A three-phase electric energy meter system with multiple communication modes and remote upgrade

Abstract

The present invention relates to the technical field of three-phase electric energy meters, and specifically to a three-phase electric energy meter system that can be remotely upgraded using multiple communication modes. The system comprises a background parallel management unit, an electric energy information collection unit, a background remote upgrade unit, an electric energy data processing unit, a data training unit, a data self-test unit, and a load prediction unit. When a cloud upgrade instruction is generated in the background parallel management unit, the cloud upgrade information instruction is transmitted to the electric energy information collection unit. After receiving the upgrade instruction, the electric energy information collection unit collects upgrade status information and historical data packets of each electric energy meter. The electric energy data processing unit cleans the historical data packets, specifically using the average value of the ratio of the power range value to the rated power value and the connection deviation rate ratio as the prediction weighted item for a model construction, so as to meet the regulation and prediction requirements of multiple risks in a future environment and improve the service life of the electric energy meter at that location.

Description

Technical Field

[0001] The present invention relates to the technical field of three-phase electric energy meters, and in particular to a three-phase electric energy meter system capable of remote upgrading in multiple communication modes. Background Art

[0002] With the continued advancement of smart grid construction, three-phase electricity meters, as key data collection and metering equipment in power systems, are becoming increasingly important in terms of advanced functionality and system stability. To meet the ever-increasing demands of power management, remote upgrade capabilities have become an indispensable feature of three-phase electricity meter systems. However, existing remote upgrade methods for three-phase electricity meters currently on the market lack significant environmental adaptability. In areas with well-developed communication infrastructure, such as cities, while offering high-speed and stable network conditions, existing upgrade solutions often rely heavily on specific communication networks, such as 4G or wired networks. The upgrade process can be hindered by brief network outages, signal interference, or even insufficient bandwidth due to regional network congestion. This can not only result in upgrade failures, but also require significant time and manpower for secondary upgrades, significantly impacting power operation and maintenance efficiency.

[0003] Existing three-phase energy meter system upgrade methods lack sufficient flexibility when dealing with energy meter terminals with different hardware configurations. Energy meters produced by different manufacturers have different hardware performance, storage capacity, and computing power. Some upgrade solutions fail to fully account for these differences. As a result, on some energy meters with limited hardware resources, the upgrade process is interrupted due to insufficient resources, or new functions fail to operate normally after the upgrade, seriously affecting the user experience and the overall intelligent level of the power system. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a three-phase electric energy meter system that can be remotely upgraded in multiple communication modes, which can effectively solve the problems in the prior art.

[0005] Technical Solution

[0006] The present invention provides a three-phase electric energy meter system that can be remotely upgraded in multiple communication modes, including: a background parallel management unit, an electric energy information collection unit, a background remote upgrade unit, an electric energy data processing unit, a data training unit, a data self-test unit and a load prediction unit. When a cloud upgrade instruction is generated in the background parallel management unit, the cloud upgrade information instruction is transmitted to the electric energy information collection unit. After receiving the upgrade instruction, the electric energy information collection unit collects the upgrade status information and historical data packets of each electric energy meter. The electric energy data processing unit cleans the historical data packets therein and performs a simulation on the processed historical data packets through the data training unit. Model construction, predicting data through the load prediction unit to obtain data group A of the electricity meter, and then performing secondary training model construction on data group A for the upgraded model to obtain data group B. The data self-checking unit judges the future electricity load based on data group B, and generates the future electricity data time series Xf and its power wave peak value. The load prediction unit compares the future electricity data time series Xf and the power wave peak value with the maximum rated power of the electricity meter. The data self-checking unit judges the working status of the electricity meter in the future time period and feeds back to the background remote upgrade unit. The background remote upgrade unit issues specific remote upgrade tasks according to the future working status of each electricity meter.

[0007] Furthermore, the method for obtaining the upgrade status information of each electricity meter is: based on the current version information of the electricity meter and the internal power supply information of the electricity meter, it is packaged into local status information, and the above local information status is fed back to the background remote upgrade unit, and an upgrade plan is generated in the background remote upgrade unit. The historical data packet contains data information of the power system, including power load data and time information data. The upgrade status information includes the version number of the current electricity meter, the running time of the version, the specific location and the internal power supply information of the electricity meter.

[0008] Furthermore, the cleaning processing step includes: making an original judgment on the historical data packets, screening and removing the electric energy data and duplicate data that have not been successfully collected, and then using the data concentrator to unify the data format, and accurate to three decimal places, and package it into a data packet.

[0009] Furthermore, a model construction includes a data preparation unit and a model training unit. The data preparation unit converts data packets into data as input to the LSTM model, determines the number of LSTM layers and the number of neurons in each LSTM layer, adds a fully connected layer to map the features output by the LSTM layer to the final prediction dimension, and the fully connected layer linearly transforms the feature vector output by the LSTM layer through a weight matrix to obtain a single prediction result, and uses a sliding window to generate time series samples. The LSTM model uses its unique gating mechanism, including input gate, forget gate and output gate, to effectively remember data information over a long period of time in the past. The LSTM model can accurately predict future load changes based on peak and off-peak electricity consumption data from days or even weeks ago. The data training unit is responsible for training the LSTM model for load forecasting, including model construction, training and evaluation: the LSTM model is constructed using the hidden state of the time step and the input of the current time step, and obtains the weight matrices of the forget gate, input gate and candidate state, and uses an activation function to activate the weight matrices of the forget gate, input gate and candidate state.

[0010] Furthermore, the method for the load prediction unit to perform data prediction is: judging the risk operation data and status performance within the primary model construction data, the operation risk data includes the power value and connection deviation rate measured at a certain moment that far exceeds the normal three-phase power range, and the status performance data includes the status performance value and the operating loss value, setting the prediction rules for the secondary model construction, and using the ratio of the value exceeding the power range to the rated power value and the average value of the connection deviation rate ratio as the prediction weighted item for the primary model construction to form data group A.

[0011] Furthermore, the secondary model construction includes a data input unit and a secondary model training unit. By inputting data group A into the LSTM model, it is possible to judge and predict future situations and obtain a secondary prediction result, which is marked as data group B. Data group B is expanded to generate the future electric energy data time series Xf and its power peak value.

[0012] Furthermore, the data self-check unit can judge the working status of the electric energy meter in the future time period, and its basis is that if the peak value of the power wave passing through in the future time period is greater than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter cannot meet the maximum passing electric energy value in the future time period, and generates an optimization upgrade signal, which is fed back to the background remote upgrade unit. If the peak value of the power wave passing through in the future time period is less than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter does not need remote upgrade.

[0013] Furthermore, the remote upgrade task delivery method includes but is not limited to software updates initiated by a remote master station, OTA updates based on a communication network, and local manual updates.

[0014] Beneficial effects

[0015] In the present invention, remote upgrade task information issued by the remote upgrade background is received; the electric energy information of the target electric energy meter is obtained; based on the electric energy meter information, the target electric energy meter is determined to determine whether it meets the preset remote upgrade task requirements, and then the upgrade plan is returned to the remote upgrade background, wherein the data is weighted according to the risk operation data and the status performance, and the upgrade plan is calculated based on the electric energy meter information and the electric energy meter battery information, and the remote upgrade task issued by the remote upgrade background according to the upgrade plan is received; the electronic control unit is upgraded according to the remote upgrade task, so that the upgrade plan of the remote upgrade task can be determined more accurately, and the flexibility of the remote upgrade task execution scenario can be improved. Compared with manually calculating the upgrade plan of the remote upgrade task, the computing efficiency can be greatly improved and the cost can be reduced. At the same time, the method makes predictions based on the original data of the electric energy meter under different usage scenarios, and the background remote upgrade unit issues specific remote upgrade tasks based on the future working status of each electric energy meter. It can be speculated based on the current usage environment and is suitable for the normal use of electric energy meters in various environments.

[0016] Compared with the existing upgrade method, the present invention uses risk operation data and status performance as enhanced items. Specifically, the average value of the ratio of the value exceeding the power range to the rated power value and the connection deviation rate ratio is used as the prediction weighted item for a single model construction, which meets the requirements for regulatory prediction of multiple risks in future environments and improves the service life of the electricity meter at that location.

[0017] In the system, the LSTM model, through its unique gating mechanism, including input gates, forget gates, and output gates, can effectively memorize data information from long periods of past use. The LSTM model can accurately predict future load changes based on peak and trough electricity consumption data from days or even weeks prior. This is one of the advantages of this system compared to existing neural networks. Furthermore, the LSTM model can automatically learn these non-stationary characteristics in the data and dynamically adjust the weighting of data at different times. During the hot summer months, high-power electrical appliances are frequently used, causing significant fluctuations in power load data. The LSTM model can analyze and predict current data more effectively based on the characteristics of summer periods in historical data, accurately capturing patterns in load fluctuations. This provides a reliable basis for power dispatch and management, and a predictive basis for future load forecasting, satisfying the functional requirements of each electricity meter in different environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0019] Figure 1 It is a flowchart of the present invention;

[0020] Figure 2 A diagram of the steps for constructing the primary and secondary training models of the present invention. DETAILED DESCRIPTION

[0021] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] The present invention will be further described below with reference to the embodiments.

[0023] Example 1: A three-phase electric energy meter system with multiple communication modes and remote upgrade, see the attached Figure 1 -Attached Figure 2 A three-phase electric energy meter system with multiple communication modes and remote upgradeability includes: a background parallel management unit, an electric energy information acquisition unit, a background remote upgrade unit, an electric energy data processing unit, a data training unit, a data self-checking unit, and a load prediction unit;

[0024] When the cloud upgrade instruction is generated in the background parallel management unit, the cloud upgrade information instruction is transmitted to the power information collection unit. After receiving the upgrade instruction, the power information collection unit collects the upgrade status information and historical data packets of each power meter.

[0025] The power data processing unit cleans the historical data packets.

[0026] Through the data training unit, a model is constructed for the processed historical data packets. A model construction includes a data preparation unit and a model training unit. The data preparation unit converts the data packet into data as the input of the LSTM model, determines the number of LSTM layers and the number of neurons in each LSTM layer, and adds a fully connected layer to map the features output by the LSTM layer to the final prediction dimension. The fully connected layer linearly transforms the feature vector output by the LSTM layer through the weight matrix to obtain a prediction result. By training on a large number of diverse data sets, the LSTM model can learn the commonalities and characteristics of data in different scenarios and has good generalization ability.

[0027] The upgrade plan is calculated based on the electricity meter information and the electricity meter battery information, and receives the remote upgrade task issued by the remote upgrade background according to the upgrade plan; upgrading the electronic control unit according to the remote upgrade task can more accurately determine the upgrade plan of the remote upgrade task, improve the flexibility of the remote upgrade task execution scenario, and compared with manually calculating the upgrade plan of the remote upgrade task, it can greatly improve computing efficiency and reduce costs.

[0028] The LSTM model can also memorize peak and valley data from days or even weeks prior, allowing accurate predictions of future load changes. This is one of the advantages of this system compared to existing neural networks. Furthermore, the LSTM model can automatically learn these non-stationary characteristics in the data and dynamically adjust the weighting of data at different times. During the hot summer months, high-power appliances like air conditioners are frequently used, causing significant fluctuations in power load data. The LSTM model can analyze and predict current data more effectively based on the characteristics of summer periods in historical data, accurately capturing patterns in load fluctuations. This provides a reliable basis for power dispatch and management, and a predictive basis for future load forecasting, ensuring that each energy meter can function in different environments.

[0029] The load prediction unit predicts data. The method for the load prediction unit to predict data is as follows: determine the risk operation data and status performance within the primary model construction data. The operation risk data includes the power value and connection deviation rate measured at a certain moment that far exceeds the normal three-phase power range. The status performance data includes the status performance value and the operating loss value. Set the prediction rules for secondary model construction, process the rated optimization value YH and the connection deviation rate CX, and use the formula: Obtain the adjustment coefficient TZ, where a1 and a2 are preset proportional coefficients. Obtain the rated power and the excess power value of the electric energy meter, perform difference processing on them, obtain the actual power deviation, and perform ratio processing on the actual power deviation with the rated power of the electric energy meter to obtain the rated optimization value, which is marked as YH. The average value of the ratio of the excess power range value to the rated power value and the connection deviation rate ratio is used as the prediction weighted item for the primary model construction to form data group A.

[0030] Then, a secondary training model is constructed for the upgrade model of data group A to obtain data group B. The secondary model construction includes a data input unit and a secondary model training unit. Data group A is input into the LSTM model to judge and predict future situations, and a secondary prediction result is obtained, which is marked as data group B. Data group B is expanded to generate the future electric energy data time series Xf and its power peak value; the remote upgrade task information issued by the remote upgrade background is received;

[0031] Obtain the electric energy information of the target electric energy meter; determine whether the target electric energy meter meets the preset remote upgrade task requirements based on the electric energy meter information, and then return the upgrade plan to the remote upgrade background, in which the data is weighted according to the risk operation data and status performance.

[0032] The method for obtaining the upgrade status information of each electric energy meter is as follows: based on the current version information of the electric energy meter and the internal power supply information of the electric energy meter, the local status information is packaged and fed back to the background remote upgrade unit, and an upgrade plan is generated in the background remote upgrade unit. The historical data packet contains the data information of the power system, including power load data and time information data. The upgrade status information includes the version number of the current electric energy meter, the running time of the version, the specific location and the internal power supply information of the electric energy meter.

[0033] The data self-checking unit determines the future power load based on the data group B and generates the future power data time series Xf and its power peak value.

[0034] The load prediction unit compares the future electric energy data time series Xf and power peak value with the maximum rated power of the electric energy meter. The data self-check unit determines the working status of the electric energy meter in the future time period and feeds back to the background remote upgrade unit.

[0035] The background remote upgrade unit issues specific remote upgrade tasks based on the future working status of each electricity meter;

[0036] This method makes predictions based on the original data of the electricity meters under different usage scenarios. The background remote upgrade unit issues specific remote upgrade tasks based on the future working status of each electricity meter. It can make inferences based on the current usage environment and is suitable for the normal use of electricity meters in various environments.

[0037] Example 2: The cleaning process includes: performing an original judgment on the historical data packets, screening and removing the electric energy data that failed to be successfully collected and duplicate data, and then using the data concentrator to unify the data format, and accurate to three decimal places, and packaging it into a data packet;

[0038] A sliding window is used to generate time series samples. The data training unit is responsible for training the LSTM model for load forecasting, including model construction, training, and evaluation. The LSTM model is constructed using the hidden state of the time step and the input of the current time step, and the weight matrices of the forget gate, input gate, and candidate state are obtained. The weight matrices of the forget gate, input gate, and candidate state are activated using an activation function.

[0039] The LSTM model, through its unique gating mechanism, including input gate, forget gate and output gate, can effectively remember data information from a long period of time in the past. The LSTM model can accurately predict future load changes based on peak and low electricity consumption data from days or even weeks ago.

[0040] Example three: The data self-inspection unit can judge the working status of the electric energy meter in the future time period, and its basis is that if the power wave peak value passing through in the future time period is greater than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter cannot meet the maximum passing electric energy value in the future time period, and generates an optimization upgrade signal to feedback to the background remote upgrade unit. If the power wave peak value passing through in the future time period is less than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter does not need remote upgrade; the remote upgrade task issuance method includes but is not limited to software updates initiated by the remote master station, OTA updates based on the communication network, and local manual updates; therefore, the risk operation data and status performance are enhanced in the present invention. Specifically, the ratio of the value exceeding the power range to the rated power value and the average value of the connection deviation rate ratio are used as the prediction weighted items for a model construction, so as to meet the requirements of regulatory prediction of multiple risks in future environments and improve the service life of the electric energy meter at this location.

[0041] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A three-phase electric energy meter system with multiple communication modes and remote upgradeability, characterized in that: It includes: background parallel management unit, power information collection unit, background remote upgrade unit, power data processing unit, data training unit, data self-test unit and load prediction unit. When the cloud upgrade instruction is generated in the background parallel management unit, the cloud upgrade information instruction is transmitted to the power information collection unit. After receiving the upgrade instruction, the power information collection unit collects the upgrade status information and historical data packets of each power meter. The power data processing unit cleans the historical data packets, and builds a model for the processed historical data packets through the data training unit. The obtained data is then used for load prediction. The unit predicts data to obtain data group A of the electric energy meter, and then performs a secondary training model construction on data group A for the upgrade model to obtain data group B. The data self-checking unit determines the future electric energy load based on data group B and generates the future electric energy data time series Xf and its power wave peak value. The load prediction unit compares the future electric energy data time series Xf and power wave peak value with the maximum rated power of the electric energy meter. The data self-checking unit determines the working status of the electric energy meter in the future time period and feeds it back to the background remote upgrade unit. The background remote upgrade unit issues a specific remote upgrade task based on the future working status of each electric energy meter; The working status refers to whether the peak power wave passing through in the future time period is greater than or equal to the maximum rated power of the electric energy meter; The load forecasting unit performs data forecasting by determining risky operation data and status performance within the primary model construction data. The risky operation data includes power values ​​and connection deviation rates measured at a certain moment that far exceed the normal three-phase power range. The status performance data includes status performance values ​​and operating loss values. Prediction rules for secondary model construction are set, and the average of the ratio of the power range value to the rated power value and the connection deviation rate ratio is used as a prediction weighting item for the primary model construction to form data group A. The secondary model construction includes a data input unit and a secondary model training unit. By inputting data group A into the LSTM model, it is possible to judge and predict future situations and obtain a secondary prediction result, which is marked as data group B. Data group B is expanded to generate the future electric energy data time series Xf and its power peak value.

2. A three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 1, characterized in that: The method for obtaining the upgrade status information of each electricity meter is as follows: based on the current version information of the electricity meter and the internal power supply information of the electricity meter, determine whether the target electricity meter meets the preset remote upgrade task requirements, package it into local status information, and feedback the above local status information to the background remote upgrade unit, generate an upgrade plan in the background remote upgrade unit, the historical data packet contains data information of the power system, including power load data and time information data, and the upgrade status information includes the version number of the current electricity meter, the running time of the version, the specific location and the internal power supply information of the electricity meter.

3. A three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 2, characterized in that: The cleaning processing steps include: making an original judgment on the historical data packets, screening and removing the electric energy data and duplicate data that have not been successfully collected, and packaging them into data packets.

4. A three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 3, characterized in that: A model construction includes a data preparation unit and a model training unit. The data preparation unit converts the data packet into data as the input of the LSTM model, determines the number of LSTM layers and the number of neurons in each LSTM layer, and adds a fully connected layer to map the features output by the LSTM layer to the final prediction dimension. The fully connected layer linearly transforms the feature vector output by the LSTM layer through the weight matrix to obtain a prediction result, and uses a sliding window to generate time series samples.

5. The three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 1 is characterized in that: The basis for judging the working status of the electric energy meter in the future time period is that if the peak value of the power wave passing through in the future time period is greater than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter cannot meet the maximum passing electric energy value in the future time period, and an optimization upgrade signal is generated and fed back to the background remote upgrade unit. If the peak value of the power wave passing in the future time period is less than or equal to the maximum rated power of the electric energy meter, it means that the electric energy meter does not need remote upgrade.

6. The three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 1, characterized in that: Remote upgrade task delivery methods include but are not limited to software updates initiated by the remote master station, OTA updates based on the communication network, and local manual updates.

7. The three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 3, characterized in that: When cleaning data, historical data packets pass through the data concentrator to unify the data format and be accurate to three decimal places.

8. The three-phase electric energy meter system with multiple communication modes and remote upgradeability according to claim 4, characterized in that: The data training unit is responsible for training the LSTM model for load forecasting, including model construction, training and evaluation: the LSTM model is constructed through the hidden state of the time step and the input of the current time step, and the weight matrix of the forget gate, input gate and candidate state is obtained, and the activation function is used to activate the weight matrix of the forget gate, input gate and candidate state.

Images