A three-phase electric energy meter automatic detection method and detection system

Abstract

A kind of three-phase electric energy meter automation detection method and detection system, using AGV car to transport three-phase electric energy meter, record conveying information data;The three-phase electric energy meter obtained by transportation is tested, and actual detection result is obtained;Using the trained performance prediction model, the performance of the three-phase electric energy meter to be detected is tested, and the performance prediction model feedback result is obtained;Relative entropy data is calculated in combination with actual detection result and performance prediction model feedback result;Based on relative entropy data and corresponding conveying information data of three-phase electric energy meter, the periodic feedback data of three-phase electric energy meter is output;Based on periodic feedback data, the built-in factory nameplate of three-phase electric energy meter is formed, registered in cloud database, and data traceability channel is formed.The present application can effectively eliminate human factors, improve test accuracy, ensure data traceability, and improve the reliability and efficiency of quality control and performance verification.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of automatic detection of electric energy meters, and particularly relates to a three-phase electric energy meter automatic detection method and a detection system. BACKGROUND

[0002] With the rapid development of smart grids, as the key metering equipment of the power system, the functions of smart electric energy meters have been expanded from traditional electric energy metering to data acquisition, remote communication, two-way interaction and load control, etc. advanced applications. In order to ensure the metering accuracy, functional reliability and long-term stability of the smart electric energy meter, the manufacturer needs to conduct strict testing before leaving the factory, covering electrical performance, communication protocol, environmental adaptability, safety protection and other dimensions. The traditional manual testing method relies on manual operation, visual inspection and scattered testing equipment, and has problems such as low efficiency, high misjudgment rate, and non-uniform testing standards. Especially in the batch production scene, manual testing is difficult to meet the requirements of high precision and high reliability, which may lead to unqualified products flowing into the market, thereby affecting the power grid company's metering settlement and user's power experience, etc.

[0003] In recent years, the introduction of automatic testing technology has provided an efficient solution for smart electric energy meter detection. Through computer control transportation, preset standard test process and automatic analysis and storage of detection data, it provides the ability to test multiple electric energy meters in parallel and automatically record test data. For example, the patent document with publication number CN106501753A completes the detection process of three-phase electric energy meters through an automatic detection line, including a feeding robot, a voltage withstand detection unit, a function detection unit, a pre / post-detection buffer unit, a multifunctional calibration unit, a laser coding, a seal, a label, etc. The detection, coding, packaging, labeling and qualified / defective separation of the electric energy meter are automatically completed. For example, the patent document with publication number CN102721940A provides a fully automatic detection system, which includes an intelligent storage interface, a feeding unit, an appearance detection, a voltage withstand test, a function error detection, a load test, a clear key, an automatic seal and a labeling unit. The system automatically adjusts the position of the electric meter through a robot and a mechanical hand and completes various detection tasks. However, in the current detection, the detection period of the electric energy meter after leaving the factory does not set a specific analysis process, and a fixed factory nameplate is used for recording. In the entire automatic detection process, transportation, fixation and disassembly and many other links often cause some damage and damage to the electric energy meter, resulting in that the subsequent detection period cannot be consistent with the laboratory test period, and lacks certain data support.

[0004] Therefore, an urgent need exists for an automated detection method for three-phase electric energy meters that combines transportation parameters, detects periodic changes, and predicts losses, accurately controls the detection process, automatically records data, and provides periodic adjustment basis, ensuring that the detection process of the electric energy meter is more stable and efficient, avoiding the influence of human factors on the test results, and achieving comprehensive and accurate performance verification and quality control. SUMMARY

[0005] To solve the technical problems of lack of flexibility and data support in the prior art without setting the detection period according to the actual situation of the electric energy meter, the present application provides a three-phase electric energy meter automated detection system and a detection method and a detection system.

[0006] To solve the above technical problems, the present application specifically adopts the following technical solutions.

[0007] The present application provides a three-phase electric energy meter automated detection method, comprising:

[0008] S1, using an AGV car to transport a three-phase electric energy meter, recording the transportation information data; testing the transported three-phase electric energy meter to obtain the actual detection result;

[0009] S2, using a trained performance prediction model to test the performance of the three-phase electric energy meter to be detected, obtaining the performance prediction model feedback result; combining the actual detection result and the performance prediction model feedback result to calculate the relative entropy data;

[0010] S3, based on the relative entropy data and the corresponding transportation information data of the three-phase electric energy meter, outputting the periodic feedback data of the three-phase electric energy meter;

[0011] S4, based on the periodic feedback data, forming the built-in factory nameplate of the three-phase electric energy meter, registering in the cloud database, and forming the data traceability channel.

[0012] Further, in S1, the transportation information data includes:

[0013] Taking a single day as a period, the power value of the AGV car in each transportation process within each period, the load value of the AGV car in each transportation process within each period, and the current transportation sequence of the AGV car in the current period;

[0014] The power value of the AGV car in each transportation process within each period is the average value of the power value at the start time and the power value at the completion of transportation.

[0015] Further, in S2, the historical transportation information historical data and the corresponding historical actual detection result are used as the input of the performance prediction model for training; the output of the performance prediction model is the performance prediction model feedback result, including the power-on condition, power consumption and electrical performance of the three-phase electric energy meter.

[0016] Further, in S2, based on the result set distribution, actual detection result distribution of actual detection result and prediction detection result distribution of performance prediction model feedback result are obtained respectively;

[0017] Based on the actual detection result distribution and the prediction detection result distribution, relative entropy data of each cycle is formed:

[0018] ;

[0019] Wherein, represents the relative entropy data between and q refers to the actual detection model, p refers to the performance prediction model of the automatic detection system, represents the actual detection result distribution of the i th three-phase electric energy meter; represents the prediction detection result distribution of the i th three-phase electric energy meter; N refers to the total number of detections of three-phase electric energy meters in each cycle.

[0020] Further, in S3, based on the relative entropy data of three-phase electric energy meters in each cycle, combined with the delivery information data corresponding to each relative entropy data, data combination between delivery information data and relative entropy data is established Wherein, refers to the power value of AGV trolley in the transportation process, refers to the load value of AGV trolley in the transportation process, refers to the current transportation of AGV trolley belonging to the i th transportation in the current cycle, refers to the corresponding relative entropy data; Based on a plurality of data combinations

[0021] , a linear fitting function between relative entropy data and is formed:

[0022]

[0023] Wherein, , , respectively refer to the regression parameters corresponding to each delivery information data, represents the error term.

[0024] ​​​Further, in S3, if the relative entropy data of the three-phase electric energy meter of different periods in the same batch appears different situations, that is, the relative entropy data calculated in the later period is in the increasing state relative to the relative entropy data calculated in the former period, it is judged that the delivery information data causes the three-phase electric energy meter to appear defects; wherein the same batch refers to the same batch of electric energy meters produced by the same equipment, and the three-phase electric energy meters of the same batch are transported in multiple periods.

[0025] If the three-phase electric energy meter of different periods in the same batch does not appear the different situations, the period feedback data of the three-phase electric energy meter to be detected is recorded as the first maintenance period time after leaving the factory.

[0026] Further, in S3, when the three-phase electric energy meter of different periods in the same batch appears different situations of relative entropy data, the period feedback data of the three-phase electric energy meter is adjusted based on the delivery information data of the three-phase electric energy meter.

[0027] Further, in S3, the period feedback data is initialized as the first maintenance period time after leaving the factory; based on the delivery information data, the three-phase electric energy meter with the same delivery information data is selected from the historical data to obtain the time of the first alarm maintenance, if the time of the first alarm maintenance is earlier than the first maintenance period time built-in the three-phase electric energy meter with defects, the difference data between the time of the first alarm maintenance and the built-in first maintenance period time is calculated, and a new data combination is formed in combination with the delivery information data.

[0028] After obtaining a plurality of new data combinations, cross-validation is used to obtain a plurality of linear fitting function models; in all the models obtained by training, the linear fitting function and parameters with the optimal mean square error loss function are selected for reservation to obtain the linear fitting function of the interpolation data with respect to the delivery information data.

[0029] Further, in S3, the period feedback data of the three-phase electric energy meter is output by using the selected linear fitting function and the delivery information data of the three-phase electric energy meter.

[0030] On the other hand, the application also proposes an automatic detection system for three-phase electric energy meter, which comprises an AGV delivery module 101, an intelligent detection module 102, a period correlation module 103 and a marking module 104, and is characterized in that:

[0031] The AGV delivery module 101 is used to build an AGV trolley working line, deliver the three-phase electric energy meter to the entrance of the automatic detection system, and upload the delivery information data at the same time;

[0032] The intelligent detection module 102 marks the corresponding three-phase electric energy meter based on the delivery information data, calls the performance prediction model feedback result of the automatic detection system, and outputs the loss function value between the performance prediction model feedback result and the actual detection result.

[0033] The cycle correlation module 103 outputs the cycle feedback data of the three-phase electric energy meter based on the loss function value between the performance prediction model feedback result and the actual detection result and the delivery information data of the corresponding three-phase electric energy meter;

[0034] The marking module 104 forms the built-in factory nameplate of the three-phase electric energy meter based on the cycle feedback data, registers in the cloud database, and forms a data traceability channel.

[0035] Compared with the prior art, the beneficial effects of the present application are:

[0036] 1、The intelligent detection module 102 of the present application realizes the comprehensive detection of the performance of the electric energy meter, avoiding the low efficiency and inaccuracy of traditional manual testing. This module can accurately test the electrical performance of the electric energy meter, reduce the occurrence of human error, and improve the repeatability and reliability of the test. The intelligent detection module not only greatly improves the test efficiency, but also can record and analyze the detection data in real time, providing detailed basis for subsequent quality management. This efficient automated detection greatly shortens the detection period and provides the necessary accuracy for mass production, ensuring that the quality of the electric energy meter meets the standards and reducing the risk of substandard products entering the market.

[0037] 2、The cycle correlation module 103 of the present application can provide dynamic adjustment basis during the detection of the smart electric energy meter by accurately monitoring the detection cycle, transportation parameters and possible losses during use of the electric energy meter. This module predicts the performance trend based on the actual use of the electric energy meter and historical detection data, thereby effectively adjusting the detection cycle to ensure that the electric energy meter always maintains high accuracy and stability during use. This periodic correlation not only optimizes the maintenance and inspection process of the electric energy meter, but also reduces the error of manual judgment, ensuring the accuracy of the electric energy meter after a long period of operation. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 is a method flow chart of an automatic detection method of a three-phase electric energy meter of the present application;

[0039] Figure 2 is a system structure diagram of an automatic detection system of a three-phase electric energy meter of the present application. DETAILED DESCRIPTION

[0040] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. The embodiments described in the present application are only a part of the embodiments of the present application, but not all the embodiments. Based on the spirit of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0041] Embodiment one

[0042] The present application provides an automatic detection method for a three-phase electric energy meter, as shown in the figure, through automatic detection and periodic adjustment, combined with transportation parameters and loss prediction, the detection process of the electric energy meter is accurately controlled. Effectively eliminate human factors, improve test accuracy, ensure data traceability, improve the reliability and efficiency of quality control and performance verification. The specific scheme is as follows. Figure 1

[0043] S1, use AGV car to transport three-phase electric energy meter, record the transportation information data; test the three-phase electric energy meter obtained by transportation, and get the actual detection result.

[0044] Specifically, one AGV car corresponds to one three-phase electric energy meter; when the AGV car works on the AGV car track, the current power state, load condition and the number of transportation of the AGV car on the same day are recorded.

[0045] Further, the transportation information data includes:

[0046] Taking a day as a period, the power value of the AGV car in each transportation process in each period, the load value of the AGV car in each transportation process in each period and the current transportation sequence of the AGV car in the current period in each period are taken as a period.

[0047] The power value of the AGV car in each transportation process in each period is the average value of the power value at the start time and the power value at the completion of transportation.

[0048] Further, the test of the three-phase electric energy meter includes connection and disconnection test, appearance test, power-on inspection test, power consumption test and electrical performance test, and the actual detection result is obtained; wherein the electrical performance test includes voltage influence test, frequency influence test, harmonic influence test and short-time overcurrent influence test.

[0049] S2, use the trained performance prediction model to test the performance of the three-phase electric energy meter to be detected, and get the performance prediction model feedback result; combine the actual detection result and the performance prediction model feedback result to calculate the loss function value.

[0050] ​Specifically, the performance prediction model can be existing technologies including but not limited to regression models, deep learning models, etc., and no further description is made herein on the structure of the performance prediction model. In order to train the performance prediction model, historical delivery information historical data and corresponding historical actual detection results are used, and the performance prediction model feedback result output by the performance prediction model includes various performances of the three-phase electric energy meter, such as power-on condition, power consumption, and electrical performance.

[0051] Further, based on the result set distribution, an actual detection result distribution of the actual detection result and a prediction detection result distribution of the performance prediction model feedback result are respectively obtained.

[0052] Based on the actual detection result distribution and the prediction detection result distribution, relative entropy data of each cycle is formed as a loss function value between the output performance prediction model feedback result and the actual detection result:

[0053] ;

[0054] Wherein, represents the relative entropy data between , q refers to the actual detection model, and p refers to the performance prediction model of the automatic detection system. represents the actual detection result distribution of the i-th three-phase electric energy meter ; represents the prediction detection result distribution of the i-th three-phase electric energy meter ; and N refers to the total number of detections of the three-phase electric energy meter in each cycle.

[0055] Specifically, since and are probability distributions including four tests, for example, the item of relative entropy , each element in and is extracted and processed, and after all elements are processed, the results of all elements are finally added to obtain the calculation result of .

[0056] Further processing, when the performance prediction model is fixed, is in a constant state, and then calculating the minimum value of the relative entropy data is simplified to calculating the maximum value of .

[0057] S3, based on the loss function value and the corresponding delivery information data of the three-phase electric energy meter, cycle feedback data of the three-phase electric energy meter is output.

[0058] Specifically, based on the loss function value of the three-phase electric energy meter per cycle, i.e., the relative entropy data, combined with the delivery information data corresponding to each relative entropy data, a data combination between the delivery information data and the relative entropy data is established , wherein, represents the power value of the AGV during the delivery process, represents the load value of the AGV during the delivery process, represents the current transportation of the AGV belonging to the th transportation of the current cycle, represents the corresponding relative entropy data;

[0059] Based on a plurality of data combinations , a linear fitting function between the relative entropy data and is formed:

[0060]

[0061] wherein, , , respectively represent the regression parameters corresponding to each delivery information data, represents the error term.

[0062] Further, the size of the fitted relative entropy data reflects the accuracy of the performance prediction model of the automatic detection system. Without changing the performance prediction model, if the relative entropy data of the three-phase electric energy meter of the same batch in different cycles appears in different situations, i.e., the relative entropy data calculated in the latter cycle is in an increasing state relative to the relative entropy data calculated in the former cycle, it is judged that the delivery information data causes the three-phase electric energy meter to appear defects. Wherein, the same batch refers to the same batch of electric energy meters produced by the same equipment, which can be regarded as completely the same electric energy meters, and the three-phase electric energy meters of the same batch are transported in multiple cycles. If the three-phase electric energy meters of the same batch in different cycles do not appear the above situation, the cycle feedback data of the three-phase electric energy meter to be detected is recorded as the first maintenance cycle time after leaving the factory.

[0063] Further, based on the delivery information data, combined with the judgment result of the appearance of defects of the three-phase electric energy meter, the cycle feedback data of the three-phase electric energy meter is adjusted.

[0064] Specifically, the cycle feedback data is initialized as the first maintenance cycle time after leaving the factory; based on the delivery information data, the three-phase electric energy meters leaving the factory under the same delivery information data are selected in the historical data to obtain the time of the first alarm maintenance, if the time of the first alarm maintenance is earlier than the first maintenance cycle time built-in the three-phase electric energy meter with defects, the difference data between the time of the first alarm maintenance and the built-in first maintenance cycle time is calculated, and a new data combination is recorded wherein, the first alarm repair time and the difference data of the built-in first repair cycle time, obtain several groups of data combination After that, in order to obtain the linear fitting function of the interpolation data about the delivery information data, a data set U is established, and the data set U is randomly divided into S mutually exclusive subsets of the same size, wherein S is a constant set by the system, S-1 is randomly selected each time as the training set of the linear fitting function, and the remaining 1 is used as the test set. After training, S-1 is randomly selected again to train the data, and the training iteration number is set to be less than S. After the training iteration number is reached, the random selection is stopped. Among all the training models obtained by training, the model and parameters with the optimal loss function are selected and reserved to obtain the linear fitting function of the interpolation data about the delivery information data. Among them, as a preferred embodiment of the present application, the mean square error loss function is used as the loss function. The cycle feedback data of the three-phase electric energy meter is output by using the selected linear fitting function and the delivery information data of the three-phase electric energy meter.

[0065] S4, based on the cycle feedback data, forming the built-in factory nameplate of the three-phase electric energy meter, at the same time, registering in the cloud database of the system end, forming the data traceability channel.

[0066] Specifically, the cycle feedback data is obtained, the actual nameplate data on the three-phase electric energy meter is not changed, the cycle detection time is adjusted, the built-in factory nameplate is generated, and the modification process is recorded as a digital path and stored on the system built-in storage chip to form data traceability.

[0067] Embodiment two

[0068] The present application provides a kind of three-phase electric energy meter automatic detection system, as shown in Figure 2 The system includes AGV conveying module 101, intelligent detection module 102, cycle correlation module 103 and marking module 104.

[0069] Specifically, AGV conveying module 101 is used to build AGV car working line, and the three-phase electric energy meter to be detected is conveyed to the entrance of automatic detection system, while uploading delivery information data.

[0070] The AGV car working line is built and includes:

[0071] The three-phase electric energy meter storage platform is externally connected with a plurality of AGV trolley tracks, the plurality of AGV trolley tracks are connected to the automatic detection system entrance, the plurality of AGV trolley tracks are internally provided with cloud chips, and the cloud chips record the current power state, the load condition and the number of deliveries of the AGV trolley in a day when the AGV trolley works on the AGV trolley track.

[0072] The AGV trolley working line further comprises a plurality of empty box bodies for adapting the plurality of AGV trolleys to carry the three-phase electric energy meters, specifically, the bottom area of the plurality of empty box bodies should be adapted to the AGV trolley hopper, and a connecting device is arranged to realize mutual connection, and one AGV trolley carries one three-phase electric energy meter.

[0073] The three-phase electric energy meter storage platform is further connected with a robot control system, an upper end of the robot control system is fixedly provided with a mechanical hand, the robot control system controls the mechanical hand to grab the three-phase electric energy meter in the three-phase electric energy meter storage platform, and the mechanical hand places the grabbed three-phase electric energy meter in the empty box body of the AGV trolley based on the recognition of the position of the AGV trolley by the recognition device.

[0074] The conveying information data comprises:

[0075] The power value of the AGV trolley in each delivery process in each cycle, the load value of the AGV trolley in each delivery process in each cycle and the current transportation order of the AGV trolley in the current cycle are taken as a cycle, and each cycle is taken as a cycle.

[0076] The power value of the AGV trolley in each delivery process in each cycle is taken as the average value of the power value at the start time and the power value at the completion of transportation.

[0077] The automatic detection system specifically comprises:

[0078] The hardware port is used for realizing the automatic detection of the three-phase electric energy meter in the hardware direction. The intelligent detection module 102 sends an instruction, the transmission layer receives the upper-level instruction, and the instruction is transmitted to each control unit, specifically including a feeding unit, a wire connecting and disconnecting unit and an appearance observation unit. The feeding unit is connected to the automatic detection system entrance to realize the automatic feeding of the three-phase electric energy meter to be detected, the wire connecting and disconnecting unit specifically realizes the wire connecting and disconnecting function of the three-phase electric energy meter, and the appearance observation unit is used for identifying and observing whether the appearance of the three-phase electric energy meter is damaged. Specifically, the intelligent detection module can be but is not limited to an embedded system, an Internet of Things and a PLC technology, as long as the functions of the intelligent detection module can be realized, and the prior art is not described in detail here.

[0079] A software port is used to implement new function checks for a three-phase electric energy meter through a hardware port, and specifically includes a power-on check unit, a power consumption test unit, and an electrical performance test unit. The power-on check unit is used to implement line power-on check for the three-phase electric energy meter, the power consumption test unit is used to test power consumption of the three-phase electric energy meter, and the electrical performance test unit is used to specifically implement voltage influence test, frequency influence test, harmonic influence test, and short-time overcurrent influence test through test programming. Specifically, the power consumption test unit can implement its functions using a digital electric energy metering chip, an analog power supply load system, etc., the voltage influence test can be implemented using an adjustable power supply, a voltage sensor, etc., the frequency influence test can be implemented using a frequency generator, an oscilloscope, etc., the harmonic influence test can be implemented using digital signal processing technology, and the short-time overcurrent influence test can be implemented using an overcurrent test device, a relay, a circuit breaker, etc. The above prior art will not be described in detail here.

[0080] The software port is externally connected with an execution layer, which is used to summarize test results, select unqualified products, and feed back test qualified rate to the performance prediction model port.

[0081] Further, the intelligent detection module 102 marks the corresponding three-phase electric energy meter based on the transportation information data, calls the performance prediction model feedback result of the automatic detection system, and outputs a loss function value between the performance prediction model feedback result and the actual detection result.

[0082] Specifically, the input of the performance prediction model includes transportation information historical data and corresponding historical detection data of the automatic detection system, and the output of the performance prediction model feedback result includes various performances of the three-phase electric energy meter, such as power-on condition, power consumption, and electrical performance. The performance prediction model can be existing technologies including but not limited to regression models, deep learning models, etc., which will not be described in detail here.

[0083] Specifically, the test results of each three-phase electric energy meter on different tests are obtained, and an actual detection result distribution is formed based on the result set distribution; the prediction results of each three-phase electric energy meter on different tests are obtained by using the performance prediction model of the automatic detection system, and a predicted detection result distribution is formed based on the result set distribution;

[0084] The relative entropy data of each cycle is formed based on the actual detection result distribution and the predicted detection result distribution, and is used as a loss function value between the output performance prediction model feedback result and the actual detection result:

[0085] ;

[0086] Wherein, represents The relative entropy data between the actual detection model q and the performance prediction model p of the automatic detection system, representing the actual detection result distribution of the i-th three-phase electric energy meter; representing the predicted detection result distribution of the i-th three-phase electric energy meter; representing the actual detection result distribution of the i-th three-phase electric energy meter; representing the predicted detection result distribution of the i-th three-phase electric energy meter;

[0087] Specifically, the softmax regression principle is used to process and , and the log calculation takes 10 as the base, i.e., lg, to form the relative entropy data of each detected three-phase electric energy meter.

[0088] Taking part of the data as an example, the actual detection result distribution is set to include: voltage influence test, frequency influence test, harmonic influence test, and short-time overcurrent influence test; the predicted detection result distribution is set to be equivalent to the actual detection result distribution, and the distributed data is processed based on the softmax regression principle to form a single data, i.e., the relative entropy data.

[0089] The calculation formula of the relative entropy data is:

[0090] ;

[0091] Specifically, since and are probability distributions including four tests, for example, the item in the relative entropy, each element in and is extracted and processed, and after all the elements are processed, the results of all the elements are added to obtain the calculation result of . After further processing, when the performance prediction model is fixed, is in a constant state, and then calculating the minimum value of the relative entropy data is simplified to calculating the maximum value of .

[0092] The intelligent detection module 102 realizes efficient detection of electric energy meters by automatically controlling the detection process. This module can monitor and record data in real time, automatically analyze test results, reduce manual intervention, and reduce the misjudgment rate. By accurately controlling each test step, it ensures the consistency and high precision of the detection, which helps to improve product quality.

[0093] Further, the cycle correlation module 103 outputs the cycle feedback data of the three-phase electric energy meter based on the loss function value between the performance prediction model feedback result and the actual detection result and the corresponding delivery information data of the three-phase electric energy meter.

[0094] Specifically, based on the loss function value of the three-phase electric energy meter per cycle, i.e. the relative entropy data, combined with the delivery information data corresponding to each relative entropy data, a data combination between the delivery information data and the relative entropy data is established wherein, denotes the power value of the AGV trolley during the delivery process, denotes the load value of the AGV trolley during the delivery process, denotes the current transportation of the AGV trolley belonging to the th transportation of the current cycle, denotes the corresponding relative entropy data;

[0095] Based on a plurality of data combinations , a linear fitting function between the relative entropy data and is formed:

[0096]

[0097] wherein, , , denote the regression parameters corresponding to each delivery information data, respectively, represents the error term.

[0098] Further, the size of the fitted relative entropy data reflects the accuracy of the performance prediction model of the automatic detection system. Without changing the performance prediction model, if the relative entropy data of the three-phase electric energy meter of the same batch in different cycles appears in different situations, i.e. the relative entropy data calculated in the later cycle is in an increasing state relative to the relative entropy data calculated in the former cycle, it is judged that the delivery information data causes the three-phase electric energy meter to appear defects. Wherein, the same batch refers to the same batch of electric energy meters produced by the same equipment, which can be regarded as completely the same electric energy meters, and the three-phase electric energy meters of the same batch are transported in multiple cycles. If the three-phase electric energy meters of the same batch in different cycles do not appear the above situation, the cycle feedback data of the three-phase electric energy meter to be detected is recorded as the first maintenance cycle time after leaving the factory.

[0099] Further, based on the delivery information data, combined with the judgment result of causing the three-phase electric energy meter to appear defects, the cycle feedback data of the three-phase electric energy meter is adjusted.

[0100] Specifically, the cycle feedback data is initialized as the first maintenance cycle time after factory; based on the delivery information data, the three-phase electric energy meter with the same delivery information data in the historical data is selected to obtain the first alarm maintenance time, and if the first alarm maintenance time is earlier than the first maintenance cycle time built-in the three-phase electric energy meter with defects, the difference data between the first alarm maintenance time and the built-in first maintenance cycle time is calculated and recorded as a new data combination wherein, the difference data between the first alarm maintenance time and the built-in first maintenance cycle time, and obtain several groups of data combinations After that, in order to obtain the linear fitting function of the interpolation data about the delivery information data, a data set U is established, and the data set U is randomly divided into S mutually exclusive subsets of the same size, wherein S is a constant set by the system, S-1 is randomly selected each time as the training set of the linear fitting function, and the remaining 1 is used as the test set. After training, S-1 is randomly selected again to train the data, and the training iteration number is set to be less than S. After the training iteration number is reached, the random selection is stopped. Among all the training models obtained by training, the model and parameters with the optimal loss function are selected and reserved to obtain the linear fitting function of the interpolation data about the delivery information data. Wherein, as a preferred embodiment of the present application, the loss function uses the mean square error loss function. The linear fitting function and the delivery information data of the three-phase electric energy meter to be detected are used to output the cycle feedback data of the three-phase electric energy meter.

[0101] The cycle correlation module 103 can dynamically adjust the detection cycle of the electric energy meter by combining the transportation parameters, the detection cycle and the loss prediction. This module analyzes the use and transportation of the electric energy meter through intelligent algorithm, accurately predicts its state change, and timely adjusts the detection strategy to avoid the deviation caused by transportation or loss, and improves the stability and accuracy of the long-term operation of the electric energy meter.

[0102] Further, the marking module 104 forms the built-in factory nameplate of the three-phase electric energy meter based on the cycle feedback data, and at the same time, registers in the cloud database of the system end to form a data traceability channel.

[0103] Specifically, the cycle feedback data is initialized as the first maintenance cycle time after factory; based on the delivery information data, the three-phase electric energy meter with the same delivery information data in the historical data is selected to obtain the first alarm maintenance time, and if the first alarm maintenance time is earlier than the first maintenance cycle time built-in the three-phase electric energy meter with defects, the difference data between the first alarm maintenance time and the built-in first maintenance cycle time is calculated and recorded as a new data combination

[0104] The marking module 104 generates the built-in factory nameplate of the three-phase electric energy meter based on the periodic feedback data, and registers in the cloud database to form a data traceability channel, which can effectively trace the detection process and adjustment history of the electric energy meter, and ensure that the periodic adjustment is accurately recorded without modifying the actual nameplate data. Through digital path recording, reliable support is provided for subsequent quality management and data verification, improving the transparency and traceability of data, and enhancing the credibility and security of the system.

[0105] The embodiments of the present application further provide a computer program product, which comprises computer program codes, and when the computer program codes run on a computer, the computer program codes make the computer implement the contents in the above-mentioned embodiments of the present application.

[0106] The embodiments of the present application further provide a computer readable storage medium, which stores computer instructions, and when the computer instructions run on a computer, the computer instructions make the computer implement the contents in the above-mentioned embodiments of the present application.

[0107] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described platforms, devices and units can refer to the corresponding processes in the foregoing embodiments, which will not be described here.

[0108] In the embodiments of the present application, the prefix words such as "first", "second" are only used to distinguish different description objects, and do not have limiting effect on the position, order, priority, quantity or content of the described objects. The use of ordinal words such as ordinal words in the embodiments of the present application does not constitute limitation on the described objects, and the description of the described objects should refer to the description in the context of claims or embodiments, and should not constitute redundant limitation because of the use of such prefix words.

[0109] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented by other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there can be another division way in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0110] In the embodiments of the present application, if there is no special description and logical conflict, the terms and / or descriptions between the embodiments are consistent and can be mutually referred, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0111] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0112] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0113] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An automated detection method for three-phase energy meters, characterized in that, include: S1, using AGV carts to transport three-phase energy meters and record the transport information data; The three-phase electricity meter that was delivered was tested, and the actual test results were obtained. The transmitted information data includes: Using a single day as a cycle, the battery level of the AGV vehicle during each transport process in each cycle, the load value of the AGV vehicle during each transport process in each cycle, and the transport sequence of the AGV vehicle in the current cycle. The power value of the AGV in each transport process within each cycle is taken as the average of the power value at the start of transport and the power value at the end of transport. S2, Use the trained performance prediction model to perform performance tests on the three-phase energy meter to be tested, and obtain the feedback results of the performance prediction model; Combine the actual test results and the feedback results of the performance prediction model to calculate the relative entropy data; Based on the distribution of the result set, the distribution of the actual detection results and the distribution of the predicted detection results from the feedback results of the performance prediction model are obtained respectively. The relative entropy data for each period is generated based on the actual detection result distribution and the predicted detection result distribution: ； in, represent The relative entropy data between them, where q refers to the actual detection model and p refers to the performance prediction model of the automated detection system. Represents the i-th three-phase energy meter The actual distribution of test results; Represents the i-th three-phase energy meter The predicted distribution of test results; N refers to the total number of three-phase energy meters tested in each cycle; S3, based on relative entropy data and the corresponding transmission information data of the three-phase energy meter, outputs the periodic feedback data of the three-phase energy meter; S4, based on periodic feedback data, forms the built-in factory nameplate of the three-phase energy meter, which is registered in the cloud database to form a data traceability channel.

2. The automated detection method for a three-phase energy meter according to claim 1, characterized in that: In S2, historical transmission information and corresponding historical actual detection results are used as inputs to the performance prediction model for training; the output of the performance prediction model is the performance prediction model feedback result, including the power supply status, power consumption, and electrical performance of the three-phase energy meter.

3. The automated detection method for a three-phase energy meter according to claim 1, characterized in that: In S3, based on the relative entropy data of the three-phase energy meters for each cycle, and combined with the corresponding transmission information data for each relative entropy data, a data combination between the transmission information data and the relative entropy data is established. ,in, This refers to the battery level of the AGV (Automated Guided Vehicle) during transportation. This refers to the load capacity of the AGV (Automated Guided Vehicle) during transportation. This refers to the AGV cart's current transport belonging to the [number]th [period] of the current cycle. Secondary transport, Refers to the corresponding relative entropy data; Based on several data combinations To form relative entropy data and Linear fitting function between: in, , , These refer to the regression parameters corresponding to each transmitted information data. This represents the error term.

4. An automated detection method for a three-phase energy meter according to claim 1 or 3, characterized in that: In S3, if three-phase energy meters from the same batch but different cycles show different relative entropy data, that is, the relative entropy data calculated in the later cycle is increasing relative to the relative entropy data calculated in the previous cycle, it is judged that the transmission information data caused the three-phase energy meter to be defective; where "same batch" refers to the same batch of energy meters produced by the same equipment, and the three-phase energy meters in the same batch are transported in multiple cycles. If the different situations do not occur in three-phase energy meters of the same batch but different cycles, the cycle feedback data of the three-phase energy meter to be tested shall be recorded as the first maintenance cycle time after leaving the factory.

5. The automated detection method for a three-phase energy meter according to claim 4, characterized in that: In S3, if three-phase energy meters of different cycles in the same batch show different relative entropy data, the periodic feedback data of the three-phase energy meters is adjusted based on the transmission information data of the three-phase energy meters.

6. The automated detection method for a three-phase energy meter according to claim 5, characterized in that: In S3, the periodic feedback data is initialized to the time of the first maintenance cycle after leaving the factory; Based on the transmission information data, select the factory-made three-phase energy meters with the same transmission information data in the historical data, and obtain the time of their first alarm maintenance. If the time of the first alarm maintenance is earlier than the built-in first maintenance cycle time of the defective three-phase energy meter, calculate the difference between the time of the first alarm maintenance and the built-in first maintenance cycle time, and combine it with the transmission information data to form a new data combination. After obtaining several new data combinations, cross-validation is used to obtain multiple linear fitting function models. Among all the models obtained through training, the linear fitting function with the optimal mean squared error loss function and parameters are selected and retained to obtain the linear fitting function of the interpolated data with respect to the transmitted information data.

7. The automated detection method for a three-phase energy meter according to claim 6, characterized in that: In S3, the selected linear fitting function and the transmission information data of the three-phase energy meter are used to output the periodic feedback data of the three-phase energy meter.

8. An automated testing system for a three-phase energy meter based on the automated testing method according to any one of claims 1-7, comprising an AGV conveying module (101), an intelligent testing module (102), a periodic correlation module (103), and a marking module (104), characterized in that: The AGV conveying module (101) is used to construct the AGV trolley work line, conveying the three-phase energy meter to the entrance of the automated detection system, and uploading the conveying information data at the same time; The intelligent detection module (102) marks the corresponding three-phase energy meters based on the transmitted information data, calls the performance prediction model feedback result of the automated detection system, and outputs the loss function value between the performance prediction model feedback result and the actual detection result; The periodic correlation module (103) outputs the periodic feedback data of the three-phase energy meter based on the loss function value between the performance prediction model feedback result and the actual detection result and the corresponding transmission information data of the three-phase energy meter. The marking module (104) forms the built-in factory nameplate of the three-phase energy meter based on the periodic feedback data, and registers it in the cloud database to form a data traceability channel.

Images