abnormal current numerical deviation; The current anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows: ； Among them, Ixs is the current anomaly assessment coefficient corresponding to the first power equipment, Ifc is the monitoring current variance, Icb is the abnormal current quantity ratio, and Ipc is the abnormal current numerical deviation.

7. The automatic data annotation method for multi-source heterogeneous power data according to claim 1 is characterized in that: The step S3 further includes the following specific steps: Step S31: Acquire equipment power data, and acquire the power monitoring time period, the first power equipment to the ath power equipment, the first equipment nameplate image to the ath equipment nameplate image, and the equipment nameplate image recognition model according to the equipment power data; Step S32: monitoring the voltage of the first power equipment in the power monitoring cycle to obtain a voltage anomaly assessment coefficient corresponding to the first power equipment; Step S33: monitoring the voltages of the second power equipment to the a-th power equipment respectively, and obtaining the voltage anomaly assessment coefficients corresponding to the second power equipment to the a-th power equipment respectively; Step S34: defining the voltage anomaly assessment coefficients respectively corresponding to the first power equipment to the ath power equipment as equipment voltage monitoring data.

8. The automatic data annotation method for multi-source heterogeneous power data according to claim 7 is characterized in that: The step S32 further includes the following specific steps: Step S321: randomly selecting a number of voltage monitoring time points within a power monitoring cycle, and naming the selected multiple voltage monitoring time points as the first voltage monitoring time point to the dth voltage monitoring time point in chronological order; Step S322: respectively obtaining voltage values ​​of the first power equipment corresponding to the first voltage monitoring time point to the dth voltage monitoring time point, and obtaining the first monitoring voltage value to the dth monitoring voltage value; Step S323: Calculate the variance of the first monitoring voltage value to the dth monitoring voltage value to obtain the monitoring voltage variance; Step S324: using the equipment nameplate image recognition model to recognize the first equipment nameplate image, obtaining the rated voltage value corresponding to the first power equipment, and numerically comparing the first monitoring voltage value to the dth monitoring voltage value with the rated voltage values ​​respectively; when the monitoring voltage value is less than or equal to the rated voltage value, marking the corresponding monitoring voltage value as a first type of monitoring voltage value; when the monitoring voltage value is greater than the rated voltage value, marking the corresponding monitoring voltage value as a second type of monitoring voltage value; Step S325: acquiring the quantity of the second type voltage values ​​to obtain the quantity of abnormal voltage points, calculating the ratio of the quantity of abnormal voltage points to d to obtain the abnormal voltage quantity ratio; Step S326: averaging multiple second-type voltage values ​​to obtain an average of abnormal voltage values, calculating a difference between the average of abnormal voltage values ​​and a rated voltage value, and taking an absolute value of the obtained difference to obtain an abnormal voltage value deviation; Step S327: obtaining a voltage anomaly assessment coefficient corresponding to the first power equipment by calculating the monitored voltage variance, the abnormal voltage quantity ratio, and the abnormal voltage numerical deviation; The voltage anomaly assessment coefficient corresponding to the first power equipment is calculated, and the specific formula is as follows: ； Among them, Vxs is the voltage anomaly assessment coefficient corresponding to the first power equipment, Vfc is the monitoring voltage variance, Vcb is the abnormal voltage quantity ratio, and Vpc is the abnormal voltage numerical deviation.

9. The automatic data annotation method for multi-source heterogeneous power data according to claim 1 is characterized in that: The step S4 further includes the following specific steps: Step S41: acquiring device current monitoring data, and acquiring current abnormality assessment coefficients corresponding to the first power device to the ath power device respectively according to the device current monitoring data; Step S42: acquiring device voltage monitoring data, and acquiring voltage anomaly assessment coefficients corresponding to the first power device to the ath power device respectively according to the device voltage monitoring data; Step S43: Calculating the current abnormality assessment coefficient and the voltage abnormality assessment coefficient corresponding to each power device to obtain the power device abnormality monitoring coefficient; The abnormal monitoring coefficient of power equipment is calculated, and the specific formula is as follows: ； Among them, Dyc is the power equipment abnormality monitoring coefficient, Ixs is the current abnormality assessment coefficient, and Vxs is the voltage abnormality assessment coefficient; Step S44: obtaining a threshold value of an abnormal monitoring coefficient of electric power equipment, comparing the abnormal monitoring coefficient of electric power equipment with the threshold value of an abnormal monitoring coefficient of electric power equipment, and marking an abnormality on electric power equipment in a power monitoring time period according to the numerical comparison; The step S44 further includes the following specific steps: Step S441: respectively obtaining a current anomaly assessment coefficient threshold and a voltage anomaly assessment coefficient threshold, and calculating the current anomaly assessment coefficient threshold and the voltage anomaly assessment coefficient threshold to obtain a power equipment abnormality monitoring coefficient threshold; Step S442: When the abnormal monitoring coefficient of the power equipment is greater than or equal to the abnormal monitoring coefficient threshold of the power equipment, the corresponding power equipment is marked as abnormal, and an abnormal power data warning is issued; Step S443: When the power equipment abnormality monitoring coefficient is less than the power equipment abnormality monitoring coefficient threshold, continue to monitor the power data of the corresponding power equipment.

10. An automatic data annotation system for multi-source heterogeneous power data, applicable to an automatic data annotation method for multi-source heterogeneous power data as claimed in any one of claims 1 to 9, characterized in that: The specific working process of each module of the automatic data annotation system is as follows: Data acquisition module: respectively acquires the power monitoring time period, the first power equipment to the ath power equipment, and the first equipment nameplate image to the ath equipment nameplate image, and creates an equipment nameplate image recognition model to obtain equipment power data; Current monitoring module: performs current monitoring on multiple power devices in the power monitoring time period according to the power data of the device, and obtains the current abnormality assessment coefficient corresponding to each power device to obtain the device current monitoring data; Voltage monitoring module: performs voltage monitoring on multiple power equipment in the power monitoring time period according to the equipment power data, and obtains the voltage anomaly assessment coefficient corresponding to each power equipment to obtain equipment voltage monitoring data; Data labeling module: by analyzing the equipment current monitoring data and equipment voltage data, the power equipment abnormal monitoring coefficient corresponding to each power equipment is obtained, the power equipment abnormal monitoring coefficient threshold is obtained, the power equipment abnormal monitoring coefficient is numerically compared with the power equipment abnormal monitoring coefficient threshold, and the power equipment in the power monitoring time period is marked as abnormal according to the numerical comparison.

Images