Intelligent power distribution cabinet control method and power distribution cabinet

Abstract

The application relates to the technical field of power distribution control, in particular to an intelligent power distribution cabinet control method and a power distribution cabinet, which comprises the following steps: collecting operation parameters in real time, performing hierarchical abnormality detection on each branch in the power distribution cabinet according to the obtained operation parameters, and setting abnormality characteristics corresponding to each branch; identifying combination components of the abnormality characteristics corresponding to each branch, judging each combination component of the abnormality characteristics, setting early warning parameters under the current abnormal power distribution, performing trend analysis according to the early warning parameters, identifying the relative change amount of the early warning parameters under the change trend, performing state deduction according to the relative change amount, and identifying the strategy control target under the change trend; and using the occurrence frequency of the strategy control target to quantize the trigger conditions of each branch, setting hierarchical control strategies based on the logical connection relationship of each branch under the trigger conditions. The response speed and control efficiency of the power distribution cabinet control are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution control, in particular to a smart power distribution cabinet control method and a power distribution cabinet. BACKGROUND

[0002] The power distribution cabinet is a device for connecting with the external power supply line to centrally supply power to the power supply line. The real-time monitoring and operation management of the entire power distribution system are realized by collecting all energy data of the power distribution cabinet. When controlling and analyzing the energy data of the power distribution cabinet, only monitoring single parameters such as voltage and current can easily cause problems such as isolated parameter analysis and poor data adaptability, resulting in insufficient action or control.

[0003] For example, Chinese Patent Application Publication No. CN116760197A discloses a monitoring method and device for a smart power distribution cabinet and a power distribution cabinet. The present application simulates the existing wiring scheme, obtains an existing power distribution model, and simulates a preset wiring scheme to obtain a preset power distribution model. Data analysis is performed on the preset power distribution model to determine whether the preset wiring scheme is reasonable, thereby realizing the ability to determine whether the preset wiring scheme meets safety standards before implementing the preset wiring scheme, and solving the problem of difficulty in predicting the risk after changing the wiring scheme of the power distribution cabinet in the prior art.

[0004] For example, Chinese Patent Application Publication No. CN118983936A discloses a monitoring method and device for a smart power distribution cabinet and a power distribution cabinet, relating to the technical field of monitoring and analysis. It includes monitoring and collecting cabinet surface information and internal line information corresponding to the target power distribution cabinet, analyzing the cabinet surface information and internal line information corresponding to the target power distribution cabinet, and further confirming the equipment aging evaluation coefficient corresponding to the target power distribution cabinet. The running state information and electrical safety information corresponding to the target power distribution cabinet are obtained. The running state information corresponding to the target power distribution cabinet is analyzed to confirm the running state evaluation coefficient corresponding to the target power distribution cabinet, and the electrical safety information corresponding to the target power distribution cabinet is analyzed to confirm the electrical safety evaluation coefficient corresponding to the target power distribution cabinet. The comprehensive evaluation coefficient corresponding to the target power distribution cabinet is confirmed, and adjustment is made based on the comprehensive evaluation coefficient corresponding to the target power distribution cabinet.

[0005] The prior art describes the processing methods for the wiring of the power distribution cabinet and the surface aging, respectively, but ignores the actual running state of the power distribution cabinet connecting branches and the strategy control method, resulting in the power distribution cabinet being unable to timely perform early warning processing according to the connection equipment when identifying abnormal parameters such as current and voltage of the connection equipment, and causing the power distribution cabinet to still have problems of delayed risk identification and excessively long processing response when controlling the equipment connection. SUMMARY

[0006] In order to solve the above technical problems, the technical scheme adopted by the present application is: a smart power distribution cabinet control method, comprising: S1, using the multiple sensors currently deployed in the power distribution cabinet, real-time collection of operation parameters, the operation parameters including but not limited to the voltage, current of each branch, the state of each branch switch and the operation time of each branch switch.

[0007] S2, according to the obtained operation parameters, layered anomaly detection is performed on each branch in the power distribution cabinet, and the abnormal characteristics corresponding to each branch are set.

[0008] S3, identifying the combination components of the abnormal characteristics corresponding to each branch, determining each combination component of the abnormal characteristics, and setting the warning parameters under the current abnormal power distribution.

[0009] S4, according to the trend analysis of the warning parameters, identifying the relative change amount of the warning parameters under the change trend, and performing state deduction with the relative change amount to identify the strategy control target under the change trend.

[0010] S5, using the occurrence frequency of the strategy control target, quantifying the trigger conditions of each branch, and setting the layered control strategy based on the logical connection relationship of each branch under the trigger condition.

[0011] A smart power distribution cabinet, comprising: a computer readable storage medium, the computer readable storage medium stores instructions, when the instructions run on the computer, make the computer execute any of the above-mentioned smart power distribution cabinet control method.

[0012] The beneficial effects of the present application are: first, the present application detects abnormal characteristics by layered anomaly detection, which detects abnormal characteristics based on physical layering and logical layering of each operation parameter, so that abnormal characteristics can contain multi-dimensional data information, avoid the reduction of current data analysis accuracy due to single parameter and small dimension, and provide data basis for subsequent data analysis.

[0013] Second, the present application uses regular expression to extract multi-parameter conditional dependency relationship based on association rules and support difference, quantifies abnormal level based on support difference, regards the combination form of abnormal level of current operation parameters as output abnormal characteristics, uses different abnormal characteristic combinations, regards the difference part as warning parameters, to improve the response rate of operation parameter specificity in some scenes, and improve the response dimension of abnormal characteristic alarm processing.

[0014] Thirdly, the application classifies the early warning parameters according to trends, simulates the parameter changes on each branch of the current power distribution cabinet, adopts a joint control strategy for the intersection existing in the same state classification, adopts a difference strategy for the different parts, avoids the over-control situation caused by the configuration and processing strategy of each branch, then dynamically adjusts the threshold according to the frequency of the strategy control target, and finally realizes the priority control of each branch after combining the trigger sequence. BRIEF DESCRIPTION OF DRAWINGS

[0015] The application will be further described below in combination with the drawings and examples.

[0016] Figure 1 It is a flowchart of a smart power distribution cabinet control method.

[0017] Figure 2 It is a flowchart of step S2 of a smart power distribution cabinet control method.

[0018] Figure 3 It is a flowchart of step S3 of a smart power distribution cabinet control method.

[0019] Figure 4 It is a flowchart of step S4 of a smart power distribution cabinet control method.

[0020] Figure 5 It is a flowchart of step S5 of a smart power distribution cabinet control method. DETAILED DESCRIPTION

[0021] The embodiments of the application will be described in detail below. The embodiments described below are exemplary and are only used to explain the application, and cannot be understood as a limitation of the application. If the specific technology or condition is not indicated in the embodiments, the technology or condition described in the literature in the art or according to the product instruction is used.

[0022] Reference Figure 1 A smart power distribution cabinet control method, comprising: S1, using a plurality of sensors deployed in a current power distribution cabinet, collecting running parameters in real time, the running parameters including but not limited to the voltage and current of each branch, the state of each branch switch and the running time of each branch switch.

[0023] S2, according to the obtained running parameters, performing layered abnormality detection on each branch in the power distribution cabinet, and setting the abnormality characteristics corresponding to each branch.

[0024] S3, identifying the combination components of the abnormality characteristics corresponding to each branch, determining each combination component of the abnormality characteristics, and setting the early warning parameters under the current abnormal power distribution.

[0025] S4. Perform trend analysis based on the warning parameters, identify the relative changes of the warning parameters under the changing trend, and use the relative changes to perform state deduction to identify the strategy control target under the changing trend.

[0026] S5 utilizes the frequency of target occurrence to quantify the triggering conditions of each branch, and sets a hierarchical control strategy based on the logical connection relationship of each branch under the triggering conditions.

[0027] At this point, the current and voltage data collected from each branch are acquired, and the opening and closing status of circuit breakers and contactors, as well as the corresponding operating duration of each branch switch, are used as the main operating parameters for identification. Current sensors, voltage sensors, cameras, and other devices are used to directly acquire these operating parameters.

[0028] In step S1, when the operating parameters are obtained, the current operating parameters need to be filled with the baseline data of the operating parameters.

[0029] The implementation of step S1 also includes: S11, sensing the operating parameters under multiple operating cycles and obtaining the value range of each operating parameter within a continuous time window; at this time, the operating cycle is described as the data under the cycles of periodic inspection, control command execution, and periodic data archiving of the distribution cabinet. This data is divided into multiple combinations and viewed in the form of continuous time windows, and the value range of the operating parameters within these operating cycles is described. For example, the value range of current and voltage, whether each branch switch is normally opened and closed, and the running time of each branch configured in the distribution cabinet under the corresponding operating cycle, etc. This data is processed separately to form data sets under multiple scenarios, which is convenient for subsequent classification and control.

[0030] S12 distributes the operating parameters to multiple data sets according to the operating scenario based on the value range of each operating parameter, and outputs the operating parameters according to the time length of the multiple data sets.

[0031] The time length of the aforementioned data set represents the time period represented by the difference between the maximum and minimum timestamps of the data in the current data set. It indicates the relative runtime of each data set in the corresponding operating scenario, and helps to determine the operating status of the equipment connected to the power distribution cabinet.

[0032] Preferably, the aforementioned hierarchical anomaly detection utilizes a three-phase power quality analyzer deployed in the distribution cabinet and a wireless temperature measurement system for the terminal circuit, forming a physical hierarchical architecture of high voltage-low voltage-terminal. Different processing methods are applied to branches under different hierarchical architectures. For example, for branches connected to the distribution cabinet, such as high-voltage branches, insulation resistance monitoring is used to record their current and voltage, while low-voltage branches are monitored for current overload. The physical hierarchy is divided using the equipment type and the actual description of each branch on the distribution cabinet. The physical hierarchy represents the different voltage and current configurations on each branch when different devices are connected. These conditions are divided into multiple physical levels according to the high-voltage and low-voltage classification standards and the corresponding branch deployment and operation status. For example, the labels on the branches connected to the distribution cabinet that describe long-term operation, intermittent operation, paused operation, and waiting for inspection are used as their operation status. The description of branches belonging to high voltage or low voltage is combined with their operation status labels to form multiple physical hierarchical descriptions, thereby determining the relative operation status of each branch.

[0033] Simultaneously, logical layering can be used to achieve layered anomaly detection for multiple branches of the power distribution cabinet. For example, based on the current usage scenario of the power distribution cabinet and the parts connected to the power distribution cabinet, abnormal data can be classified according to clustering algorithms such as K-Means, and the risk level can be quantified by combining the operating status of the power distribution cabinet.

[0034] The hierarchical control strategy implemented at this point consists of three levels. The first level is local rapid response, which involves real-time monitoring of branch switch status and immediate circuit breaker tripping when current exceeds limits. The second level emphasizes the logical relationships between multiple branches, adjusting the load and operation of each branch connected to the distribution cabinet. The third level involves long-term series analysis in segments, using trend analysis to perform peak-shaving control and ensure that the configured current of each branch can be stably output.

[0035] like Figure 2 As shown, the implementation method of step S2 includes: S21, using the numerical range of the operating parameters belonging to low pressure and high pressure, and combining the operating status of the operating parameters, to obtain the physical layer corresponding to the current operating parameters.

[0036] S22, the operating parameters under the physical layer are used as the input features of the logical layer, and the branches in the distribution cabinet are clustered. The deviation values ​​of the operating parameters after clustering are used to set the logical layer of each operating parameter. The abnormal level represented by each layer in the logical layer is used as the output abnormal feature.

[0037] Preferably, when performing the above-mentioned logical layering, the input voltage and current need to be normalized and then clustered. The state of each branch switch is used to indicate the closed or open state of the switch. The state of the branch switch is essentially determined by combining the state of the switch and the running time with the clustered current and voltage after completing the voltage and current clustering analysis. It is then checked whether the current switch has performed the corresponding disconnection process after the branch current abnormal alarm, and whether the corresponding switch has undergone relevant state changes when the current and voltage values ​​are too large. These data are combined as the abnormal features of the output.

[0038] The implementation of clustering in step S22 also includes: S221, using the current and voltage of each branch in the distribution cabinet as part of the clustering, comparing the current and voltage with historical baseline data to obtain multiple clusters.

[0039] S222, the standard deviation of each value in the cluster from the cluster center is regarded as the deviation value of the operating parameter, and the abnormality level is set for the current and voltage in the current operating parameters.

[0040] S223 combines each anomaly level with the state and operating time of the branch switch, and treats the combined data as the output anomaly characteristic.

[0041] Preferably, during the clustering process described above, the current and voltage are clustered according to the K-Means clustering method, and the standard deviation calculated for each cluster is compared with the standard deviation in the historical baseline data. At this time, the historical baseline data are the current and voltage values ​​during normal operation in the historical data. When the standard deviation calculated for the input current and voltage differs significantly from the historical baseline data, it indicates that an abnormality in the current and voltage is detected in the current distribution cabinet.

[0042] For example, data whose current standard deviation is less than the historical baseline standard deviation is considered to be at level one anomaly, which indicates normal operation. Data whose current standard deviation is 1-2 times the historical baseline standard deviation is considered to be at level two anomaly, which is considered to be data that needs to be closely monitored. Data whose current standard deviation is more than 2 times the historical baseline standard deviation is considered to be at level three anomaly, which represents a serious issue that needs to be addressed. This is how the anomaly levels for current and voltage are determined.

[0043] As for abnormal output characteristics, the abnormal levels of current and voltage will be combined with the state and operating time of the corresponding branch switch in sequence to obtain the abnormal output characteristics.

[0044] Preferably, the combined abnormal features in step S223 further include: using the confidence level of the branch switch's state and running time in historical data as a basic condition, taking the product of the confidence level of the branch switch's state and running time with the current switch's load rate as a state variable, performing clustering processing on the branch switch's state and running time, setting the abnormality level based on the standard deviation of the state variable relative to the standard deviation of the historical baseline data, and combining the abnormality levels contained in the current operating parameters to output the abnormal features.

[0045] It should be noted that the confidence level of the branch switch state and runtime is calculated by switching state → runtime, and the ratio of the number of itemsets combining switching state and runtime to the number of all itemsets containing the corresponding switching state is used to measure the probability that the branch switch contains the corresponding runtime when it is closed or open.

[0046] The aforementioned load rate can be obtained by calculating the power from the real-time input voltage and current and dividing it by the rated power to obtain the state quantity corresponding to the current branch switch. When the branch switch is closed, the current operating status of the switch can be directly described. At this time, the confidence level of the branch switch's state and operating time is expressed in the form of instantaneous values ​​to describe its state quantity. When the switch is open, the state quantity will be represented as 0. At this time, its operating status can still be directly described. When the switch is open, it will not be compared with the historical baseline data. Only when the state quantity is non-zero will the anomaly level be calculated with the historical baseline data. The anomaly level is handled in the same way as the voltage and current.

[0047] It should be noted that when using K-Means clustering, the number of clusters K to be selected needs to be determined. At this time, only when the sum of squared errors relative to the cluster center after statistical analysis of multiple clusters is the smallest, and the silhouette coefficient is greater than 0.5, the corresponding number of clusters is regarded as the clusters for the current clustering process. At the same time, historical baseline data can also be calculated based on data from multiple time windows adjacent to the current input data to achieve real-time updates of historical baseline data.

[0048] Preferably, since the runtime represents the operating time of the equipment in a specific cycle, the confidence score is obtained by using the state of the branch switch and the runtime to quantify the state correlation of the equipment under continuous operation. It is necessary to combine multiple parameters such as load rate to explain the length of time the equipment operates in a single operating cycle in order to determine whether the current distribution cabinet can operate stably.

[0049] Specifically, the current data collection and processing cycle is based on days and hours, with each cycle divided into time lengths of 1 day, 3 days, 7 days, and 15 days, and the data collected every hour is used as the content of the current analysis.

[0050] In one embodiment of the present invention, in step S3, it is necessary to combine and determine the operating parameters after classifying the abnormality level, and check whether the corresponding abnormal features are interdependent parts, so as to obtain the obvious abnormal feature combination under the feature combination, and output these feature combinations as early warning parameters, representing the parts that the current power distribution cabinet needs to control and process.

[0051] like Figure 3 As shown, the implementation of step S3 includes: S31, analyzing the conditional dependencies of each component in the abnormal features based on the feature combination of the abnormal features, and collecting the combination judgment information of each abnormal feature.

[0052] S32, based on the conditional dependencies associated with the combined judgment information, obtain the processing record of the abnormal features by combining the data in the abnormal features in the order of the input of the running parameters.

[0053] S33 utilizes the processing record of abnormal features to compare each abnormal feature, and regards the difference subset of each component in the abnormal feature as the output warning parameter.

[0054] Preferably, the combined components are used to describe the values ​​of voltage, current, switch status, and runtime in each abnormal feature, indicating the data set corresponding to the current abnormal feature. The conditional dependencies of the above-mentioned combined components are used to describe the situation where voltage, current, the status of each branch switch, and the runtime corresponding to each branch switch trigger an early warning after combination. At this time, it is necessary to calculate the similarity between the data under multiple abnormal levels contained in the abnormal feature and the data that meets the early warning in the historical data. For example, using the calculation method of cosine similarity or Pearson correlation coefficient, after normalizing the data after combining multiple abnormal levels, the three abnormal levels are combined pairwise according to the abnormal level corresponding to voltage, the abnormal level corresponding to current, the abnormal level corresponding to the status of each branch switch, and the abnormal level corresponding to the runtime of each branch switch. The Pearson correlation coefficient is used to calculate the similarity after the combination of the corresponding abnormal levels, and these similarities are used as the identification method of the conditional dependencies under their combination. If cosine similarity is used to calculate the similarity, the data under the combination of the three abnormal levels is converted into vector form, and the similarity of these data is directly calculated with the early warning part of the historical data. When the calculated similarity is greater than 0.6, the combined components of the corresponding anomalous features are considered to have conditional dependencies. The difference subset of the combined components in multiple anomalous features is considered as the output warning parameter. This warning parameter is used to describe the different parts of multiple anomalous features in the data set of voltage, current, switching state and running time combinations. These parts may highlight potential anomalies.

[0055] The core purpose of combining judgment information is to identify conditional dependencies with statistical significance or potential risks by quantifying the similarity between the current combination of abnormal features and historical early warning data, thereby providing a dynamic and interpretable basis for early warning decisions.

[0056] As for the processing record, it is formed by combining the abnormal features of the input parameter sequence into a set of association rule data such as voltage → current → switch state → runtime. This part of the data is combined with multiple sets of data when using conditional dependency to judge its combination judgment information to record multiple sets of data that meet the conditions of data normalization → similarity calculation → early warning triggering.

[0057] When comparing features in the processing records and identifying differences in anomalous features, the process records are further examined to determine how the anomalous features trigger warnings. At this point, the Apriori algorithm can be used to analyze the support of multiple conditionally dependent operating parameters within the processing records. This support is then used to describe the differences in the current parameters under warning conditions. The support will determine the support of multiple combinations of anomalous features in historical data related to warnings, indicating the likelihood of the current operating parameter values ​​approaching a warning. The difference in support between multiple combinations of anomalous levels within the processing records is considered the feature difference of the anomalous features at this point, and the corresponding data is output as the warning parameters.

[0058] It should be noted that the current support is based on the ratio of the number of itemsets of voltage, current, switch state, and runtime combinations under the corresponding values ​​in the association rule to the number of itemsets of all data combinations, which is used to illustrate the probability of voltage, current, switch state, and runtime occurring under the corresponding values.

[0059] The implementation of step S33 also includes: using the processing record of the current abnormal feature, viewing multiple combinations of conditionally dependent components under the current abnormal feature, setting association rules corresponding to each abnormal feature, and the association rules representing the combination relationship between multiple data with conditional dependencies. The association rules are set based on regular expressions, indicating the form of other parameters when a certain value in the running parameters changes, to explain the main associated parts in the current abnormal feature; for example, association rules between voltage and current, association rules between switch state and running time, and association rules for combinations of voltage, current and switch state, etc. The association rules are usually expressed in the form of X→Y, indicating the combination of voltage, current, the state of each branch switch and the running time corresponding to each branch switch in the abnormal feature.

[0060] Calculate the support of each abnormal feature under the corresponding association rule. Generate a feature difference record of the abnormal feature by the difference between the support of each abnormal feature and the average support. Synchronize all feature difference records to the early warning parameters and determine that each early warning parameter corresponds to the abnormality level divided by the operating parameters.

[0061] The purpose of marking feature differences is to subsequently examine which data in the input operating parameters exhibit conditional correlations, and to identify data combinations with significantly higher differences than other combinations, thus indicating data segments prone to high risk. Support difference is calculated by averaging the correlation degrees of multiple currently processed anomaly features after calculating the correlation degrees of the association rules corresponding to the anomaly features. The difference between the support degree of each association rule and the average value is compared, and processing is carried out sequentially according to this difference. It is emphasized that the parts with large support difference values ​​are processed first to quickly locate the anomalies in each branch within the current distribution cabinet.

[0062] The purpose of adapting support when obtaining the difference subset is to further describe the probability of the occurrence of the itemset of the combination of voltage, current, switching state and running time in the historical data. This probability of occurrence is combined with the difference subset of each current combination to indicate the set of values ​​of the operating parameters that are different when a warning is issued.

[0063] In one embodiment of the present invention, such as Figure 4 As shown, the implementation of step S4 includes: S41, converting the warning parameters into time series data, arranging the warning parameters according to timestamps, and calculating the relative change of each data in the warning parameters.

[0064] S42, based on the changing trend of the relative change, identify the state category corresponding to the relative change, and construct the state transition diagram corresponding to the warning parameter based on the warning parameter associated under each state category.

[0065] S43. Using the warning parameters in the state transition diagram, perform policy matching on each warning parameter. After policy matching, for warning parameters belonging to the same state category, determine the intersection of these warning parameters as the policy control target.

[0066] S44. For warning parameters that do not belong to the same state category, the difference between each warning parameter after policy matching shall be regarded as the policy control target.

[0067] It should be noted that the state transition diagram will classify the state according to the relative change of each warning parameter, connect and display multiple warning parameters according to the timestamp of the warning parameters, and label the state category on each warning parameter, and then combine them to form the state transition diagram corresponding to each branch on the distribution cabinet.

[0068] The aforementioned relative changes will categorize the warning parameters into three states: stable, deteriorating, and recovering. These three states represent the changing trends of the relative changes. For the relative changes in voltage and current, the stable state is [-10%, 10%], the deteriorating state is greater than 30% and continues to rise, and the recovering state is less than -30% and returns to the normal range. The identification of switch status and running time is based on the corresponding state quantities, and the non-zero parts are detected. The method used can be the same as that used for voltage and current to set the corresponding state. As for the numerical range not included between the stable, deteriorating, and recovering states, it represents the relative fluctuation state. This part of the data is not used as the part for direct state classification. After setting its label using rising fluctuation and falling fluctuation, its label is used as the corresponding state classification to divide the state categories corresponding to stable, deteriorating, recovering, rising fluctuation, and falling fluctuation.

[0069] Then, using multiple sets of early warning parameters corresponding to stable, deteriorating, recovering, rising fluctuations, and falling fluctuations, the early warning parameters on the corresponding branches are connected by the timestamps of the early warning parameters to form a state transition diagram. For each state in the state transition diagram, a preset strategy is configured, such as current limiting, alarm, and power restoration. When matching strategies, the cosine similarity is calculated between the early warning parameters under each state category and the parameters marked in the preset strategies. The preset strategy corresponding to the maximum value after the cosine similarity calculation is used as the strategy for the corresponding early warning parameter. When calculating the cosine similarity, the early warning parameters need to be normalized and the similarity calculation is completed in the form of vectors. As for selecting the intersection and difference parts, it is biased towards dynamically identifying the similarities and differences in the strategies configured by each branch after the early warning parameters are identified, in order to determine the strategy to be adopted by different branches under the strategy control, and to prevent the limitation of a single strategy from causing the state judgment of each branch to be incorrect. The output strategy control targets will further calculate the intersection and difference sets of the operating parameters included in these preset strategies. These data will be used as the output data during strategy matching analysis to emphasize the intersection part that is centrally reviewed and the difference part that is auxiliaryly reviewed during strategy setting. This will complete the selection of strategy control targets under multiple state classifications. The output strategy control targets will include the specific values ​​of operating parameters, anomaly levels, association rules, etc. The association between these data will be output as strategy control targets. When setting trigger conditions for strategy control targets, the trigger conditions can directly correspond to the anomaly levels and association rules. When executing strategies, the frequency of each operating parameter reaching the anomaly level under the warning can be used to indicate the strategy control targets that need to be focused on, as well as the data that are not effective targets and are processed using difference analysis.

[0070] Preferably, the correlation described above represents the form of high voltage rise → switch disconnection, indicating that the switch state and operating time are directly related to current and voltage. Alternatively, the correlation can be based on the conditional dependencies between multiple sets of warning parameters, illustrating the logical direct correlation between them. This part can be directly extracted using regular expressions set in the database. That is, after filtering warning parameters through correlation rules, their state classifications and the corresponding parts of the correlation rules are considered as multiple combined categories. The intersection of parameters within the same category after the combination of correlation rules and state classifications is used as the primary strategy target. Parameters not within the same category emphasize the differences to determine the differences during strategy execution and adjust the strategies configured for each branch. For example, the intersection is considered the core strategy control target, processed using methods such as simultaneously limiting voltage and current; the differences are considered auxiliary strategy control targets, processed using methods such as adjusting only voltage without interfering with the switch.

[0071] In one embodiment of the present invention, such as Figure 5 As shown, the implementation of step S5 also includes: S51, when the strategy control target is a valid target, the frequency of occurrence of the strategy control target within a preset time period is used as the frequency threshold of the strategy control target, and the valid target is data that meets the triggering conditions.

[0072] S52, when the policy control target is not a valid target, data greater than the frequency threshold is regarded as the part that meets the frequency threshold, and the policy control target that meets the frequency threshold is used as the trigger condition for each branch.

[0073] S53 treats the order in which each branch is triggered when the triggering condition is met as a logical connection relationship, sets the strategy for each branch in the order of triggering, and outputs the set strategies one by one according to the abnormality level of the triggering condition to obtain the hierarchical control strategy.

[0074] In step S5, policy control targets that have been verified in historical data to accurately trigger or handle early warnings are considered valid targets. These policy control targets represent effective handling methods in the control of the distribution cabinet. For example, if the policy control target under the intersection is voltage + current dual limit, and this target triggers 5 over-limits per unit time in the current data, it will mean that voltage + current needs to be warned and policy control needs to be set. At this time, the corresponding policy control target is considered a valid target and relevant thresholds are set. The part that is not a valid target is the difference in the policy control target when it is acquired. If the frequency of the corresponding data exceeds the frequency of obvious early warning under the current frequency threshold, it may represent a potential early warning event in the current scenario. It is necessary to determine whether the corresponding policy control target is a potential valid target.

[0075] As for setting trigger conditions, it is based on the logic of whether subsequent settings are for potential valid targets. The triggering situations of each branch are processed according to the triggering order. According to the triggering order of the intersection and difference parts in the strategy control target, the current set strategy method is set one by one. The intersection part is regarded as high priority and the difference part is regarded as low priority. If there are still problems after the strategy processing of the intersection part is completed, the strategy corresponding to the difference part is used to complete the deployment of the current hierarchical control strategy. The abnormal level of the original running parameters after deployment is mapped to explain the overall execution process of the current strategy.

[0076] This invention also provides an intelligent power distribution cabinet, comprising: a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to execute an intelligent power distribution cabinet control method as described above, achieving the functions described below: real-time acquisition of operating parameters using various sensors deployed in the current power distribution cabinet, including but not limited to voltage and current of each branch, status of each branch switch, and operating time corresponding to each branch switch; hierarchical anomaly detection of each branch within the power distribution cabinet based on the acquired operating parameters, setting anomaly characteristics corresponding to each branch; identification of the combined components of the anomaly characteristics corresponding to each branch, judgment of each combined component of the anomaly characteristics, and setting early warning parameters under the current abnormal power distribution; trend analysis based on the early warning parameters, identification of the relative change of the early warning parameters under the changing trend, and state deduction based on the relative change to identify the strategy control target under the changing trend; quantification of the triggering conditions of each branch using the frequency of occurrence of the strategy control target, and setting a hierarchical control strategy based on the logical connection relationship of each branch under the triggering conditions; and finally, completion of the control processing of the power distribution cabinet.

[0077] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A method for controlling an intelligent power distribution cabinet, characterized in that, include: S1 uses various sensors deployed in the current distribution cabinet to collect operating parameters in real time. The operating parameters include the voltage and current of each branch, the status of each branch switch, and the running time of each branch switch. S2, based on the acquired operating parameters, perform layered anomaly detection on each branch in the distribution cabinet and set the corresponding anomaly characteristics for each branch; The implementation of step S2 includes: S21, using the numerical range of operating parameters belonging to low voltage and high voltage, and combining the operating status of the operating parameters, obtaining the physical layer corresponding to the current operating parameters; S22, using the operating parameters under the physical layer as the input features of the logical layer, performing clustering processing on each branch in the distribution cabinet, and setting the logical layer of each operating parameter based on the deviation value of the operating parameters after clustering processing; and using the abnormal level represented by each layer in the logical layer as the output abnormal feature. S3, identify the combination components of abnormal features corresponding to each branch, determine each combination component of abnormal features, and set the early warning parameters under the current abnormal power distribution. S4. Perform trend analysis based on the warning parameters, identify the relative change of the warning parameters under the changing trend, and use the relative change to perform state deduction to identify the strategy control target under the changing trend. S5 utilizes the frequency of target occurrence to quantify the triggering conditions of each branch, and sets a hierarchical control strategy based on the logical connection relationship of each branch under the triggering conditions.

2. The intelligent power distribution cabinet control method according to claim 1, characterized in that, The implementation of step S1 also includes: S11, perform data sensing on the operating parameters under multiple operating cycles, and obtain the value range of each operating parameter under a continuous time window; S12 distributes the operating parameters to multiple data sets according to the operating scenario based on the value range of each operating parameter, and outputs the operating parameters according to the time length of the multiple data sets.

3. The intelligent power distribution cabinet control method according to claim 1, characterized in that, The implementation methods for clustering in step S22 also include: S221 uses the current and voltage of each branch in the distribution cabinet as part of the clustering, compares the current and voltage with historical baseline data respectively, and obtains multiple clusters; S222, the standard deviation of each value in the cluster from the cluster center is regarded as the deviation value of the operating parameter, and the abnormality level is set for the current and voltage in the current operating parameters; S223 combines each anomaly level with the state and operating time of the branch switch, and treats the combined data as the output anomaly characteristic.

4. The intelligent power distribution cabinet control method according to claim 3, characterized in that, The combined abnormal features in step S223 also include: Using the confidence level of the branch switch's status and runtime in historical data as the basic condition, the product of the confidence level of the branch switch's status and runtime and the current switch's load rate is regarded as the status variable. The status and runtime of the branch switch are clustered, and the anomaly level is set by the standard deviation of the status variable relative to the standard deviation of the historical baseline data. The anomaly levels contained in the current operating parameters are combined and output as anomaly features.

5. The intelligent power distribution cabinet control method according to claim 1, characterized in that, Step S3 can be implemented in the following ways: S31, Based on the feature combination of abnormal features, analyze the conditional dependence of each combination component in the abnormal features, and collect the combination judgment information of each abnormal feature. S32, based on the conditional dependencies associated with the combined judgment information, obtain the processing record of the abnormal features by combining the data in the abnormal features according to the order of the input of the running parameters; S33 utilizes the processing record of abnormal features to compare each abnormal feature, and regards the difference subset of each component in the abnormal feature as the output warning parameter.

6. The intelligent power distribution cabinet control method according to claim 5, characterized in that, The implementation of step S33 also includes: Based on the processing record of the current abnormal feature, view the multiple combined components that have conditional dependencies under the current abnormal feature, and set the association rules corresponding to each abnormal feature; Calculate the support of each abnormal feature under the corresponding association rule. Generate a feature difference record of the abnormal feature by the difference between the support of each abnormal feature and the average support. Synchronize all feature difference records to the early warning parameters and determine that each early warning parameter corresponds to the abnormality level divided by the operating parameters.

7. The intelligent power distribution cabinet control method according to claim 1, characterized in that, The implementation methods of step S4 include: S41, convert the warning parameters into time series data, arrange the warning parameters according to the timestamp, and calculate the relative change of each data in the warning parameters; S42, based on the changing trend of the relative change, identify the state classification corresponding to the relative change, and construct the state transition diagram corresponding to the warning parameter based on the warning parameter associated under each state classification; S43, using the warning parameters in the state transition diagram, perform policy matching on each warning parameter. After policy matching, for warning parameters belonging to the same state category, determine the intersection of these warning parameters as the policy control target. S44. For warning parameters that do not belong to the same state category, the difference between each warning parameter after policy matching shall be regarded as the policy control target.

8. The intelligent power distribution cabinet control method according to claim 1, characterized in that, The implementation of step S5 also includes: S51, when the strategy control target is a valid target, the frequency of the strategy control target in the preset time period is used as the frequency threshold of the strategy control target, and the valid target is the data that meets the triggering condition. S52, when the policy control target is not a valid target, data greater than the frequency threshold is regarded as the part that meets the frequency threshold, and the policy control target that meets the frequency threshold is used as the trigger condition for each branch. S53 treats the order in which each branch is triggered when the triggering condition is met as a logical connection relationship, sets the strategy for each branch in the order of triggering, and outputs the set strategies one by one according to the abnormality level of the triggering condition to obtain the hierarchical control strategy.

9. An intelligent power distribution cabinet, characterized in that, include: A computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform a smart distribution cabinet control method as described in any one of claims 1 to 8.

Images