Control method, system and device for power replacement and meter replacement without power outage

Abstract

The present application relates to a control method, system and equipment for non-stop power replacement and meter replacement, and relates to the technical field of power equipment, including: obtaining a target circuit load sequence to be replaced, calculating a circuit load variation coefficient; after switching the target circuit to the bypass, analyzing the circuit fluctuation and circuit heating according to the sequence, obtaining a circuit fluctuation coefficient and a circuit heating parameter; analyzing circuit abnormalities according to the two coefficients, obtaining and displaying the circuit abnormality rate; configuring an energy loss prediction coefficient according to the circuit abnormality rate, predicting energy flow parameters according to the circuit heating parameters, and performing energy compensation on the new meter after the power replacement and meter replacement are completed. The present invention solves the problem that in the traditional non-stop power replacement and meter replacement process, the circuit fluctuation abnormality affects the non-stop power effect due to the limited temporary carrying capacity of the bypass, and the energy loss affects the accuracy of energy metering, thereby improving the reliability and metering accuracy of the non-stop power replacement and meter replacement operations.

Description

Technical Field

[0001] The present application relates to the field of power equipment, and in particular to a control method, system and equipment for non-stop power replacement and meter replacement. Background Art

[0002] With the development of intelligent power system operation and maintenance technology, bypass safety control during non-stop meter replacement has become a key step in ensuring power supply reliability and metering accuracy. Currently, traditional non-stop power replacement and meter replacement technologies rely primarily on empirical procedures and static parameter settings, resulting in problems such as a lack of bypass capacity assessment, delayed circuit anomaly warnings, and extensive compensation for power loss.

[0003] The existing control method for battery replacement and meter replacement relies solely on manual judgment of the bypass access status, without real-time monitoring of circuit load fluctuations and heating parameters. This results in circuit abnormalities easily caused by bypass overload during the meter replacement process. At the same time, there is a lack of quantitative analysis of the power loss caused by bypass heating, making it difficult to accurately compensate for metering errors. This not only affects the effectiveness of meter replacement without power outages, but also damages user rights due to deviations in electricity metering. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides a control method, system and equipment for power replacement and meter replacement without power outage, which improves the accuracy of bypass temporary carrying capacity assessment, realizes effective response to abnormal circuit fluctuations, and solves the problem of inaccurate metering caused by power loss.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a method for controlling power switching and meter replacement without power outage, the method comprising:

[0007] Obtain the circuit load sequence of the target circuit to be replaced, and analyze and obtain the circuit load variation coefficient;

[0008] After switching the target circuit to bypass, performing circuit fluctuation analysis and circuit heating analysis according to the circuit load sequence to obtain a circuit fluctuation coefficient and circuit heating parameters;

[0009] Conduct circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain the circuit abnormality rate, and display it;

[0010] According to the circuit abnormality rate, the power loss prediction coefficient is configured, and the power loss is predicted according to the circuit heating parameters to obtain the power loss parameters. After the battery and meter are replaced, the new meter is compensated for power according to the power loss parameters.

[0011] Furthermore, a load test is performed on the target circuit to be replaced to obtain a circuit load sequence, wherein the target circuit is equipped with an old meter before the replacement;

[0012] A circuit load variation coefficient is obtained by calculation according to the circuit load sequence.

[0013] Furthermore, an average circuit load is calculated based on the circuit load sequence;

[0014] Obtaining a bypass performance predictor trained based on historical data of non-stop power switching, wherein the bypass performance predictor includes a circuit fluctuation prediction branch and a circuit heating prediction branch;

[0015] The circuit load sequence is input into the bypass performance predictor, and the circuit fluctuation coefficient and circuit heating parameters are obtained by prediction output.

[0016] Furthermore, based on the historical data of non-stop power replacement, a sample circuit load set is collected, and the proportion of the number of circuit fluctuations caused by bypass under different sample circuit loads is collected, and a sample circuit fluctuation coefficient set is obtained by annotation;

[0017] Collecting heating parameters of the bypass under different sample circuit loads to obtain a sample heating parameter set;

[0018] Based on machine learning, we build circuit fluctuation prediction branches and circuit heating prediction branches.

[0019] The sample circuit load set is used as input training data, and the sample circuit fluctuation coefficient set and the sample heating parameter set are used respectively to perform supervised training on the circuit fluctuation prediction branch and the circuit heating prediction branch until convergence, thereby obtaining a bypass performance predictor.

[0020] Furthermore, based on the historical data of non-stop power replacement, a set of sample circuit fluctuation coefficients and a set of sample circuit load variation coefficients are collected, and the proportion of bypass abnormalities under different sample circuit fluctuation coefficients and sample circuit load variation coefficients is collected, and the sample circuit abnormality rate set is obtained by annotation;

[0021] Based on a decision tree, a circuit anomaly classifier is constructed using the sample circuit fluctuation coefficient set, the sample circuit load variation coefficient set, and the sample circuit anomaly rate set;

[0022] The circuit fluctuation coefficient and the circuit load variation coefficient are input into the circuit anomaly classifier, and the circuit anomaly rate is obtained by classification and displayed.

[0023] Furthermore, the circuit abnormality rate is configured as a power loss prediction coefficient;

[0024] Obtain power loss prediction branch group;

[0025] Using the power loss prediction coefficient as a configuration ratio, selecting the power loss prediction branch of the configuration ratio, inputting the circuit heating parameter, predicting and outputting multiple branch power loss parameters, and calculating the average to obtain the power loss parameter;

[0026] After the battery and meter are replaced, the new meter is compensated for the energy loss according to the energy loss parameters.

[0027] Furthermore, based on the historical data of non-stop power swapping, a set of bypass sample circuit heating parameters is collected, and the power loss parameters during the power swapping process under different circuit heating parameters are tested and obtained, and the sample power loss parameter sets are marked;

[0028] randomly dividing the sample data of a preset proportion within the sample circuit heating parameter set and the sample power loss parameter set with replacement, and training a first power loss prediction branch based on machine learning;

[0029] Continue to train multiple power loss prediction branches to obtain a power loss prediction branch group.

[0030] In a second aspect, an embodiment of the present application provides a control system for power switching and meter replacement without power outage, the system comprising:

[0031] The load sequence analysis module is used to obtain the circuit load sequence of the target circuit to be replaced and analyze and obtain the circuit load variation coefficient;

[0032] A bypass parameter analysis module is used to perform circuit fluctuation analysis and circuit heating analysis according to the circuit load sequence after switching the target circuit to bypass, so as to obtain a circuit fluctuation coefficient and a circuit heating parameter;

[0033] An abnormality rate calculation module is used to perform circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain the circuit abnormality rate, and display it;

[0034] The electric energy compensation module is used to configure the electric energy loss prediction coefficient according to the circuit abnormality rate, predict the electric energy loss according to the circuit heating parameters, obtain the electric energy loss parameters, and after the battery and meter are replaced, perform electric energy compensation processing on the new meter according to the electric energy loss parameters.

[0035] In a third aspect, an embodiment of the present application provides a computer device, the device comprising:

[0036] memory for storing computer programs;

[0037] The processor is used to read and execute the computer program, thereby realizing a control method for power replacement and meter replacement without power outage in the first aspect.

[0038] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0039] The present application proposes a control method, system and equipment for non-stop battery replacement and meter replacement, which realizes accurate measurement of the battery replacement process through multi-dimensional parameter analysis and dynamic compensation mechanism. First, the load sequence of the target circuit to be replaced is obtained, and the load variation coefficient is obtained by analysis to provide data support for subsequent bypass switching and abnormality analysis. Then the circuit is switched to the bypass, and the fluctuation and heating analysis is performed based on the load sequence. The fluctuation coefficient and heating parameters are obtained through the bypass performance predictor trained with historical data. Then, the two types of parameters are combined for abnormal analysis, and the abnormality rate is output by the abnormality classifier constructed by the decision tree to intuitively display the circuit status. During the control process, the abnormality rate is configured as the energy loss prediction coefficient, combined with the heating parameter input to the multi-branch prediction group, and the loss parameter is obtained by calculating the mean through integrated learning. Finally, the new meter is compensated after the battery replacement is completed.

[0040] The technical solution of this application solves the problem of metering deviation caused by bypass heating in the traditional battery replacement and meter replacement process by integrating multiple steps such as circuit feature detection, multi-model prediction and dynamic compensation, realizes intelligent control of battery replacement and meter replacement without power outage, improves metering accuracy and battery replacement safety, and provides technical support for accurate metering and reliable operation of power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A flowchart of a method for controlling power replacement and meter replacement without power outages provided in an embodiment of the present application;

[0043] Figure 2 A schematic structural diagram of a control system for non-stop power replacement and meter replacement provided in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of the present application.

[0045] In the accompanying drawings, the components represented by the reference numerals are described as follows:

[0046] Load sequence analysis module 01 , bypass parameter analysis module 02 , abnormality rate calculation module 03 , power compensation module 04 , computer device 300 , memory 310 , processor 320 , computer program 311 . DETAILED DESCRIPTION

[0047] The present application provides a control method, system and equipment for non-stop power replacement and meter replacement, which are used to solve the technical problems existing in the prior art such as the lack of assessment of temporary bypass carrying capacity, abnormal circuit fluctuations affecting the non-stop power effect, and inaccurate metering due to power loss.

[0048] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0049] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically specified.

[0050] In the description of this application, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art will recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0051] Example 1, as shown in the attached Figure 1 As shown, the present application provides a control method for power replacement and meter replacement without power outage, the method comprising the following steps:

[0052] S110: Obtaining a circuit load sequence of a target circuit to be replaced, and analyzing to obtain a circuit load variation coefficient;

[0053] In the embodiment of the present application, in the intelligent control scenario of battery replacement and meter replacement without power outage, in order to achieve accurate evaluation and abnormal analysis of circuit status during battery replacement and meter replacement, it is necessary to obtain dynamic operation data of the target circuit through load testing.

[0054] Specifically, for the target circuit that is to be replaced and has an old meter installed, a high-precision current detection device is used to collect real-time load data to ensure that continuous operating data including parameters such as current is obtained in the energized state.

[0055] Furthermore, the collected load data is structured and organized in time series to form a circuit load sequence that can reflect the circuit operation characteristics.

[0056] At the same time, the amplitude and frequency of the load fluctuation of the circuit are calculated based on the circuit load sequence, so as to obtain the circuit load variation coefficient representing the circuit load variation trend.

[0057] This step provides basic data support reflecting the original operating characteristics of the circuit for subsequent circuit fluctuation analysis, heating parameter prediction and abnormality rate calculation after bypass switching through load data collection and dynamic analysis in the energized state.

[0058] Step S110 of the method provided in the embodiment of the present application includes:

[0059] Performing a load test on a target circuit to be replaced to obtain a circuit load sequence, wherein the target circuit is equipped with an old meter before the replacement;

[0060] A circuit load variation coefficient is obtained by calculation according to the circuit load sequence.

[0061] In the embodiment of the present application, in order to achieve accurate analysis of the circuit status and parameter optimization during the battery replacement process, it is necessary to obtain the dynamic operation data of the target circuit through systematic load testing.

[0062] Specifically, first, for the target circuit that needs to be replaced and has an old meter installed, based on its voltage level and load type, a high-precision ammeter with an appropriate range is selected to collect real-time circuit load data under the energized operating state.

[0063] For example, for a three-phase four-wire low-voltage circuit, a three-phase clamp ammeter with a level 0.5 accuracy is used to synchronously sample the currents of the L1, L2, L3 phase lines and the N line, and the sampling frequency is not less than 100 Hz to ensure that the transient fluctuation characteristics of the load are captured.

[0064] Furthermore, the acquired raw data, such as current, is preprocessed. Specifically, a sliding average filter is used to remove sampling noise, and then the instantaneous values ​​of each parameter are organized into two-dimensional sequence data in timestamp order, thereby forming a circuit load sequence that reflects the real-time operating status of the circuit.

[0065] Furthermore, based on the obtained circuit load sequence, a quantitative analysis is performed on it to obtain a corresponding circuit load variation coefficient.

[0066] Specifically, the circuit load sequence is first divided into multiple time periods with a fixed duration (e.g., 10 minutes, which can be adjusted based on load characteristics). The current data for each time period is then extracted. The difference ΔI between the maximum and minimum RMS current values ​​within each time period is calculated as the current fluctuation amplitude within that time period.

[0067] For example, if the maximum current value in a 10-minute time period is 50 A and the minimum current value is 30 A, then the current fluctuation amplitude in this time period is ΔI=50-30=20 A.

[0068] Furthermore, the current fluctuation ratio of the target circuit within the time period is calculated based on the obtained current fluctuation amplitude and the rated current of the target circuit.

[0069] The current fluctuation ratio is calculated as "current fluctuation ratio = current fluctuation amplitude / rated current." The specific value of the rated current is determined based on the actual power supply specifications of the target circuit, the rated parameters of the electrical equipment, and the national grid standards.

[0070] For example, if a target circuit is a three-phase 380V industrial power supply system, the rated current is determined to be 60A according to the GB / T15283-1994 standard, and the current fluctuation amplitude is ΔI=20A within a 10-minute time period, then the corresponding current fluctuation ratio = 20A / 60A≈0.33.

[0071] Finally, the arithmetic mean of the current fluctuation ratios in all time periods in the target circuit is calculated to obtain the circuit load variation coefficient (i.e., circuit load variation coefficient = Σcurrent fluctuation ratio of a single time period / number of time periods) to achieve a quantitative assessment of the degree of dynamic fluctuation of the circuit load.

[0072] For example, if the circuit load sequence of a target circuit is divided into five time periods, and the current fluctuation ratios of each time period are 0.356, 0.32, 0.28, 0.31, and 0.33, respectively, then the circuit load variation coefficient of the target circuit = (0.356+0.32+0.28+0.31+0.33) / 5=0.3192.

[0073] Among them, the larger the circuit load variation coefficient, the greater the fluctuation of the circuit load before battery replacement and the more drastic the change in current; conversely, the smaller the coefficient, the more stable the circuit load, the smaller the fluctuation amplitude of the current, and the more stable the circuit operation state.

[0074] The circuit load variation coefficient obtained through the above steps can intuitively reflect the severity of the circuit load fluctuation before the battery replacement, providing a quantitative basis for the subsequent circuit status analysis during bypass switching and the operation evaluation after the new meter is connected.

[0075] S120: After switching the target circuit to bypass, performing circuit fluctuation analysis and circuit heating analysis according to the circuit load sequence to obtain a circuit fluctuation coefficient and circuit heating parameters;

[0076] In the embodiment of the present application, in the intelligent control scenario of power replacement and meter replacement without power outage, in order to achieve accurate assessment and risk prediction of the circuit status after bypass switching, it is necessary to conduct dynamic characteristic analysis of circuit fluctuations and heating conditions based on the circuit load sequence.

[0077] Specifically, after the target circuit is switched to temporary bypass, the average circuit load in the sequence is first calculated based on the acquired circuit load sequence to provide a basic reference value for subsequent prediction analysis.

[0078] Furthermore, the bypass performance predictor trained based on the historical data of non-stop power replacement is called, and the circuit load sequence is input into the predictor. Through the calculation and processing of the machine learning model, the circuit fluctuation coefficient and circuit heating parameters that can reflect the dynamic characteristics of the circuit are output.

[0079] Among them, the predictor integrated circuit fluctuation prediction branch and the circuit heating prediction branch can realize the synchronous prediction of dual-dimensional parameters.

[0080] This step converts the circuit load sequence into quantifiable fluctuation and heating characteristic parameters through the process of "data input - model prediction - dual parameter output", providing data support for subsequent circuit anomaly analysis and power loss prediction, ensuring that the operating status assessment of the bypass system during the battery replacement process is scientific and reliable.

[0081] Step S120 in the method provided in the embodiment of the present application includes:

[0082] Calculating an average circuit load according to the circuit load sequence;

[0083] Obtaining a bypass performance predictor trained based on historical data of non-stop power switching, wherein the bypass performance predictor includes a circuit fluctuation prediction branch and a circuit heating prediction branch;

[0084] The circuit load sequence is input into the bypass performance predictor, and the circuit fluctuation coefficient and circuit heating parameters are obtained by prediction output.

[0085] In the embodiment of the present application, in the intelligent control scenario of power replacement and meter replacement without power outage, in order to achieve accurate assessment and risk prediction of the circuit status after bypass switching, it is necessary to carry out dynamic characteristic analysis based on the circuit load sequence.

[0086] Specifically, the average circuit load is first calculated based on the circuit load sequence. The average circuit load can reflect the load level of the target circuit before battery replacement and provide a basic reference value for subsequent prediction analysis.

[0087] For example, if the circuit load sequence of a target circuit includes current data for five 10-minute time periods, namely 45 A, 50 A, 38 A, 42 A, and 47 A, the average circuit load is (45 + 50 + 38 + 42 + 47) / 5 ≈ 44.4 A.

[0088] The average circuit load parameter directly reflects the actual operating load of the target circuit before the battery swap. A larger parameter indicates a higher actual operating load and a higher operating intensity. Conversely, a smaller parameter indicates a lower actual operating load and a lower operating intensity.

[0089] Furthermore, the average circuit load parameters are sequentially input into a bypass performance predictor trained based on historical data of non-stop power switching to predict corresponding parameters that can reflect the dynamic characteristics of the circuit.

[0090] The training steps of the “bypass performance predictor” in the method provided in the embodiment of the present application include:

[0091] Based on the historical data of non-stop power replacement, a sample circuit load set is collected, and the proportion of the number of circuit fluctuations caused by bypass under different sample circuit loads is collected, and the sample circuit fluctuation coefficient set is obtained by annotation;

[0092] Collecting heating parameters of the bypass under different sample circuit loads to obtain a sample heating parameter set;

[0093] Based on machine learning, we build circuit fluctuation prediction branches and circuit heating prediction branches.

[0094] The sample circuit load set is used as input training data, and the sample circuit fluctuation coefficient set and the sample heating parameter set are used respectively to perform supervised training on the circuit fluctuation prediction branch and the circuit heating prediction branch until convergence, thereby obtaining a bypass performance predictor.

[0095] In the embodiment of the present application, in order to achieve accurate prediction of bypass performance, it is necessary to construct a bypass performance predictor through machine learning, and drive model training through multi-dimensional historical data to predict and output reliable circuit fluctuation coefficients and circuit heating parameters.

[0096] Specifically, a complete data set of circuit loads is firstly screened out from the historical data of battery swapping without power outages, and is used as a sample circuit load set.

[0097] Among them, the historical battery replacement data is obtained by real-time monitoring and collection of the circuit operation status during past non-stop battery replacement operations. The monitoring equipment includes high-precision ammeters, etc. to ensure the accuracy and completeness of the data.

[0098] Furthermore, for each sample circuit load, by simulating the circuit operating conditions in the actual battery replacement scenario, the proportion of the number of circuit fluctuations occurring in the bypass under the load is collected, and after labeling and integration, a set of sample circuit fluctuation coefficients is obtained.

[0099] The calculation formula for the circuit fluctuation coefficient is "circuit fluctuation coefficient = number of bypass fluctuations / monitoring time (minutes)". The size of this value directly reflects the frequency of circuit fluctuations in the bypass system per unit time.

[0100] For example, when simulating a 40A circuit load condition, the bypass system is monitored to fluctuate three times within one hour. Based on this, the corresponding circuit fluctuation coefficient is marked as 0.05 (3 / 60), indicating that the bypass under this load fluctuates an average of 0.05 times per minute, which is a relatively low-frequency fluctuation state and the circuit stability is relatively good.

[0101] Similarly, for each identical sample circuit load, by deploying temperature sensors for real-time monitoring under simulated working conditions, the heating parameters of each bypass are collected to obtain a sample heating parameter set.

[0102] Specifically, when simulating the circuit operating conditions of an actual battery swapping scenario, high-precision infrared temperature sensors are installed at key locations such as bypass connection points and cables. A sampling frequency of once per second is set to monitor the temperature in real time, and the temperature change data of the bypass under a specific load is recorded. The highest temperature value is used as the heating parameter of the bypass.

[0103] For example, in a simulation test of a 40A current load, an infrared temperature sensor is used to continuously monitor for 30 minutes to obtain a temperature data sequence of the bypass connection point. After data processing, the highest temperature value extracted is 65°C. This value is used as the heating parameter of the bypass under the load of the sample circuit, which can intuitively reflect the heating status of the bypass under the corresponding load.

[0104] Furthermore, based on machine learning, a circuit fluctuation prediction branch and a circuit heating prediction branch are constructed. That is, the random forest algorithm is used as the basic framework, and the prediction capabilities of multiple decision trees are integrated to achieve regular learning of the coupling relationship between circuit load characteristics and bypass performance parameters.

[0105] Specifically, the sample circuit load set is used as input features, the sample circuit fluctuation coefficient set and the sample heating parameter set are used as output labels. By constructing multiple decision trees and introducing a random sampling mechanism (including random sampling of samples and random selection of features), each branch model can automatically analyze the key influencing factors of circuit fluctuation and heating under different load conditions.

[0106] First, the collected sample data is divided into training, validation, and test sets in a ratio of 7:2:1 to ensure that the data distribution covers circuit characteristics under different operating conditions such as light load, medium load, and heavy load. For example, from 1,000 sets of historical samples, 700 sets of training sets, 200 sets of validation sets, and 100 sets of test sets are extracted. Each set of data contains circuit load parameters and the circuit fluctuation coefficient and heating parameters under the corresponding load.

[0107] Furthermore, when the random forest model is trained using the training set, 80% of the sample data is randomly selected from the sample circuit load set each time, and 60% of the feature parameters are randomly selected from the load feature dimension and input into the model.

[0108] Taking the circuit fluctuation prediction branch as an example, the circuit load parameters (such as the average current of 40A) are used as input features, and the circuit fluctuation coefficient (such as 0.05 times / minute) is used as the output label. The mapping relationship between the load parameters and the fluctuation coefficient is learned through the automatic splitting mechanism of the decision tree.

[0109] Furthermore, during the training process, the initial number of decision trees was set at 50, and the model performance was evaluated using the validation set every time 10 trees were added. After the first round of training, the average error between the predicted and actual values ​​of the volatility coefficient on the validation set was 0.03. When the number of decision trees was increased to 100, the error narrowed to 0.015.

[0110] To prevent overfitting, an early stopping mechanism was implemented. Training was stopped when the validation set error fluctuated less than 0.002 for five consecutive rounds. For example, if the validation set error stabilized at 0.012 during the 30th round of training and did not decrease for five consecutive rounds, the fluctuation prediction branch model was considered converged.

[0111] Similarly, the heat prediction branch is trained using the same mechanism, taking circuit load parameters as input features and heat parameters (such as maximum temperature) as output labels to build a converged heat prediction branch. For example, when the load feature "average current of 55A" is input, the ensemble prediction of 150 decision trees outputs a maximum temperature prediction of 68°C.

[0112] Finally, the trained bypass performance predictor can receive circuit load sequence input, and through parallel calculation of the circuit fluctuation prediction branch and the heat prediction branch, output the corresponding circuit fluctuation coefficient and heat parameter prediction value, providing a quantitative basis for risk assessment during the battery replacement process.

[0113] For example, when performing a non-stop power swap on a target circuit, the average current of the circuit load sequence is obtained as 55A. This circuit load is then fed into the trained bypass performance predictor. The predictor, using a parallel calculation branch for circuit fluctuation prediction and a circuit heating prediction branch, outputs a circuit fluctuation coefficient of 0.06 times / minute and a maximum heating parameter of 68°C.

[0114] Furthermore, based on the output circuit fluctuation coefficient and heating parameters, the operating status of the circuit after bypass switching can be evaluated. If the fluctuation coefficient is large or the temperature is too high, the operator will be prompted to take corresponding measures, such as adjusting the battery replacement operation steps or strengthening heat dissipation to ensure the safety and reliability of the battery replacement process.

[0115] S130: Perform circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain a circuit abnormality rate, and display it;

[0116] In the embodiment of the present application, in order to avoid bypass anomalies (such as short circuit and overload) caused by circuit fluctuations and load changes, it is necessary to construct an anomaly classifier by combining the circuit fluctuation coefficient and the load change coefficient two-dimensional combined analysis to output the real-time circuit anomaly rate and display it.

[0117] Specifically, first, based on the historical data of non-stop power replacement, a set of sample circuit fluctuation coefficients and a set of sample circuit load variation coefficients are collected, and the proportion of bypass anomalies under different sample circuit fluctuation coefficients and sample circuit load variation coefficients are collected, and the sample circuit anomaly rate set is obtained by annotation.

[0118] Furthermore, based on the decision tree algorithm, the above three sets are used to construct a circuit anomaly classifier, and a mapping relationship of "current fluctuation-load change-abnormality probability" is established to achieve accurate classification of real-time circuit anomaly rate.

[0119] Furthermore, the real-time circuit fluctuation coefficient and circuit load variation coefficient are input into the trained circuit anomaly classifier. After hierarchical splitting and feature analysis of the decision tree, the corresponding circuit anomaly rate is output and displayed to the operator in a visual interface.

[0120] This step uses machine learning modeling driven by historical data to convert abstract current fluctuation characteristics and load change patterns into quantifiable abnormal risk indicators, realizing real-time assessment and risk classification of bypass operating status, and providing data support for safe operations during non-stop power replacement.

[0121] Step S130 in the method provided in the embodiment of the present application includes:

[0122] Based on the historical data of non-stop power replacement, the sample circuit fluctuation coefficient set and the sample circuit load variation coefficient set are collected. The proportion of bypass abnormalities under different sample circuit fluctuation coefficients and sample circuit load variation coefficients is collected, and the sample circuit abnormality rate set is obtained by annotation.

[0123] Based on a decision tree, a circuit anomaly classifier is constructed using the sample circuit fluctuation coefficient set, the sample circuit load variation coefficient set, and the sample circuit anomaly rate set;

[0124] The circuit fluctuation coefficient and the circuit load variation coefficient are input into the circuit anomaly classifier, and the circuit anomaly rate is obtained by classification and displayed.

[0125] In the embodiment of the present application, in the intelligent control scenario of power replacement and meter replacement without power outage, in order to achieve accurate identification and risk classification of abnormal conditions of the bypass system, it is necessary to build a circuit abnormality classifier based on the circuit fluctuation coefficient and load variation coefficient for quantitative analysis.

[0126] Specifically, firstly, based on the historical data of non-stop power replacement, a sample circuit fluctuation coefficient set and a sample circuit load change coefficient set are collected.

[0127] Among them, the circuit fluctuation coefficient is obtained by analyzing the historical circuit load sequence through the bypass performance predictor, reflecting the severity of the current fluctuation after the bypass switching; the circuit load variation coefficient is obtained by calculating the fluctuation amplitude and frequency of the historical circuit load sequence, characterizing the dynamic change degree of the circuit load before battery replacement.

[0128] Furthermore, for different combinations of sample circuit fluctuation coefficients and sample circuit load variation coefficients, the proportion of bypass anomalies (such as short circuits and overloads) is statistically analyzed and annotated to form a sample circuit anomaly rate set.

[0129] For example, in historical data, when the circuit fluctuation coefficient is 0.3 and the load variation coefficient is 0.2, the probability of bypass short circuit is 35%. This ratio is marked and included in the sample circuit abnormality rate set.

[0130] Furthermore, based on the decision tree algorithm, the above three sets are used to construct a circuit anomaly classifier. This classifier automatically mines the mapping relationship between "circuit fluctuation coefficient - load variation coefficient - circuit anomaly probability" through supervised learning.

[0131] For example, the classifier might learn the splitting rule "When the circuit fluctuation coefficient is greater than 0.3 and the load variation coefficient is greater than 0.25, the bypass anomaly rate exceeds 40%." This rule is derived from the feature extraction of a 42% bypass short circuit probability when the fluctuation coefficient is 0.35 and the load variation coefficient is 0.28 in historical samples.

[0132] Furthermore, during the training process, the sample circuit fluctuation coefficient set and the sample circuit load variation coefficient set are used as input features, and the sample circuit anomaly rate set is used as the output label. Through the hierarchical splitting and feature allocation of the decision tree, accurate modeling of the anomaly probability is achieved.

[0133] Specifically, the sample data was first divided into training, validation, and test sets in a ratio of 7:2:1 to ensure that the data covered abnormal situations under different combinations of fluctuation coefficients and load variation coefficients. For example, 700 training sets were extracted from 1000 sets of historical samples, containing parameter combinations and corresponding abnormality rates under different operating conditions such as light load (<30A), medium load (30-60A), and heavy load (>60A).

[0134] During training, the decision tree starts from the root node and selects the feature (circuit fluctuation coefficient or load variation coefficient) that best distinguishes the abnormality rate for splitting based on information gain.

[0135] For example, the first split is based on the "circuit fluctuation coefficient > 0.2" condition, dividing the samples into two groups: a group with a higher fluctuation coefficient (average anomaly rate of 35%) and a group with a lower fluctuation coefficient (average anomaly rate of 15%). The split is then continued for each child node until the preset stopping condition is met (i.e., the anomaly rate accuracy meets the target).

[0136] During the training process, the model performance is evaluated through the validation set. An evaluation is performed every time 10 decision trees are added. When the prediction error of the validation set anomaly rate fluctuates by less than 2% for five consecutive rounds, the early stopping mechanism is triggered to avoid model overfitting and determine that the model has converged.

[0137] After training is complete, the real-time circuit fluctuation coefficient and circuit load variation coefficient are input into the converged circuit anomaly classifier. The classifier uses the learned splitting rules to make hierarchical judgments, outputs the corresponding circuit anomaly rate, and displays it to the operator through a visual interface.

[0138] For example, when the input real-time circuit fluctuation coefficient is 0.25 and the load change coefficient is 0.15, the abnormality classifier outputs an abnormality rate of 22%, and displays the bypass as a "low-risk abnormality"; on the contrary, if the abnormality rate exceeds the preset threshold (such as 30%), the bypass is displayed as a "high-risk abnormality", which will automatically trigger the suspension of battery replacement mechanism and prompt the operator to inspect the circuit.

[0139] This step converts abstract circuit parameters into quantifiable abnormality indicators through a closed-loop process of "historical data collection - abnormality classifier construction - real-time abnormality classification - classification display", realizes early prediction of potential bypass faults, and provides data support for the safety and reliability of the non-stop power replacement process.

[0140] S140: According to the circuit abnormality rate, configure the power loss prediction coefficient, predict the power loss according to the circuit heating parameters, obtain the power loss parameters, and after the battery and meter are replaced, perform power compensation processing on the new meter according to the power loss parameters.

[0141] In the embodiment of the present application, in the intelligent control scenario of battery replacement and meter replacement without power outage, in order to ensure the accuracy of electricity metering during the battery replacement process, it is necessary to combine the circuit abnormality rate and circuit heating parameters to predict the energy loss, so as to achieve accurate compensation of the new meter after the battery replacement.

[0142] Specifically, the circuit abnormality rate is first configured as the power loss prediction coefficient, which is used to dynamically adjust the number of models involved in the power loss prediction.

[0143] Furthermore, based on the historical data of battery replacement without power outage, a set of bypass sample circuit heating parameters is collected, and the energy loss parameters of battery replacement under different heating parameters are obtained through real-time measurement, and they are labeled and integrated to form a corresponding set of sample energy loss parameters.

[0144] Furthermore, using random partitioning with replacement, the sample data is divided into a training set and a validation set according to a preset ratio. Based on the machine learning algorithm, multiple power loss prediction branches are trained to construct a prediction branch cluster. Each prediction branch can learn the mapping relationship between specific heating characteristics and power loss.

[0145] Furthermore, when predicting power loss, the power loss prediction branch that matches the power loss prediction coefficient ratio is selected, and the real-time circuit heating parameter is used as input. After each branch independently outputs the predicted value, the average is calculated to obtain the final power loss parameter.

[0146] Finally, after the battery and meter replacement operations are completed, the energy loss parameter is used to control the energy compensation processing of the new meter, and the compensation value is written into the new meter to ensure the accuracy of energy measurement.

[0147] Through the above steps, the abnormal degree of the circuit operation status and the bypass heating characteristics are converted into quantifiable compensation data, which effectively solves the problem of energy metering deviation caused by bypass heating during the non-stop power replacement process.

[0148] Step S140 in the method provided in the embodiment of the present application includes:

[0149] Configuring the circuit abnormality rate as a power loss prediction coefficient;

[0150] Obtain power loss prediction branch group;

[0151] Using the power loss prediction coefficient as a configuration ratio, selecting the power loss prediction branch of the configuration ratio, inputting the circuit heating parameter, predicting and outputting multiple branch power loss parameters, and calculating the average to obtain the power loss parameter;

[0152] After the battery and meter are replaced, the new meter is compensated for energy loss according to the energy loss parameters.

[0153] In the embodiment of the present application, in order to solve the problem of energy metering deviation caused by bypass heating during the battery replacement process, it is necessary to establish a dynamic energy loss prediction mechanism based on the circuit abnormality rate, and realize accurate compensation of the new meter through integrated calculation of multiple branch models.

[0154] Specifically, the circuit abnormality rate is first directly configured as the power loss prediction coefficient, and the coefficient is used to determine the proportion of the number of branches participating in the power loss prediction.

[0155] Furthermore, according to the power loss prediction coefficient, multiple power loss prediction branches are trained through historical data to obtain a power loss prediction branch group.

[0156] The step of “obtaining a power loss prediction branch group” in the method provided in the embodiment of the present application includes:

[0157] Based on the historical data of non-stop power replacement, the bypass sample circuit heating parameter set is collected, and the power loss parameters during the battery replacement process under different circuit heating parameters are tested and obtained, and the sample power loss parameter set is marked;

[0158] randomly dividing the sample data of a preset proportion within the sample circuit heating parameter set and the sample power loss parameter set with replacement, and training a first power loss prediction branch based on machine learning;

[0159] Continue to train multiple power loss prediction branches to obtain a power loss prediction branch group.

[0160] In the embodiment of the present application, in order to achieve accurate prediction of power loss during the battery swapping process, it is necessary to construct a power loss prediction branch group based on machine learning to improve the robustness and accuracy of the prediction through multi-branch integration.

[0161] Specifically, first, a set of bypass sample circuit heating parameters is collected from the historical operation data of non-stop power replacement, including key parameters such as the maximum temperature of the bypass connection point.

[0162] At the same time, for the working conditions of different circuit heating parameters, the energy loss parameters (i.e. the amount of lost electricity) during the battery replacement process are obtained through actual testing, and they are standardized and labeled to form a sample energy loss parameter set.

[0163] Specifically, energy loss parameters are acquired by using a high-precision energy meter to record the difference in energy consumption (i.e., the amount of energy lost) during the battery swap process without power outages. After standardizing the energy loss parameters by unitizing their numerical units and removing abnormal data, a set of sample energy loss parameters is generated that corresponds one-to-one with the sample circuit heating parameter set.

[0164] For example, in the historical data of non-stop battery replacement, when the maximum bypass temperature is 65°C, the energy loss during 30 minutes of battery replacement is 0.5 degrees. This group of data is annotated and included in the sample energy loss parameter set.

[0165] Furthermore, using the decision tree algorithm as the basic framework, a power loss prediction branch is constructed using a set of sample circuit heating parameters and a set of sample power loss parameters. This branch automatically discovers the mapping relationship between circuit heating parameters and power loss parameters through supervised learning.

[0166] Furthermore, during the training process, the sample circuit heating parameter set is used as input features, and the sample power loss parameter set is used as output labels. Through the hierarchical splitting and feature allocation of the decision tree, accurate prediction of the power loss parameters is achieved.

[0167] Specifically, a random sampling method with replacement is used to divide the training, validation, and test sets from the sample circuit heating parameter sets and the sample power loss parameter sets into a preset ratio (e.g., 7:2:1) to ensure that the data covers the power loss scenarios under different heating parameter combinations. For example, 700 training sets are drawn from 1000 sets of historical samples, containing heating parameters and corresponding power loss parameters under different operating conditions, such as low heat (<50°C), medium heat (50-70°C), and high heat (>70°C).

[0168] Furthermore, during training, the decision tree starts from the root node and selects the heating feature (i.e., the highest temperature) that best distinguishes the power loss parameters for splitting based on information gain.

[0169] For example, the first split is based on the "maximum temperature > 60°C" condition, dividing the samples into two groups: a high-temperature group (average power loss of 0.6°C) and a low-temperature group (average power loss of 0.2°C). The split is then continued for each child node until the preset stopping condition is met (i.e., the accuracy of power loss prediction reaches the target).

[0170] Similarly, during the training process, the model performance is evaluated through the validation set, and an evaluation is performed every time 10 decision trees are added. When the energy loss prediction error of the validation set fluctuates by less than 3% for five consecutive rounds, the early stopping mechanism is triggered to avoid model overfitting and determine that the prediction branch has converged.

[0171] Finally, after training is complete, the real-time circuit heating parameters are fed into the converged first power loss prediction branch, which performs hierarchical judgment based on the learned splitting rules and outputs the corresponding power loss parameter prediction value.

[0172] For example, when the input real-time maximum temperature is 68°C, the predicted branch output power loss parameter is 0.8 degrees; if the input parameter is the maximum temperature of 55°C, the output predicted value is 0.3 degrees.

[0173] Similarly, we continue to use the same sample division method and decision tree algorithm, adjust parameters such as the splitting threshold and tree depth, train multiple different power loss prediction branches, and finally obtain a power loss prediction branch group, providing multi-model support for subsequent dynamic power loss prediction based on circuit abnormality rate.

[0174] Furthermore, when the circuit abnormality rate is configured as the power loss prediction coefficient, a corresponding number of branches are automatically selected from the power loss prediction branch group according to the ratio of the coefficient, and the real-time circuit heating parameters are input into each branch for parallel calculation.

[0175] Among them, when the circuit abnormality rate is high, in order to improve the prediction accuracy under complex working conditions, a higher proportion of prediction branches are selected to participate in the calculation; when the abnormality rate is low, a lower proportion of branches are selected to improve the calculation efficiency.

[0176] For example, if the current circuit abnormality rate is 40%, 70 power loss prediction branches are selected for the calculation. Real-time collected circuit heating parameters (such as a maximum temperature of 68°C) are input into each branch, and each branch independently outputs a power loss prediction value based on the trained decision tree model.

[0177] Furthermore, after obtaining multiple power loss prediction values ​​output by each branch, these prediction values ​​are arithmetic averaged to obtain the final power loss prediction value, thereby achieving accurate quantification of power loss during the battery replacement process.

[0178] For example, if 70 power loss prediction branches are selected to participate in the calculation, the predicted values ​​output by each branch are 70 values ​​such as 0.78 degrees, 0.82 degrees, and 0.79 degrees. After the arithmetic mean calculation, the average value of these predicted values ​​is 0.8 degrees. This average value is used as the final power loss parameter for compensation processing of the new meter.

[0179] Finally, after the battery and meter swap is complete, the final energy loss parameter (e.g., 0.8 kWh) is automatically retrieved and the compensation value is added to the new meter's initial reading via the meter communication interface. For example, if the new meter initially reads 100 kWh, the compensation will display 100.8 kWh, thus calibrating the energy loss caused by bypass heating during the battery swap.

[0180] The embodiments of the present application achieve the following technical effects through the above specific implementation methods:

[0181] The present application proposes a control method for battery replacement and meter replacement without power outage. First, a load test is performed on the target circuit to be replaced, the circuit load sequence is obtained, and the load variation coefficient is calculated to provide a data basis for subsequent analysis. Secondly, the circuit is switched to the bypass, the average load is calculated based on the load sequence, and the bypass performance predictor including the circuit fluctuation prediction branch and the circuit heating prediction branch is called to output the circuit fluctuation coefficient and the circuit heating parameter. Then, combining the circuit fluctuation coefficient and the load variation coefficient, an anomaly analysis is performed using a circuit anomaly classifier constructed based on a decision tree, the circuit anomaly rate is output and visualized, and the bypass fault is predicted in advance. During the control process, the circuit anomaly rate is configured as the energy loss prediction coefficient to obtain the energy loss prediction branch group, and the number of branches is selected according to the ratio of the coefficient. After the circuit heating parameter is input, the energy loss parameter is obtained by arithmetic average calculation. After the battery replacement and meter replacement are completed, the energy loss parameter is written into the new meter as a compensation value to achieve energy loss measurement calibration.

[0182] The method provided in the embodiment of the present application solves the problems of circuit fluctuation abnormalities and metering deviation caused by limited temporary carrying capacity and bypass heating in traditional battery replacement through a closed-loop process of "load analysis - bypass prediction - abnormality assessment - loss compensation", improves the safety and metering accuracy during non-stop battery replacement and meter replacement, and provides technical support for the reliable operation of the power system.

[0183] Example 2, as shown in the attached Figure 2 As shown, based on the inventive concept of a control method for non-stop power switching and meter replacement provided in Example 1, the present application also provides a non-stop power switching and meter replacement control system, specifically including:

[0184] Load sequence analysis module 01 is used to obtain the circuit load sequence of the target circuit to be replaced and analyze and obtain the circuit load variation coefficient;

[0185] Bypass parameter analysis module 02, configured to perform circuit fluctuation analysis and circuit heating analysis according to the circuit load sequence after switching the target circuit to bypass mode, and obtain circuit fluctuation coefficient and circuit heating parameter;

[0186] The abnormality rate calculation module 03 is used to perform circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain the circuit abnormality rate, and display it;

[0187] The electric energy compensation module 04 is used to configure the electric energy loss prediction coefficient according to the circuit abnormality rate, predict the electric energy loss according to the circuit heating parameters, obtain the electric energy loss parameters, and after the battery and meter are replaced, perform electric energy compensation processing on the new meter according to the electric energy loss parameters.

[0188] In one embodiment, the load sequence analysis module 01 is further configured to:

[0189] Performing a load test on a target circuit to be replaced to obtain a circuit load sequence, wherein the target circuit is equipped with an old meter before the replacement;

[0190] A circuit load variation coefficient is obtained by calculation according to the circuit load sequence.

[0191] In one embodiment, the bypass parameter analysis module 02 is further configured to:

[0192] Calculating an average circuit load according to the circuit load sequence;

[0193] Obtaining a bypass performance predictor trained based on historical data of non-stop power switching, wherein the bypass performance predictor includes a circuit fluctuation prediction branch and a circuit heating prediction branch;

[0194] The circuit load sequence is input into the bypass performance predictor, and the circuit fluctuation coefficient and circuit heating parameters are obtained by prediction output.

[0195] The training steps of the bypass performance predictor include:

[0196] Based on the historical data of non-stop power replacement, a sample circuit load set is collected, and the proportion of the number of circuit fluctuations caused by bypass under different sample circuit loads is collected, and the sample circuit fluctuation coefficient set is obtained by annotation;

[0197] Collecting heating parameters of the bypass under different sample circuit loads to obtain a sample heating parameter set;

[0198] Based on machine learning, we build circuit fluctuation prediction branches and circuit heating prediction branches.

[0199] The sample circuit load set is used as input training data, and the sample circuit fluctuation coefficient set and the sample heating parameter set are used respectively to perform supervised training on the circuit fluctuation prediction branch and the circuit heating prediction branch until convergence, thereby obtaining a bypass performance predictor.

[0200] In one embodiment, the abnormality rate calculation module 03 is further used to:

[0201] Based on the historical data of non-stop power replacement, the sample circuit fluctuation coefficient set and the sample circuit load variation coefficient set are collected. The proportion of bypass abnormalities under different sample circuit fluctuation coefficients and sample circuit load variation coefficients is collected, and the sample circuit abnormality rate set is obtained by annotation.

[0202] Based on a decision tree, a circuit anomaly classifier is constructed using the sample circuit fluctuation coefficient set, the sample circuit load variation coefficient set, and the sample circuit anomaly rate set;

[0203] The circuit fluctuation coefficient and the circuit load variation coefficient are input into the circuit anomaly classifier, and the circuit anomaly rate is obtained by classification and displayed.

[0204] In one embodiment, the electric energy compensation module 04 is further configured to:

[0205] Configuring the circuit abnormality rate as a power loss prediction coefficient;

[0206] Obtain power loss prediction branch group;

[0207] Using the power loss prediction coefficient as a configuration ratio, selecting the power loss prediction branch of the configuration ratio, inputting the circuit heating parameter, predicting and outputting multiple branch power loss parameters, and calculating the average to obtain the power loss parameter;

[0208] After the battery and meter are replaced, the new meter is compensated for the energy loss according to the energy loss parameters.

[0209] Among them, obtaining the power loss prediction branch group includes:

[0210] Based on the historical data of non-stop power replacement, the bypass sample circuit heating parameter set is collected, and the power loss parameters during the battery replacement process under different circuit heating parameters are tested and obtained, and the sample power loss parameter set is marked;

[0211] randomly dividing the sample data of a preset proportion within the sample circuit heating parameter set and the sample power loss parameter set with replacement, and training a first power loss prediction branch based on machine learning;

[0212] Continue to train multiple power loss prediction branches to obtain a power loss prediction branch group.

[0213] Example 3, as shown in the attached Figure 3 As shown, based on the inventive concept of a control method for non-stop power switching and meter replacement provided in Example 1, the present application further provides a computer device 300, which comprises:

[0214] Memory 310 for storing computer programs 311;

[0215] The processor 320 is configured to read and execute the computer program 311 .

[0216] When the computer program is executed by the processor, a control method for power replacement and meter replacement without power outage in the first embodiment is implemented.

[0217] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0218] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0219] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A control method for power replacement and meter replacement without power outage, characterized in that: The method comprises: Obtain the circuit load sequence of the target circuit to be replaced, and analyze and obtain the circuit load variation coefficient; After switching the target circuit to bypass mode, perform circuit fluctuation analysis and circuit heating analysis based on the circuit load sequence to obtain circuit fluctuation coefficient and circuit heating parameters, including: Calculating an average circuit load according to the circuit load sequence; Obtaining a bypass performance predictor trained based on historical data of non-stop power switching, wherein the bypass performance predictor includes a circuit fluctuation prediction branch and a circuit heating prediction branch; Inputting the circuit load sequence into the bypass performance predictor, and predicting output to obtain a circuit fluctuation coefficient and a circuit heating parameter; Conduct circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain the circuit abnormality rate, and display it; According to the circuit abnormality rate, an energy loss prediction coefficient is configured, and energy loss is predicted according to the circuit heating parameters to obtain energy loss parameters. After the battery and meter are replaced, energy compensation processing is performed on the new meter according to the energy loss parameters, including: Configuring the circuit abnormality rate as a power loss prediction coefficient; Obtain power loss prediction branch group; Using the power loss prediction coefficient as a configuration ratio, selecting the power loss prediction branch of the configuration ratio, inputting the circuit heating parameter, predicting and outputting multiple branch power loss parameters, and calculating the average to obtain the power loss parameter; After the battery and meter are replaced, the new meter is compensated for the energy loss according to the energy loss parameters.

2. The control method for non-stop power switching and meter replacement according to claim 1, characterized in that: Obtain the circuit load sequence of the target circuit to be replaced, and analyze and obtain the circuit load variation coefficient, including: Performing a load test on a target circuit to be replaced to obtain a circuit load sequence, wherein the target circuit is equipped with an old meter before the replacement; A circuit load variation coefficient is obtained by calculation according to the circuit load sequence.

3. The control method for non-stop power switching and meter replacement according to claim 1, characterized in that: The training step of the bypass performance predictor comprises: Based on the historical data of non-stop power replacement, a sample circuit load set is collected, and the proportion of the number of circuit fluctuations caused by bypass under different sample circuit loads is collected, and the sample circuit fluctuation coefficient set is obtained by annotation; Collecting heating parameters of the bypass under different sample circuit loads to obtain a sample heating parameter set; Based on machine learning, we build circuit fluctuation prediction branches and circuit heating prediction branches. The sample circuit load set is used as input training data, and the sample circuit fluctuation coefficient set and the sample heating parameter set are used respectively to perform supervised training on the circuit fluctuation prediction branch and the circuit heating prediction branch until convergence, thereby obtaining a bypass performance predictor.

4. The control method for non-stop power switching and meter replacement according to claim 1, characterized in that: According to the circuit fluctuation coefficient and the circuit load variation coefficient, a circuit abnormality analysis is performed to obtain a circuit abnormality rate and display the result, including: Based on the historical data of non-stop power replacement, the sample circuit fluctuation coefficient set and the sample circuit load variation coefficient set are collected. The proportion of bypass abnormalities under different sample circuit fluctuation coefficients and sample circuit load variation coefficients is collected, and the sample circuit abnormality rate set is obtained by annotation. Based on a decision tree, a circuit anomaly classifier is constructed using the sample circuit fluctuation coefficient set, the sample circuit load variation coefficient set, and the sample circuit anomaly rate set; The circuit fluctuation coefficient and the circuit load variation coefficient are input into the circuit anomaly classifier, and the circuit anomaly rate is obtained by classification and displayed.

5. The control method for non-stop power switching and meter replacement according to claim 1, characterized in that: Obtain power loss prediction branch groups, including: Based on the historical data of non-stop power replacement, the bypass sample circuit heating parameter set is collected, and the power loss parameters during the battery replacement process under different circuit heating parameters are tested and obtained, and the sample power loss parameter set is marked; randomly dividing the sample data of a preset proportion within the sample circuit heating parameter set and the sample power loss parameter set with replacement, and training a first power loss prediction branch based on machine learning; Continue to train multiple power loss prediction branches to obtain a power loss prediction branch group.

6. A control system for power replacement and meter replacement without power outage, characterized in that: The system is used to execute the control method for non-stop power switching and meter replacement according to any one of claims 1 to 5, and the system comprises: The load sequence analysis module is used to obtain the circuit load sequence of the target circuit to be replaced and analyze and obtain the circuit load variation coefficient; A bypass parameter analysis module is used to perform circuit fluctuation analysis and circuit heating analysis according to the circuit load sequence after switching the target circuit to bypass, so as to obtain a circuit fluctuation coefficient and a circuit heating parameter; An abnormality rate calculation module is used to perform circuit abnormality analysis based on the circuit fluctuation coefficient and the circuit load variation coefficient, obtain the circuit abnormality rate, and display it; The electric energy compensation module is used to configure the electric energy loss prediction coefficient according to the circuit abnormality rate, predict the electric energy loss according to the circuit heating parameters, obtain the electric energy loss parameters, and after the battery and meter are replaced, perform electric energy compensation processing on the new meter according to the electric energy loss parameters.

7. A computer device, characterized in that: The device includes a processor and a memory: memory for storing computer programs; A processor is used to read and execute the computer program, thereby implementing the control method for non-stop power replacement and meter replacement as described in any one of claims 1-5.

Images