Low-voltage bus duct safe power supply system and method

Abstract

The application belongs to the technical field of cable safety management, and specifically discloses a low-voltage bus duct safety power supply system and method, which comprises: a reference threshold generation module, which generates a reference temperature rise safety threshold by performing cluster analysis on joint temperature rise, load current and environmental temperature data during historical safe operation; a threshold dynamic adjustment module, which performs dynamic interpolation adjustment on the reference temperature rise safety threshold based on real-time load and environmental temperature, and generates multiple candidate safety thresholds; a threshold evaluation and analysis module, which performs backtracking test on each candidate safety threshold, and calculates its prediction accuracy index and safety margin index; a threshold screening and determination module, which screens an optimal target safety threshold based on the above indexes; and a safety monitoring execution terminal, which performs real-time monitoring and early warning by using the target threshold. The application realizes adaptive optimization of the temperature rise safety threshold, and improves the accuracy of temperature rise monitoring and early warning and the safety of the power supply system.

Description

Technical Field

[0001] This invention belongs to the field of cable safety management technology, and more specifically, relates to a low-voltage busbar trunking safety power supply system and method. Background Technology

[0002] Low-voltage busbar trunking consists of a series of conductors of standard length, insulating materials, metal shells, connectors, and supports. As the main power supply artery, the safety of the busbar trunking is directly related to the stability of the entire power supply system.

[0003] In actual operation, the resistance heating effect of busbar conductors and joints varies significantly under different load conditions, causing their normal temperature rise range to change dynamically.

[0004] Currently, temperature rise monitoring typically relies on fixed temperature rise thresholds to determine whether busbar trunking is at risk of overheating. However, this method has the following limitations:

[0005] Under light load or favorable conditions, the temperature rise of the busbar joint body is relatively low. If a fixed high threshold set based on the worst operating conditions is used, the normal state at this time cannot be identified. It may generate alarms for harmless but slightly high temperature fluctuations, causing the true alarm to be ignored.

[0006] Under harsh operating conditions such as heavy loads, high temperatures, or severe harmonics, the temperature rise of the joint may rapidly approach but not yet exceed a fixed threshold. Although this does not trigger an early warning, the equipment is actually in a high-risk state. If the temperature rise continues to accumulate, it may cause joint aging, insulation deterioration, or even short-circuit faults, resulting in untimely early warnings and an inability to effectively prevent power grid accidents.

[0007] In summary, existing fixed threshold temperature rise monitoring methods are difficult to adapt to the complex and ever-changing operating conditions of busbar trunking. They are not only prone to false alarms under normal operating conditions, but also pose a risk of delayed or missed warnings under abnormal operating conditions, affecting the safety and operational effectiveness of the power supply system. Summary of the Invention

[0008] In view of this, in order to solve the above problems, a safe power supply system and method for low-voltage busbar trunking is proposed.

[0009] The objective of this invention can be achieved through the following technical solution: This invention provides a safe power supply system for low-voltage busbar trunking, which includes: a reference threshold generation module, which performs cluster analysis on the joint temperature rise data, load current data and ambient temperature data of the low-voltage busbar trunking during historical safe operation to generate a reference temperature rise safety threshold.

[0010] The threshold dynamic adjustment module dynamically adjusts the reference temperature rise safety threshold based on the real-time load current and real-time ambient temperature of the bus trunking, generating multiple candidate safety thresholds.

[0011] The threshold assessment and analysis module retrieves fault logs and performs backtesting tests on each candidate safety threshold. Based on the test results, it calculates prediction accuracy and safety margin indicators.

[0012] The threshold selection and determination module selects the optimal candidate safety threshold as the target safety threshold based on the prediction accuracy index and the safety margin index.

[0013] The safety monitoring execution terminal performs real-time safety monitoring and early warning of low-voltage busbar trunking based on target safety thresholds.

[0014] The present invention also provides a method for safe power supply of low-voltage busbar trunking, the method comprising: acquiring joint temperature rise data, load current data and ambient temperature data of low-voltage busbar trunking during historical safe operation, and generating a benchmark temperature rise safety threshold through cluster analysis.

[0015] The load current and ambient temperature of the current busbar trunking are collected in real time, and the reference temperature rise safety threshold is dynamically adjusted accordingly to generate multiple candidate safety thresholds.

[0016] Retrieve fault logs, perform backtracking tests on each candidate safety threshold based on the fault logs, and calculate prediction accuracy and safety margin indicators based on the test results.

[0017] The optimal candidate safety threshold is selected as the target safety threshold based on the prediction accuracy index and the safety margin index.

[0018] Real-time safety monitoring and early warning of low-voltage busbar trunking are achieved by using target safety thresholds.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention identifies and quantifies typical operating conditions and their corresponding normal temperature rise levels under different load and ambient temperature combinations by cluster analysis of historical safe operation data, and establishes a dynamic benchmark that fits the actual operating state. This overcomes the problem that when only a single fixed threshold is used as the judgment standard, it is impossible to effectively distinguish between normal temperature fluctuations and real anomalies under light load or good environmental conditions due to the threshold setting being too high. At the same time, it also reduces the probability of false alarm events.

[0020] (2) This invention dynamically adjusts the reference temperature rise safety threshold based on real-time load current and real-time ambient temperature, so that the monitoring threshold can be tightened synchronously as operating conditions deteriorate. This solves the problem that under heavy load, high temperature and other harsh operating conditions, the fixed threshold is too high and cannot capture the approaching temperature rise risk in time, thereby improving the timeliness of early warning of potential overheating faults.

[0021] (3) This invention uses historical fault records to quantitatively verify and evaluate multiple dynamically generated candidate safety thresholds, thereby selecting the target safety threshold. This breaks through the limitations of a single fixed threshold or a dynamic threshold that has not been historically verified, which may lead to inaccurate warnings. This helps to reduce the blindness in threshold selection and strives to reduce the risk of false alarms while also reducing the omission of warnings due to improper threshold settings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0024] Figure 2 This is a schematic diagram of the overall implementation process of the present invention.

[0025] Figure 3 This is a schematic diagram of the process for determining the target safety threshold of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Please see Figure 1 As shown, the present invention provides a safe power supply system for low-voltage busbar trunking, which includes: a reference threshold generation module, a threshold dynamic adjustment module, a threshold evaluation and analysis module, a threshold screening and determination module, and a safety monitoring execution terminal.

[0028] In the above, the threshold dynamic adjustment module is connected to the benchmark threshold generation module and the threshold evaluation and analysis module, respectively, and the threshold screening and determination module is connected to the threshold evaluation and analysis module and the security monitoring execution terminal, respectively.

[0029] The modules of the low-voltage busbar trunking safety power supply system described in this invention can be implemented in software, hardware, firmware, or any combination thereof. For example, they can be integrated into the processor of the busbar trunking monitoring device, and the functions of each module can be implemented by executing stored program instructions. Figure 1 The connections shown represent the data flow and calling relationships between modules.

[0030] The benchmark threshold generation module is used to perform cluster analysis on the joint temperature rise data, load current data and ambient temperature data of the low-voltage bus trunking during historical safe operation to generate a benchmark temperature rise safety threshold.

[0031] Specifically, the baseline threshold generation module first connects to the low-voltage busbar data storage terminal and calls the archived data of the target busbar during its historical safe operation period (usually no less than one complete load cycle, such as one year). The historical safe operation period refers to the continuous operation period during which the low-voltage busbar has not recorded any fault alarms or protective power outages caused by excessive joint temperature rise.

[0032] The archived data includes, but is not limited to, connector temperature rise data, load current data, ambient temperature data, and maintenance event logs.

[0033] It should be added that the specific process for collecting the joint temperature rise data, load current data, and ambient temperature data is as follows:

[0034] Temperature sensors (such as infrared sensors or contact thermocouples) are deployed at the joints of the low-voltage busbar trunking to collect the measured joint temperature values ​​at a set sampling period. The sampling period can be set to 1 minute to 10 minutes, and the present invention preferably sets it to 10 minutes.

[0035] Simultaneously, ambient temperature values ​​are collected by ambient temperature sensors deployed in the busbar installation area, and load current values ​​are collected by current measuring devices (such as current transformers or smart meters) associated with the busbar power supply circuit.

[0036] The measured joint temperature value is subtracted from the corresponding timestamp of the ambient temperature value to eliminate the influence of ambient temperature on the joint temperature and obtain the joint temperature rise data.

[0037] Furthermore, the aforementioned joint temperature rise data, load current data, and ambient temperature data are all archived according to time, for use in the subsequent generation of the reference temperature rise safety threshold.

[0038] Since the temperature rise of the low-voltage busbar joints is determined by both the load current (heat generation) and the ambient temperature (heat dissipation conditions), its operating conditions are complex and varied. If a single fixed threshold is used, it cannot accurately reflect the reasonable differences in the safety boundary under different combinations of operating conditions (such as light load at normal temperature and heavy load at high temperature), thus leading to false alarms under light load and missed alarms under heavy load.

[0039] Based on this, the present invention performs cluster analysis on the joint temperature rise data, load current data, and ambient temperature data of low-voltage busbar trunking during historical safe operation, and generates a baseline temperature rise safety threshold accordingly. The specific cluster analysis process is as follows:

[0040] S1. Extract time-aligned connector temperature rise data, load current data, and ambient temperature data to construct a historical operation dataset.

[0041] S2. Considering that the temperature rise of the busbar joint is mainly determined by the Joule heat generated by the load current passing through it and the thermal balance between the heat dissipation of the joint to the environment, the ambient temperature directly affects the heat dissipation efficiency. Therefore, this invention uses the load current value and the ambient temperature value as feature components, and performs standardization processing on the load current and ambient temperature (such as Z-score standardization) to eliminate the influence of the different dimensions and numerical ranges of the two on the clustering results.

[0042] S3. For each historical moment in the historical operation dataset, combine its standardized load current value with the ambient temperature value to construct a two-dimensional feature vector, which represents the operating condition at that moment.

[0043] S4. Using the feature vectors of all historical moments as input, Euclidean distance is used as a metric to measure the similarity between vectors. The smaller the Euclidean distance, the more similar the combined operating conditions of load current and ambient temperature. Euclidean distance is a common existing similarity measurement method, and its specific formula will not be shown.

[0044] S5. The historical operating dataset is divided into multiple sets by a clustering algorithm, and each set constitutes a typical operating condition cluster.

[0045] S6. After completing the division of operating conditions, for each typical operating condition cluster, independently analyze the joint temperature rise data during the historical safe operation period within the cluster. Calculate the preset percentile value of all temperature rise data within the cluster as the initial safety threshold for that operating condition cluster. The preset percentile value is preferably the 95th percentile, and the implementer can adjust it between 90% and 99% according to the requirements for power supply safety level.

[0046] S7. Based on the frequency of occurrence of each typical operating condition cluster in historical data, that is, the proportion of the number of data points in each cluster to the total number of data points in the historical operating dataset, the initial safety threshold of all clusters is linearly weighted and summed, and the calculation result is output as the reference temperature rise safety threshold of the low-voltage bus trunking.

[0047] It should be noted that if the joint temperature rise data, load current data, and ambient temperature data during the historical safe operation period are insufficient, the equipment rated parameters or the fixed temperature rise threshold recommended by industry standards can be used as the reference temperature rise safety threshold.

[0048] By using the benchmark threshold generation module to perform cluster analysis on historical operating datasets, typical operating conditions under different combinations of load current and ambient temperature can be automatically identified and quantified, and a dynamic benchmark for the normal temperature rise level corresponding to each operating condition can be established.

[0049] Based on this dynamic benchmark, the present invention effectively overcomes the inherent defects of using a single fixed threshold. Specifically, under light load or favorable environmental conditions, it can make judgments based on an initial safety threshold that better suits the conditions, thereby improving the tolerance to harmless temperature fluctuations and avoiding frequent false alarms caused by excessively high threshold settings. Simultaneously, under harsh conditions such as heavy load or high temperature, this benchmark also provides a more accurate starting point for subsequent real-time adjustments, enhancing the sensitivity and timeliness of identifying real overheating risks from the source. This provides a data foundation for subsequent adaptive safety monitoring.

[0050] The following supplementary explanations are required during the specific execution of steps S1 to S7 above:

[0051] Clustering algorithms can be any one or a combination of common unsupervised learning algorithms such as K-means clustering, DBSCAN clustering, and hierarchical clustering to classify the working conditions of historical operating datasets.

[0052] Preferably, this invention uses the K-means clustering algorithm as an example to explain in detail the process of dividing the historical operating dataset. In a specific configuration, the initial number of clusters K can be preset to 3, thereby automatically dividing the historical operating conditions into three basic modes, such as light load conditions, rated load conditions, and heavy load conditions (here, load is a general concept that combines the effects of current and temperature).

[0053] For example, the specific partitioning rules of the K-means clustering algorithm are as follows: First, the algorithm uses Euclidean distance as a similarity measure to calculate the distance between the feature vector of each data point in the dataset and the current K cluster centers, and assigns the data point to the cluster to which the nearest cluster center belongs. Then, the cluster centers of each cluster are updated, with the new center being the mean of the feature vectors of all data points within that cluster. The above assignment and update steps are iterated repeatedly, forming an optimization process. The algorithm terminates when the moving distance of all cluster centers in the previous two iterations is less than a preset threshold (e.g., 0.01), or when the number of iterations reaches a preset upper limit (e.g., 200), indicating that the clustering results have converged. Finally, the algorithm outputs K mutually exclusive data sets, each defined as a typical operating condition cluster, representing an operating mode with a stable combination of load and temperature characteristics. For example, the final result might be three typical operating condition clusters: low temperature light load, normal temperature medium load, and high temperature heavy load.

[0054] The threshold dynamic adjustment module is used to dynamically adjust the reference temperature rise safety threshold based on the real-time load current and real-time ambient temperature of the bus trunking, and generate multiple candidate safety thresholds.

[0055] The real-time operating parameters are collected using the same method as those used to construct the historical operating dataset, and all data are synchronized with a unified timestamp.

[0056] Specifically, the dynamic adjustment of the reference temperature rise safety threshold includes: A1. Real-time acquisition of the load current value and ambient temperature value at the current moment, and standardization processing of the load current value and ambient temperature value (same as the aforementioned standardization processing method), and combining the standard processing results of the load current value and ambient temperature value into the operating point.

[0057] A2. Calculate the Euclidean distance between the real-time operating point and the cluster centers of each typical operating condition cluster generated in the baseline threshold generation module. The smaller the Euclidean distance, the more similar the real-time operating point is to the typical pattern of that cluster. Simultaneously, select the typical operating condition cluster with the smallest Euclidean distance as the target operating condition cluster, and determine one or more typical operating condition clusters with Euclidean distances second only to the target cluster as adjacent operating condition clusters. For example, in practical applications, the typical operating condition cluster with the second smallest Euclidean distance, or the two typical operating condition clusters with the second and third smallest Euclidean distances, can be selected as adjacent operating condition clusters.

[0058] A3. The initial safety threshold corresponding to the target operating condition cluster is taken as the first candidate safety threshold.

[0059] A4. Based on the Euclidean distance between the real-time operating point and the target operating condition cluster and at least one adjacent operating condition cluster cluster center, calculate the interpolation weights, where the interpolation weights are inversely proportional to the Euclidean distances. That is, set the initial weights of each cluster to be inversely proportional to their distances, and then obtain the final interpolation weights through normalization. The normalization can be achieved by dividing the initial weights of each cluster by the sum of the initial weights of all clusters.

[0060] A5. Obtain the initial safety thresholds for the target operating condition cluster and the selected adjacent operating condition clusters. Using the calculated normalized interpolation weights, perform a linear weighted summation of the initial safety thresholds to form one or more second candidate safety thresholds:

[0061] A6. The first candidate security threshold and the one or more second candidate security thresholds are combined to form a plurality of candidate security thresholds.

[0062] In actual operation, the actual operating conditions (load and environment) of busbar trunking are continuously changing, while the typical operating condition clusters obtained by clustering historical data are discrete representative points in this continuous space. If the threshold corresponding to the single operating condition cluster with the closest Euclidean distance is directly selected, when the real-time operating condition is in the boundary region between two typical clusters, its slight fluctuations will cause the selected threshold to repeatedly switch between two different fixed values. This may lead to unstable warning signal output, thereby interfering with subsequent operation and maintenance judgments.

[0063] To address this issue, the aforementioned dynamic adjustment implementation achieves a leap from static mapping to dynamic generation of safety thresholds. Furthermore, it not only identifies the target operating condition cluster that best matches the real-time operating conditions, but also generates a set of candidate safety thresholds by calculating the distances between the real-time operating point and the target cluster and the centers of adjacent clusters, and performing inverse distance weighted interpolation. This effectively eliminates threshold jumps caused by simple classification in the critical operating condition region, enabling the safety thresholds to be continuously adjusted according to changes in operating conditions.

[0064] Secondly, the generated multiple candidate thresholds provide a better decision space for the subsequent threshold selection and determination module, making the final selected target safety threshold more adaptable to the current actual operating state, thereby improving the early warning and adaptive capability.

[0065] It should also be noted that in practical applications, if the Euclidean distance between the real-time operating point and the cluster center of all typical operating condition clusters exceeds the preset maximum distance threshold (e.g., 1.5 times the maximum distance between all historical feature vectors), an alternative strategy can be enabled. For example, the maximum value of the initial safety threshold in all typical operating condition clusters can be used as the temporary target safety threshold, and an abnormal operating condition prompt can be issued to remind the operation and maintenance personnel to pay attention.

[0066] The threshold evaluation and analysis module is used to retrieve fault logs and perform backtesting tests on each candidate safety threshold through the fault logs. Based on the test results, it calculates the prediction accuracy index and the safety margin index.

[0067] Specifically, the backtracking test process is as follows: B1. Obtain the joint temperature rise sequence and the list of actual fault event trigger times from the operation and maintenance event logs for a continuous historical period after the most recent completion of the baseline threshold generation and adjustment and before the current real-time monitoring is started.

[0068] B2. For each candidate safety threshold, compare it point by point with the joint temperature rise time series. When the temperature rise value of a certain data point is greater than or equal to the current candidate safety threshold, mark an over-limit event. Merge consecutively over-limit data points to form a simulated early warning event. This event includes the start time (timestamp of the first over-limit) and the duration (length of consecutive over-limit).

[0069] B3. After traversing the entire joint temperature rise time sequence, a series of simulated early warning events corresponding to the current candidate safety threshold can be obtained, forming its simulated early warning event sequence.

[0070] B4. For each fault time point in the list of real fault events, search the simulated warning event sequence for all simulated warning events whose start time is earlier than that fault time point and whose time difference with the fault time point is less than a preset proximity time (e.g., 30 minutes). If found, select the simulated warning event with the earliest start time, determine it as the matching event for that fault time point, and record it as a successful matching event.

[0071] B5. For each candidate security threshold, perform the following sub-steps:

[0072] B51. Among the simulated early warning events generated by the system, those that are successfully matched are classified as true positive events, and the remaining events are classified as false positive events of the candidate safety threshold.

[0073] B52. Count the number of true positive events and the number of false positive events. Compare the number of true positive events with the total number of real failure events to obtain the true positive rate. Compare the number of false positive events with the total number of simulated early warning events generated by the corresponding candidate safety threshold to obtain the false positive rate.

[0074] The true positive rate measures the probability that a candidate safety threshold will successfully issue an alert when a real fault occurs. A higher true positive rate indicates higher sensitivity and fewer missed alerts. The false positive rate measures the probability that a candidate safety threshold will falsely issue an alert when no real fault occurs. A lower false positive rate indicates fewer false alarms.

[0075] B53. For each true positive event, calculate the time difference between the start time of the corresponding simulated early warning event and the corresponding fault time point to obtain the effective advance time. Calculate the average effective advance time of all true positive events corresponding to the candidate safety threshold, and use it as the average effective advance time of the true positive early warning for the candidate safety threshold.

[0076] B6. The true positive rate, false positive rate, and average effective lead time of true positive warning for each candidate safety threshold are used as the results of the backtest.

[0077] Considering that the candidate safety thresholds output by the dynamic threshold adjustment module each represent a different balance point between reducing false alarms and reducing missed alarms, relying solely on data statistics, such as using the mean, median, or mode, ignores the differences between the different candidate thresholds. In actual historical operation, the selected final target safety threshold will neither effectively warn of real faults nor generate a large number of invalid alarms.

[0078] Therefore, this invention quantifies the value of each candidate safety threshold by retrieving fault records and conducting backtesting tests, thereby transforming the value of each candidate safety threshold into quantifiable early warning performance indicators (such as prediction accuracy and safety margin). This ensures that subsequent selection of target safety thresholds is based on historical empirical evidence, selecting the safety thresholds that achieve the best balance between false alarms and missed alarms in real history. This overcomes the limitations of potentially inaccurate early warnings associated with single fixed thresholds or dynamic thresholds that have not been historically verified, thus helping to reduce the blindness in threshold selection and striving to reduce the risk of false alarms while also reducing early warning omissions caused by improper threshold settings.

[0079] Furthermore, the specific steps for calculating the prediction accuracy index and the safety margin index are as follows: Set a preset effective lead time threshold, which can be set according to the shortest time required for operation and maintenance response, usually from 5 minutes to 30 minutes, for example, 10 minutes.

[0080] The average effective lead time of each candidate safety threshold obtained from the backtracking test is compared with the effective lead time threshold. Only candidate safety thresholds with an average effective lead time greater than or equal to the preset threshold are marked as effective candidate safety thresholds.

[0081] For each selected valid candidate safety threshold, the false positive rate and true positive rate calculated in the backtesting test are extracted. Multiple coordinate points are plotted on a two-dimensional coordinate plane with the false positive rate on the x-axis and the true positive rate on the y-axis. All coordinate points are sorted and connected in ascending order of false positive rate to form a warning performance curve. Subsequently, the area under this warning performance curve is calculated using a numerical integration method (such as the trapezoidal rule), and the area value is used as an indicator of the prediction accuracy of the valid candidate safety threshold. The closer the area value is to 1, the stronger the ability of the valid candidate safety threshold to distinguish between normal and fault states.

[0082] Obtain the joint temperature rise time series from the verification data used for backtesting. Calculate the maximum temperature rise value (peak value) of this series throughout the entire verification period. For each valid candidate safety threshold, calculate the difference between this threshold and the maximum temperature rise value. Use this difference as the safety margin indicator corresponding to that valid candidate safety threshold. A difference greater than 0 indicates that the valid candidate safety threshold is higher than the historical peak value, and a safety buffer exists. A difference less than or equal to 0 indicates that the valid candidate safety threshold is equal to or lower than the historical peak value, and the safety buffer is insufficient or even non-existent. In other words, the larger the difference, the greater the safety buffer space the threshold has relative to the highest historically observed temperature rise level, making it more robust and safer in the face of unforeseen fluctuations or measurement noise.

[0083] This invention first ensures that candidate thresholds meet the basic timeliness requirements for early warning through timely screening. Then, it quantifies and evaluates the overall discrimination accuracy of the threshold group on historical data using early warning performance curves and their area under the curve. Finally, it measures the safety margin of each threshold relative to historical extreme cases using a safety margin index. This ensures that the final determined target safety thresholds possess both sufficient early warning lead time and safety redundancy, thereby comprehensively improving the reliability of early warnings.

[0084] The threshold selection and determination module selects the optimal candidate safety threshold as the target safety threshold based on the prediction accuracy index and the safety margin index.

[0085] Specifically, please refer to Figure 3 As shown, selecting the optimal candidate safety threshold as the target safety threshold includes: D1, normalizing the prediction accuracy index and safety margin index of all valid candidate safety thresholds respectively, mapping them to the same numerical range of [0,1] to eliminate differences in units and numerical ranges, making them comparable. A preferred normalization method is Min-Max normalization, where the maximum and minimum values ​​involved in the normalization correspond to the maximum and minimum values ​​of the prediction accuracy index and safety margin index among all candidate thresholds, respectively.

[0086] D2. A nonlinear weighting function is used to calculate the comprehensive safety score for each candidate safety threshold, so as to reflect the synergistic or trade-off effects between indicators.

[0087] D3. Based on the calculated comprehensive security score, sort all valid candidate security thresholds in descending order and select the highest score from them.

[0088] D4. If the difference between the comprehensive score of multiple candidate safety thresholds and the highest score is within a preset tolerance range (e.g., 0.1), then the candidate safety threshold with the highest safety margin index is selected as the final target safety threshold.

[0089] D5. Otherwise, the candidate security threshold corresponding to the highest score will be used as the target security threshold.

[0090] Considering that candidate safety thresholds with high prediction accuracy may have a tight safety margin, while candidate safety thresholds with a loose safety margin may have insufficient accuracy, it is impossible to obtain the threshold with the best overall performance if only a single indicator is used for screening. In addition, the highest score may only differ slightly from several similar score thresholds in reality, and directly selecting the highest score may lead to unstable results due to computational inaccuracies or data noise.

[0091] Based on this, the present invention establishes a threshold screening module. This module first normalizes the prediction accuracy index and the safety margin index to the same dimension, making them comparable. Then, it calculates the comprehensive score of each candidate safety threshold using a nonlinear weighting function to quantify the overall performance between early warning accuracy and safety redundancy. Finally, by setting a tolerance range, a secondary judgment is made on candidate thresholds with similar scores, prioritizing the selection of thresholds with higher safety margins.

[0092] The above steps ensure that the final target safety threshold can achieve an effective balance between early warning accuracy and safety redundancy, thereby improving the accuracy and reliability of the final screening results.

[0093] It should be noted that the following additional explanations are required during the execution of steps D1 to D5 above:

[0094] In one specific embodiment, a preferred functional form of the nonlinear weighting function is a product-weighted geometric mean, the formula of which is: ,in, This indicates the overall safety score. and These represent the normalized prediction accuracy index and the safety margin index, respectively. and These represent the weights of the prediction accuracy index and the safety margin index, respectively. A low score for either index in this function will significantly lower the total score, forcing the screening results to seek a balance between the two.

[0095] in, and The values ​​range from 0 to 1, and the sum of both is 1. The values ​​are set according to the emphasis on prediction accuracy and security in the actual operation and maintenance scenario. For example, in a data center or precision manufacturing workshop, false alarms can cause significant interference or downtime losses. In this case, the operation and maintenance prioritizes accuracy (low false alarms), and the values ​​can be set accordingly. ,at this time, and The values ​​can be 0.7 and 0.3 respectively. In high-risk industries such as chemical and metallurgical industries, safety is absolutely paramount. In these cases, operation and maintenance focus on safety, and the settings can be adjusted accordingly. ,at this time and The values ​​can be 0.4 and 0.6 respectively.

[0096] The safety monitoring execution terminal performs real-time safety monitoring and early warning of the low-voltage busbar trunking based on the target safety threshold.

[0097] Specifically, the safety monitoring execution terminal obtains the target safety threshold output by the threshold filtering and determination module, and compares the real-time temperature data of each monitored connector with the target safety threshold set for it in real time. When the real-time temperature value of a connector continuously exceeds its corresponding target safety threshold, and the duration of the excess reaches a preset minimum stable alarm duration (e.g., 10 seconds), the connector is determined to be in an abnormal overheating state.

[0098] When a connector is determined to be in an abnormally overheated state, a structured early warning event record is generated. This record includes at least the following information: early warning trigger timestamp (i.e., the time of the fault), the specific busbar number where the abnormality occurred and the connector location identifier, the real-time temperature value, and the target safety threshold value exceeded. Simultaneously, an audible and visual alarm signal is triggered through the human-machine interface, and the early warning event record is pushed to the designated operation and maintenance management platform or mobile terminal in real time.

[0099] After the warning is triggered, continue to monitor the temperature of the connector. Only when the real-time temperature of the connector drops below the target safety threshold and remains stable for a preset recovery confirmation period (e.g., 30 seconds) is the abnormal state determined to be resolved, the audible and visual alarm stopped, and the recovery time marked in the warning event log.

[0100] Please see Figure 2 As shown, the present invention also provides a method for safe power supply of low-voltage busbar trunking, the method comprising: acquiring joint temperature rise data, load current data and ambient temperature data of low-voltage busbar trunking during historical safe operation, and generating a benchmark temperature rise safety threshold through cluster analysis.

[0101] The load current and ambient temperature of the current busbar trunking are collected in real time, and the reference temperature rise safety threshold is dynamically adjusted accordingly to generate multiple candidate safety thresholds.

[0102] Retrieve fault logs, perform backtracking tests on each candidate safety threshold based on the fault logs, and calculate prediction accuracy and safety margin indicators based on the test results.

[0103] The optimal candidate safety threshold is selected as the target safety threshold based on the prediction accuracy index and the safety margin index.

[0104] Real-time safety monitoring and early warning of low-voltage busbar trunking are achieved by using target safety thresholds.

[0105] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A low-voltage busbar trunking safety power supply system, characterized in that, The system includes: The baseline threshold generation module performs cluster analysis on the joint temperature rise data, load current data, and ambient temperature data of the low-voltage busbar trunking during its historical safe operation period to generate a baseline temperature rise safety threshold. The threshold dynamic adjustment module dynamically adjusts the reference temperature rise safety threshold based on the real-time load current and real-time ambient temperature of the bus trunking, generating multiple candidate safety thresholds. The threshold assessment and analysis module retrieves fault logs and performs backtesting tests on each candidate safety threshold. Based on the test results, it calculates the prediction accuracy index and safety margin index. The threshold screening and determination module screens the optimal candidate safety threshold as the target safety threshold based on the prediction accuracy index and the safety margin index. The safety monitoring execution terminal performs real-time safety monitoring and early warning of low-voltage busbar trunking based on target safety thresholds.

2. The low-voltage busbar trunking safety power supply system as described in claim 1, characterized in that: The historical safe operation period refers to the continuous operation period during which the low-voltage busbar trunking did not record any fault alarms or protective power outages caused by excessive joint temperature rise.

3. The low-voltage busbar trunking safety power supply system as described in claim 1, characterized in that: The steps for performing cluster analysis are as follows: Based on the joint temperature rise data, load current data, and ambient temperature data, a historical operation dataset is constructed; The historical operating dataset is divided into operating conditions using a clustering algorithm, forming multiple typical operating condition clusters. For each of the typical operating condition clusters, the preset percentile value of all temperature rise data within it is calculated as the initial safety threshold for that operating condition cluster. The weight of each typical operating condition cluster is determined based on the number of data points in the dataset. The initial safety thresholds of all typical operating condition clusters are weighted and summed to output the reference temperature rise safety threshold of the low-voltage busbar trunking.

4. A low-voltage busbar trunking safety power supply system as described in claim 3, characterized in that: The process of dividing operating conditions specifically includes: Standardize the load current and ambient temperature; For each historical moment in the historical operation dataset, its standardized load current value and ambient temperature value are combined into a feature vector; Based on the feature vectors of all historical moments as input, and using Euclidean distance as a metric for the similarity between vectors, the historical operating dataset is divided into multiple sets through a clustering algorithm, and each set constitutes a typical operating condition cluster.

5. A low-voltage busbar trunking safety power supply system as described in claim 3, characterized in that: The dynamic adjustment of the reference temperature rise safety threshold includes: The real-time load current value and the ambient temperature value are combined to form the real-time operating point; Calculate the Euclidean distance between the real-time operating point and the corresponding cluster center of each typical operating point cluster, and identify the typical operating point cluster with the smallest Euclidean distance as the target operating point cluster. Then, identify one or more typical operating point clusters with an Euclidean distance second only to the target operating point cluster as adjacent operating point clusters. The initial safety threshold corresponding to the target operating condition cluster is used as the first candidate safety threshold. The interpolation weights are calculated based on the Euclidean distance between the real-time operating point and the target operating condition cluster and at least one adjacent operating condition cluster cluster center. Based on the interpolation weights, the initial safety thresholds corresponding to the target working condition cluster and at least one adjacent working condition are weighted and calculated to generate one or more second candidate safety thresholds. The first candidate security threshold and one or more second candidate security thresholds are combined to form multiple candidate security thresholds.

6. A low-voltage busbar trunking safety power supply system as described in claim 1, characterized in that: The backtracking test of each candidate security threshold includes: Obtain the joint temperature rise sequence and the list of actual fault event trigger times for a continuous historical period after the most recent completion of the baseline threshold generation and adjustment and before the current real-time monitoring is started; For each candidate safety threshold, a series of simulated early warning events are generated based on the comparison results with the temperature rise time series. Each early warning event includes a start time and a duration, thereby obtaining the simulated early warning event sequence corresponding to each threshold. Based on the matching results between the simulated early warning event sequence and the list of actual fault event trigger times, the true positive rate, false positive rate, and average effective advance time of the true positive early warning for each candidate safety threshold are calculated and used as the backtracking test results.

7. A low-voltage busbar trunking safety power supply system as described in claim 6, characterized in that: The true positive rate, false positive rate, and average effective lead time for true positive warnings for each candidate safety threshold are calculated through the following steps: For each fault time point in the list of real fault events, find all simulated early warning events that precede the fault time point and whose time difference with the fault time point is less than the preset proximity time. If at least one is found, the simulated early warning event with the earliest start time is identified as the matching event for the fault time point and recorded as a successful matching event; For each candidate security threshold, perform the following sub-steps: Among the simulated early warning events generated by the system, those that are successfully matched are classified as true positive events, and the remaining events are classified as false positive events of the candidate safety threshold. The true positive rate and false positive rate of the candidate safety threshold are obtained by comparing the number of true positive events with the total number of actual failure events and the number of false positive events with the total number of simulated early warning events generated by the candidate safety threshold. For each true positive event, calculate the time difference between the start time of its corresponding simulated early warning event and the corresponding fault time point to obtain the effective advance time; The average effective lead time for all true positive events corresponding to the candidate safety threshold is calculated as the average effective lead time for true positive warning.

8. A low-voltage busbar trunking safety power supply system as described in claim 6, characterized in that: The calculated prediction accuracy indicators and safety margin indicators include: The candidate safety threshold whose average effective lead time for a true positive warning is greater than or equal to the preset effective lead time threshold is defined as the effective candidate safety threshold. For the effective candidate safety threshold, a warning performance curve for the effective candidate safety threshold is constructed with the false positive rate on the x-axis and the true positive rate on the y-axis, and the area under the curve is calculated as the prediction accuracy index. Calculate the difference between the maximum temperature rise value in the joint temperature rise time sequence and each effective candidate safety threshold, and use the difference as a safety margin index.

9. A low-voltage busbar trunking safety power supply system as described in claim 1, characterized in that: The selection of the optimal candidate security threshold as the target security threshold includes: The prediction accuracy index and safety margin index are normalized. Based on the normalized prediction accuracy index and safety margin index, a comprehensive safety score for each candidate safety threshold is calculated through nonlinear weighting. All candidate security thresholds are sorted according to the calculated comprehensive score, and the highest score is selected from them. If the difference between the combined score of multiple candidate safety thresholds and the highest score is within the preset tolerance range, then the candidate safety threshold with the highest safety margin index is selected as the final target safety threshold. Otherwise, the candidate security threshold corresponding to the highest score will be used as the target security threshold.

10. A method for safe power supply to a low-voltage busbar trunking system, characterized in that: The method includes: Acquire joint temperature rise data, load current data, and ambient temperature data of low-voltage busbar trunking during historical safe operation, and generate a baseline temperature rise safety threshold through cluster analysis; The load current and ambient temperature of the current busbar trunking are collected in real time, and the reference temperature rise safety threshold is dynamically adjusted accordingly to generate multiple candidate safety thresholds. Retrieve fault logs, perform backtracking tests on each candidate safety threshold based on the fault logs, and calculate prediction accuracy and safety margin indicators based on the test results. The optimal candidate safety threshold is selected as the target safety threshold based on the prediction accuracy index and the safety margin index. Real-time safety monitoring and early warning of low-voltage busbar trunking are achieved by using target safety thresholds.

Images