Power distribution network fault sensing system based on big data driving

Abstract

The invention discloses a power distribution network fault sensing system based on big data driving, and relates to the technical field of power distribution networks. The power distribution network fault sensing system based on big data driving comprises a time sequence data acquisition unit, a time sequence data analysis unit, a fault sensing analysis unit and a fault sensing judgment unit, the power distribution network fault early warning method comprises the following steps: respectively acquiring and processing operation state time sequence data of a power distribution node, constructing an operation state time sequence index set, generating a comprehensive fault sensing index of a power distribution network, and realizing abnormal early warning by constructing the operation state time sequence index set including thermal load offset, voltage disturbance coupling, grounding risk and sheath abnormal adjustment; according to the method, the response dimension and the logic granularity of power distribution network fault recognition are improved, the early potential abnormal trend can be effectively recognized, high-sensitivity recognition of slight state changes with early warning value is achieved, and therefore the adaptability and the sensing accuracy of a power distribution network system in a complex operation state are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution networks, and more particularly to a distribution network fault perception system driven by big data. Background Art

[0002] With the continuous complexity of urban power grid structures and the gradual improvement of distribution automation levels, the role of distribution networks in the power system has become increasingly prominent. Their operation safety and power supply reliability are directly related to the overall stability of the power system and the user power consumption experience. During actual operation, due to numerous nodes, wide structural distribution, and complex load types, distribution networks are extremely susceptible to the combined effects of various factors such as environmental temperature changes, equipment aging, load fluctuations, grounding abnormalities, and deterioration of cable sheath insulation, thus leading to problems such as short circuits, grounding faults, and insulation breakdowns. Therefore, how to real-time master the operation status of each node in the distribution network, identify potential anomalies, and achieve early warning has become a key technical issue in distribution operation and maintenance management.

[0003] In recent years, with the development of the Internet of Things for electricity and intelligent sensing technologies, more and more distribution equipment has achieved automatic collection of operation data, making intelligent fault perception based on big data analysis possible. By performing time-series analysis on multi-source operation data such as current, voltage, temperature, grounding resistance, and sheath leakage current, and combining means such as machine learning, statistical modeling, or expert systems to establish a perception mechanism that can identify the trend of operation risks is becoming an important research direction for intelligent operation and maintenance of modern distribution networks. However, how to reasonably construct an interpretable, deployable, and scalable multi-dimensional perception model remains one of the key challenges in this field.

[0004] The prior art, such as the distribution network fault perception system driven by big data disclosed in the invention patent application with publication number CN117741333B, includes an information acquisition unit, an information processing unit, a data acquisition unit, a model analysis unit, and a fault warning unit; related to the technical field of distribution networks, the distribution network fault perception system driven by big data accurately obtains the corresponding temperature range through the information processing unit, facilitating the comparison and perception of line faults, effectively ensuring the safe operation of the distribution network, and solving the problem in the prior art that temperature monitoring is affected by environmental conditions, resulting in large errors in detection results and being unable to accurately reflect the temperature value corresponding to the cable fault, and thus being unable to effectively perform fault perception on the target cable based on temperature. It can quickly and effectively obtain the corresponding perception signal, display it to the operation and maintenance personnel through the fault warning unit, and perform operation and maintenance execution, effectively solving the problems caused by untimely perception of line faults, such as line fires or other impacts.

[0005] Based on the above solution, it is found that the limitations of the existing technology include at least the following problems. In the process of fault identification of distribution network, the existing technology generally adopts a static interval judgment mechanism based on single physical quantities such as temperature and resistance. The core problem is that it ignores the multi-dimensional evolution characteristics of the operating status of the distribution node, and it is difficult to establish a fusion fault identification model that fully reflects the health of the node operation. Specifically, during the operation, the current load, voltage fluctuation, grounding status, sheath leakage and other physical quantities of the distribution node show a dynamic change trend, and often have a coupling effect that affects each other, but the existing system only performs temperature and other indicators. It is difficult to capture the combined risk characteristics of various dimensions through interval judgment, especially in the early stage of the fault, although some parameters have not exceeded the limit, their linkage characteristics have already shown risk trends. In addition, the existing system lacks a mechanism to normalize different types of parameters into a unified time series index, resulting in inconsistent fault judgment dimensions, rigid structure, and poor adaptability. It is difficult to adapt to the personalized judgment needs of different nodes under different loads, environments, and cable types. This model is particularly obvious in the face of distributed distribution networks and multi-node diversified configuration scenarios, which can easily lead to problems such as false alarms and missed alarms, and is not conducive to achieving the goal of intelligent and universal fault perception management. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a distribution network fault perception system driven by big data, which solves the problems in the prior art of lacking unified timing modeling and fusion perception of the multi-source operating status of distribution nodes, resulting in a single fault identification dimension, delayed response and poor adaptability.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a distribution network fault perception system driven by big data, comprising the following steps: a time series data acquisition unit, used to continuously acquire the operating status time series data of each distribution node in the distribution network, and perform preprocessing, wherein the operating status time series data includes load operating time series data, voltage operating time series data, grounding operating time series data, and sheath operating time series data; a time series data analysis unit, used to perform comprehensive analysis on the preprocessed operating time series data of each distribution node in the distribution network, and obtain the operating status time series data of each distribution node in the distribution network. An index set, wherein the operating state timing index set includes a thermal load offset timing index, a voltage disturbance coupling timing index, a grounding risk timing index, and a sheath abnormality adjustment timing index; a fault perception analysis unit, used to comprehensively analyze the operating state timing index set of each distribution node in the distribution network to obtain a comprehensive fault perception index of the distribution network; a fault perception judgment unit, used to judge and analyze the comprehensive fault perception index of the distribution network and a preset fault perception interval, and when the comprehensive fault perception index of the distribution network is outside the preset fault perception interval, the distribution network is regarded as an operating fault and a fault alarm is sent.

[0008] Further, the specific steps to obtain the comprehensive fault perception index of the distribution network are as follows: Obtain the sheath discrimination thresholds of each distribution node in the distribution network, where the sheath discrimination thresholds include the sheath disturbance sensitivity threshold and the sheath anomaly activation threshold; Input the sheath discrimination thresholds and the operation status time series index sets of each distribution node in the distribution network into the fault perception analysis model respectively to obtain the fault perception index of each distribution node in the distribution network; Conduct a comprehensive analysis of the fault perception indices of each distribution node in the distribution network to obtain the comprehensive fault perception index of the distribution network.

[0009] Further, the fault perception analysis model is specifically as follows: Among them, GzG is the fault perception index of a certain distribution node in the distribution network, RfZ is the thermal load offset time series index of a certain distribution node in the distribution network, DyR is the voltage disturbance coupling time series index of a certain distribution node in the distribution network, JdF is the grounding risk time series index of a certain distribution node in the distribution network, HyT is the sheath anomaly adjustment time series index of a certain distribution node in the distribution network, RmG is the sheath disturbance sensitivity threshold of a certain distribution node in the distribution network, α is the load offset influence coefficient stored in the database, YcJ is the sheath anomaly activation threshold of a certain distribution node in the distribution network, and χ is the sheath anomaly influence coefficient stored in the database.

[0010] Further, the load operation time series data includes the input current values at the node input end, the output current values at the node output end, and the operation temperature values of the node cable joints at several time points. The specific steps to obtain the thermal load offset time series index of each distribution node in the distribution network are as follows: Conduct a comprehensive analysis of the load operation time series data of each distribution node in the distribution network respectively to obtain the load operation status set of each distribution node in the distribution network, where the load operation status set includes the input current operation index at the node input end, the output current operation index at the node output end, and the operation temperature index of the node cable joints; Obtain the operation temperature safety index of the node cable joints of each distribution node in the distribution network, and conduct a comprehensive analysis in combination with the load operation status set of the corresponding distribution node respectively to obtain the thermal load offset time series index of each distribution node in the distribution network.

[0011] Further, the specific steps to obtain the load operation state set of each distribution node in the distribution network are as follows: Read the input current values at the node input ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the input current operation index at the node input ends of each distribution node in the distribution network; Read the output current values at the node output ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the output current operation index at the node output ends of each distribution node in the distribution network; Read the operating temperature values of the node cable joints of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the operating temperature index of the node cable joints of each distribution node in the distribution network.

[0012] Further, the specific formula for calculating the thermal load offset time series index of a certain distribution node in the distribution network is as follows: Wherein, RfZ is the thermal load offset time series index of a certain distribution node in the distribution network, RdL is the input current operation index at the node input end of a certain distribution node in the distribution network, CdL is the output current operation index at the node output end of a certain distribution node in the distribution network, β is the operating current adjustment factor stored in the database, δ1 is the current influence coefficient stored in the database, YwZ is the operating temperature index of the node cable joints of a certain distribution node in the distribution network, YwA is the operating temperature safety index of the node cable joints of a certain distribution node in the distribution network, and δ2 is the temperature influence coefficient stored in the database.

[0013] Further, the voltage operation time series data includes the input voltage values of the node buses at several time points. The specific steps to obtain the voltage disturbance coupling time series index of each distribution node in the distribution network are as follows: Based on the voltage operation time series data of each distribution node in the distribution network, analyze the input voltage fluctuation index of the node buses of each distribution node in the distribution network; Obtain the high-frequency disturbance energy values within the node setting areas of each distribution node in the distribution network, and conduct comprehensive analysis by combining them with the input voltage fluctuation indexes of the node buses of the corresponding distribution nodes respectively to obtain the voltage disturbance coupling time series index of each distribution node in the distribution network.

[0014] Further, the specific steps to analyze the input voltage fluctuation index of the node buses of each distribution node in the distribution network are as follows: Conduct comprehensive analysis on the input voltage values of the node buses of each distribution node in the distribution network at several time points respectively to obtain the input voltage mean value of the node buses of each distribution node in the distribution network; Combine the input voltage values of the node buses of each distribution node in the distribution network at several time points with the input voltage mean values of the node buses of the corresponding distribution nodes respectively for comprehensive analysis to obtain the input voltage fluctuation index of the node buses of each distribution node in the distribution network.

[0015] Further, the grounding operation timing data includes the node grounding resistance values at several time points. The specific steps to obtain the grounding risk timing index for each distribution node in the distribution network are as follows: comprehensively analyze the node grounding resistance values at several time points for each distribution node in the distribution network to obtain the average node grounding resistance and the change rate of the node grounding resistance for each distribution node in the distribution network; obtain the rated output current operation index of the node output end for each distribution node in the distribution network, and comprehensively analyze the average node grounding resistance, the change rate of the node grounding resistance, and the output current operation index of the node output end of the corresponding distribution node respectively to obtain the grounding risk timing index for each distribution node in the distribution network.

[0016] Further, the sheath operation timing data includes the leakage current values of the node cable sheaths at several time points. The specific steps to obtain the sheath abnormal adjustment timing index for each distribution node in the distribution network are as follows: based on the leakage current values of the node cable sheaths at several time points for each distribution node in the distribution network, analyze the maximum leakage current value and the average leakage current of the node cable sheath for each distribution node in the distribution network respectively; comprehensively analyze the maximum leakage current value and the average leakage current of the node cable sheath for each distribution node in the distribution network respectively to obtain the sheath abnormal adjustment timing index for each distribution node in the distribution network.

[0017] The present invention has the following beneficial effects:

[0018] (1) The distribution network fault perception system based on big data drive breaks through the limitation of the prior art that only relies on a single physical quantity, such as temperature and resistance, for static judgment by constructing an operation state timing index set including thermal load offset, voltage disturbance coupling, grounding risk, and sheath abnormal adjustment. It realizes the unified index modeling of the multi-source operation states of distribution nodes. During the operation of the system, multiple-dimensional parameters such as node current, voltage, grounding resistance, and sheath leakage current are dynamically collected in a timing manner, and through feature extraction and comprehensive analysis, they are transformed into standardized operation indexes. This index structure not only improves the response dimension and logical granularity of fault recognition but also can effectively identify early potential abnormal trends, realizing highly sensitive recognition of slight but warning-worthy state changes, thus greatly enhancing the adaptability and perception accuracy of the distribution network system under complex operation states.

[0019] (2) The big data-driven distribution network fault perception system realizes a segmented fusion mechanism based on dynamic adjustment of the sheath state by introducing a sheath anomaly adjustment timing index into the fault perception analysis model and setting a sheath disturbance sensitivity threshold and a sheath anomaly activation threshold. When the sheath state of the distribution node is in the normal or mild disturbance stage, the model uses a smooth structure to fuse three types of indexes: thermal load, voltage disturbance, and grounding risk, ensuring the system stability. When the sheath anomaly adjustment index exceeds the preset activation threshold, the model automatically switches to an enhanced response structure, so as to amplify the sensitivity and optimize the warning response for high-risk nodes without affecting the accuracy of normal state judgment. This structure has good structural elasticity and segmented recognition ability, effectively solving the problems of rigid traditional model structure and discontinuous response intensity, and significantly enhancing the practical value of the system in early fault identification and key node discrimination.

[0020] (3) The big data-driven distribution network fault perception system realizes comprehensive quantification of node-level fault risks and threshold-driven intelligent judgment logic by constructing a closed-loop linkage mechanism between the fault perception analysis unit and the fault perception judgment unit. When the operation state timing index set of the distribution node is input into the analysis model, the corresponding fault perception index can be quickly generated, and the judgment unit compares it with the preset fault perception interval to timely judge whether the current node has entered the abnormal state interval. This process requires no manual intervention and can automatically complete the complete closed-loop process from multi-source data perception, index fusion operation to fault level identification and alarm. At the same time, the system has the ability to flexibly configure the alarm interval, supporting the setting of hierarchical response thresholds according to the characteristics of the regional power grid or business requirements, so as to avoid false alarms and missed alarms while improving the response priority for high-risk nodes. This mechanism significantly enhances the online discrimination ability and automatic response ability of the system, providing efficient and accurate fault identification and warning services for the intelligent operation and maintenance of the distribution network.

[0021] Of course, it is not necessary for any product implementing the present invention to simultaneously achieve all the above-mentioned advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a block diagram of the big data-driven distribution network fault perception system of the present invention.

[0023] Figure 2 It is a specific step flowchart for obtaining the thermal load offset timing index of each distribution node in the distribution network in the big data-driven distribution network fault perception system of the present invention.

[0024] Figure 3 It is a timing diagram of the input current at the node input end of a certain distribution node in the distribution network of the big data-driven distribution network fault perception system of the present invention.

[0025] Figure 4 It is the output current time series diagram of the output terminal of a certain distribution node in the distribution network of the distribution network fault perception system driven by big data according to the present invention.

[0026] Figure 5 It is the operating temperature time series diagram of the node cable joint of a certain distribution node in the distribution network of the distribution network fault perception system driven by big data according to the present invention.

[0027] Figure 6 It is the specific step flow chart for obtaining the sheath anomaly adjustment time series index of each distribution node in the distribution network in the distribution network fault perception system driven by big data according to the present invention. Specific Embodiment

[0028] Please refer to Figure 1 , an embodiment of the present invention provides a technical solution: a distribution network fault perception system driven by big data, including the following steps: a time series data acquisition unit, configured to continuously acquire the operating state time series data of each distribution node in the distribution network and perform preprocessing. The operating state time series data includes load operating time series data, voltage operating time series data, grounding operating time series data, and sheath operating time series data; a time series data analysis unit, configured to respectively perform comprehensive analysis on the operating time series data of each distribution node in the preprocessed distribution network to obtain an operating state time series index set of each distribution node in the distribution network. The operating state time series index set includes a thermal load offset time series index, a voltage disturbance coupling time series index, a grounding risk time series index, and a sheath anomaly adjustment time series index; a fault perception analysis unit, configured to perform comprehensive analysis on the operating state time series index set of each distribution node in the distribution network to obtain a comprehensive fault perception index of the distribution network; a fault perception judgment unit, configured to perform judgment analysis on the comprehensive fault perception index of the distribution network and a preset fault perception interval, and when the comprehensive fault perception index of the distribution network is outside the preset fault perception interval, regard the distribution network as an operating fault and send a fault alarm to relevant staff.

[0029] Specifically, as Figure 2 shown, the specific steps for obtaining the comprehensive fault perception index of the distribution network are as follows: obtain the sheath discrimination threshold of each distribution node in the distribution network, and the sheath discrimination threshold includes a sheath disturbance sensitivity threshold and a sheath anomaly activation threshold; input the sheath discrimination threshold and the operating state time series index set of each distribution node in the distribution network into the fault perception analysis model respectively to obtain the fault perception index of each distribution node in the distribution network; perform comprehensive analysis (i.e., mean analysis) on the fault perception index of each distribution node in the distribution network to obtain the comprehensive fault perception index of the distribution network.

[0030] Among them, the sheath disturbance sensitivity threshold is used to identify whether there is a slight insulation offset or abnormal leakage at the cable sheath of the distribution node. This threshold is mainly used to distinguish the critical state between normal fluctuations and potential abnormalities. When the sheath anomaly adjustment index exceeds this threshold, it is determined that there may be a slight anomaly. It is statistically modeled through the historical sheath leakage current data under the normal operation state on-site. Within a selected time window (such as 7 days or 30 days), the sheath anomaly adjustment time series index sequence for all time periods is calculated, and then its 90th percentile (in a set of data, 90% of the data values are less than or equal to it, and only 10% of the data values are greater than it) is taken as the preliminary judgment threshold for this node, reflecting the upper limit of the normal fluctuation range in most cases.

[0031] The sheath anomaly activation threshold is used to judge the possibility that the cable sheath state has shown a high degree of anomaly, structural deterioration, or insulation breakdown. When the sheath anomaly adjustment index exceeds this value, the system will execute an amplified fusion structure to promptly respond to major potential hazards. It is obtained by statistically analyzing actual known fault samples or test simulation data. Select a batch of nodes that have experienced typical faults such as sheath insulation damage, moisture intrusion, and severe leakage in history, extract the sheath anomaly adjustment index of these nodes within 30 minutes to 5 minutes before the fault occurs, and calculate its average value or the lower maximum value as the trigger threshold. If there is a lack of measured fault data, the 99th percentile of the sheath anomaly adjustment index in the system (in a set of data, 99% of the data values are less than or equal to it, and only 1% of the data values are greater than it) can be used as an alternative to approximately approach the extreme anomaly startup critical value.

[0032] The fault perception analysis model is specifically as follows: Among them, GzG is the fault perception index of a certain distribution node in the distribution network, RfZ is the thermal load offset time series index of a certain distribution node in the distribution network, DyR is the voltage disturbance coupling time series index of a certain distribution node in the distribution network, JdF is the grounding risk time series index of a certain distribution node in the distribution network, HyT is the sheath anomaly adjustment time series index of a certain distribution node in the distribution network, RmG is the sheath disturbance sensitivity threshold of a certain distribution node in the distribution network, α is the load offset influence coefficient stored in the database, YcJ is the sheath anomaly activation threshold of a certain distribution node in the distribution network, and χ is the sheath anomaly influence coefficient stored in the database.

[0033] Among them, tanh(x) is the hyperbolic tangent function, and its structure is: tanh(x) = (e x -e -x ) / (e x +e -x ), with the domain being all real numbers and the range being (-1, 1).

[0034] It should be noted that the specific steps for obtaining the load offset influence coefficient α and sheath anomaly influence coefficient χ stored in the database are as follows: collect the input and output currents, sheath leakage current, cable temperature and other parameters of multiple distribution nodes under different operating states (normal, slightly abnormal, severely abnormal), and record whether actual faults occur during the corresponding period; subsequently, through regression analysis or Bayesian correlation modeling, the relationship between load imbalance (such as input current minus output current) and the probability of subsequent abnormal events is parameterized and extracted as the load offset influence coefficient, and the correlation degree between the sheath leakage current characteristics (such as range ratio or mutation rate) and the fault risk is extracted as the sheath anomaly influence coefficient; finally, both are stored in the database as constant coefficients in the model for use by the fault perception model.

[0035] In this implementation plan, by introducing the sheath disturbance sensitivity threshold and sheath anomaly activation threshold, the hierarchical judgment and dynamic response control of the cable sheath state are realized, effectively improving the system's ability to identify and distinguish between slight anomalies and severe faults. The sheath disturbance sensitivity threshold is constructed based on long-term normal operation data, which can accurately define the boundary between the initial insulation offset and natural fluctuations, avoiding false alarms; while the sheath anomaly activation threshold is obtained by modeling historical fault samples, ensuring sufficient sensitivity and response priority for severe sheath deterioration. At the same time, the fault perception analysis model combines the hyperbolic tangent function structure and the nested exponential fusion mechanism, which can automatically adjust the exponential fusion form at different stages of the sheath state, realizing the non-linear mapping between the exponential response amplitude and the anomaly level. By introducing the load offset influence coefficient and the sheath anomaly influence coefficient, the model's ability to adjust the weights of load imbalance and insulation deterioration is further improved, making the fault index construction process more physically reasonable and scenario adaptable, thus significantly enhancing the accuracy, reliability and generalization ability of the distribution network fault identification.

[0036] Specifically, as Figure 2 shown, the load operation time series data includes the input current values at the node input end, the output current values at the node output end, and the operating temperature values of the node cable joints at several time points. The specific steps for obtaining the thermal load offset time series index of each distribution node in the distribution network are as follows: comprehensively analyze the load operation time series data of each distribution node in the distribution network to obtain the load operation state set of each distribution node in the distribution network. The load operation state set includes the input current operation index at the node input end, the output current operation index at the node output end, and the operating temperature index of the node cable joints; obtain the operating temperature safety index of the node cable joints of each distribution node in the distribution network, and comprehensively analyze it in combination with the load operation state set of the corresponding distribution node to obtain the thermal load offset time series index of each distribution node in the distribution network.

[0037] Among them, the input current value at the node input end can be measured and obtained through a current transformer (CT) installed at the incoming line end of the distribution node.

[0038] The output current value at the node output end can be measured and obtained through a current transformer (CT) installed at each branch or outgoing line end.

[0039] The operating temperature value of the node cable joint can be measured and obtained through a thermocouple or an infrared temperature sensor arranged at the cable joint or the insulation connection.

[0040] The operating temperature safety index of the node cable joint is a standardized evaluation value used to evaluate whether the current temperature rise of the distribution node cable joint is within the safe range. The steps to obtain it are as follows: First, continuously collect the joint operating temperature data for a period of time through the thermocouple or infrared temperature sensor arranged at the cable joint, and calculate its moving average temperature value; Second, combine the rated maximum operating temperature specified in the factory technical specification corresponding to the equipment model of this joint, and consider the current ambient temperature. According to experience or a model, introduce a compensation factor λ to calculate the temperature safety threshold: rated maximum operating temperature - (compensation factor × (current ambient temperature - 25°C)); Finally, perform ratio normalization processing on the current moving temperature average value and the safety threshold to obtain the temperature safety index. The closer this index is to 1, the closer the temperature rise is to the limit, and the more likely it is to cause a failure.

[0041] The compensation factor is used to reflect the correction effect of the ambient temperature on the heat-carrying capacity of the cable joint. The steps to obtain it are as follows: First, collect the historical temperature operation data of multiple distribution node cable joints at different ambient temperatures, and record the corresponding equipment surface temperature rise conditions; Second, perform difference analysis on the joint operating temperature under the same load condition and the control data under the standard environment (such as 25°C), and calculate the temperature change amount ΔT of the joint caused by every 1°C increase in the ambient temperature to form a temperature rise - ambient correspondence table; Third, select node samples with stable temperature rise sensitivity, and extract the average slope as the ambient temperature correction coefficient λ through linear regression or fitting methods; Finally, perform empirical grading correction on λ according to the joint installation location (such as indoor / outdoor) and ventilation conditions to improve its adaptability in different on-site environments, and then use it to dynamically calculate the operating temperature safety threshold.

[0042] The specific steps to obtain the load operation status set of each distribution node in the distribution network are as follows: Read the input current values at the node input ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis (i.e., mean analysis) to obtain the input current operation index at the node input ends of each distribution node in the distribution network; Read the output current values at the node output ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis (i.e., mean analysis) to obtain the output current operation index at the node output ends of each distribution node in the distribution network; Read the operating temperature values of the node cable joints of each distribution node in the distribution network at several time points, and conduct comprehensive analysis (i.e., mean analysis) to obtain the operating temperature index of the node cable joints of each distribution node in the distribution network.

[0043] Among them, the specific implementation examples for calculating the input current operation index at the node input end, the output current operation index at the node output end, and the operating temperature index of the node cable joint of a certain distribution node in the distribution network are as follows. There are the following data, including the input current values, output current values, and operating temperature values of the node cable joints at the node input ends at six time points, as shown in Table 1 and Figures 3 - 5 shown:

[0044] Table 1 Example of load operation time series data of a certain distribution node in the distribution network

[0045]

[0046]

[0047] Conduct mean analysis on the data in Table 1 respectively, and obtain:

[0048] The input current operation index at the node input end of a certain distribution node in the distribution network is approximately: 360.218A.

[0049] The output current operation index at the node output end of a certain distribution node in the distribution network is approximately: 290.443A.

[0050] The operating temperature index of the node cable joint of a certain distribution node in the distribution network is approximately: 68.420°C.

[0051] The specific formula for calculating the thermal load offset time series index of a certain distribution node in the distribution network is as follows: Among them, RfZ is the thermal load offset time series index of a certain distribution node in the distribution network, RdL is the input current operation index of the node input end of a certain distribution node in the distribution network, CdL is the output current operation index of the node output end of a certain distribution node in the distribution network, β is the operation current adjustment factor stored in the database, which is used to prevent the denominator from being 0, δ1 is the current influence coefficient stored in the database, YwZ is the operation temperature index of the node cable joint of a certain distribution node in the distribution network, YwA is the operation temperature safety index of the node cable joint of a certain distribution node in the distribution network, and δ2 is the temperature influence coefficient stored in the database.

[0052] It should be explained that the specific acquisition steps of the current influence coefficient δ1 and the temperature influence coefficient δ2 stored in the database are as follows: Select multiple representative distribution nodes, and synchronously collect the time series data of the input current, output current and cable joint temperature during different operation cycles; Secondly, according to the numerical response relationship between the load mutation amplitude of the node in each time period and the temperature rise change caused by it, use the least square fitting method or correlation coefficient analysis to calculate the temperature response change rate caused by the unit current offset, and respectively form a set of influence slope coefficient sets of current on temperature offset; Then, normalize the analysis results of multiple nodes under the same operation model to obtain the standardized current influence coefficient δ1 and temperature influence coefficient δ2 that can be used for unified model call; Finally, store them in the system database as adjustment factors in the subsequent calculation of the thermal load offset index, which is used to reflect the sensitivity differences of the nodes to current / temperature fluctuations.

[0053] Among them, the specific implementation example of calculating the thermal load offset time series index of a certain distribution node in the distribution network is as follows. The following are the existing parameters:

[0054] The input current operation index of the node input end of a certain distribution node in the distribution network is approximately: 360.218A.

[0055] The output current operation index of the node output end of a certain distribution node in the distribution network is approximately: 290.443A.

[0056] The operation current adjustment factor stored in the database is approximately: 0.1.

[0057] The current influence coefficient stored in the database is approximately: 0.833.

[0058] The operation temperature index of the node cable joint of a certain distribution node in the distribution network is approximately: 68.420℃.

[0059] The operation temperature safety index of the node cable joint of a certain distribution node in the distribution network is approximately: 65.0℃.

[0060] The temperature influence coefficient δ2 stored in the database is approximately: 0.974.

[0061] Substitute the above data into the specific formula for calculating the thermal load offset time series index of a certain distribution node in the distribution network respectively, and we get:

[0062] The thermal load offset time series index of a certain distribution node in the distribution network = |(360.218 - 290.443) / (360.218 + 0.1)|^0.833 × (68.420 / 65.0)^0.974 ≈ 0.2796.

[0063] In this implementation plan, by constructing the thermal load offset time series index, the core operating parameters such as the input current, output current, and cable joint temperature of the distribution node are systematically integrated, forming an evaluation mechanism that can dynamically perceive the change trend of the node thermal load. Compared with the existing method that only uses temperature over - limit as the basis for fault judgment, this method not only considers the transmission offset between current input and output, but also introduces the temperature safety index as the normalization judgment benchmark for thermal risk, realizing an essential leap from "value judgment" to "index integration". By introducing a compensation factor to dynamically correct the temperature safety threshold, it ensures that the model can accurately reflect the thermal risk under different ambient temperature conditions; at the same time, both the current influence coefficient and the temperature influence coefficient are obtained through multi - node historical data modeling, with engineering reproducibility and regional adaptability. This method significantly improves the response sensitivity to hidden thermal anomalies such as load deterioration and poor conductor contact, providing a higher - credibility pre - fault prediction ability for the operation state of the distribution network.

[0064] Specifically, the voltage operation time series data includes the input voltage values of the node busbars at several time points. The specific steps to obtain the voltage disturbance coupling time series index of each distribution node in the distribution network are as follows: Based on the voltage operation time series data of each distribution node in the distribution network, analyze the input voltage fluctuation index of the node busbar of each distribution node in the distribution network; obtain the high - frequency disturbance energy value within the node set area of each distribution node in the distribution network, and conduct comprehensive analysis by combining the input voltage fluctuation index of the node busbar of the corresponding distribution node respectively, so as to obtain the voltage disturbance coupling time series index of each distribution node in the distribution network.

[0065] The specific steps to analyze the input voltage fluctuation index of the node busbar of each distribution node in the distribution network are as follows: Conduct comprehensive analysis (i.e., mean value analysis) on the input voltage values of the node busbars of each distribution node in the distribution network at several time points respectively to obtain the input voltage mean value of the node busbar of each distribution node in the distribution network; combine the input voltage values of the node busbars of each distribution node in the distribution network at several time points with the input voltage mean value of the node busbar of the corresponding distribution node respectively for comprehensive analysis (i.e., variance analysis) to obtain the input voltage fluctuation index of the node busbar of each distribution node in the distribution network.

[0066] Among them, the input voltage value of the node bus refers to the instantaneous voltage value when each distribution node in the distribution network is connected to the main bus within a set time period, which can be measured and obtained through a potential transformer (PT) installed on the incoming line side of the distribution node or an intelligent terminal with voltage sampling function (such as DTU / FTU).

[0067] The high-frequency disturbance energy value within the node set area is used to identify whether there are non-power frequency disturbance phenomena such as short-time pulses, arc discharges, and parallel transients near the distribution node. The acquisition steps are as follows: Arrange electromagnetic induction probes with high-frequency detection capabilities (such as high-frequency current sensors HFCT or transient voltage recorders TVR) in the distribution node set area (such as in-situ transformer, inside the ring network cabinet, or on the side of the distribution box), continuously collect voltage / current disturbance waveforms at a sampling rate dozens of times higher than the power frequency (such as 1 - 10 kHz), and use the short-time Fourier transform (STFT) or wavelet packet energy decomposition method to extract high-frequency components. Finally, take the sum of the frequency-domain energy within the time window as the high-frequency disturbance energy value of this node.

[0068] The specific formula for calculating the voltage disturbance coupling time series index of a certain distribution node in the distribution network is as follows: Among them, DyR is the voltage disturbance coupling time series index of a certain distribution node in the distribution network, DyB is the input voltage fluctuation index of the node bus of a certain distribution node in the distribution network, RdN is the high-frequency disturbance energy value within the node set area of a certain distribution node in the distribution network, μ1 is the anti-interference gain coefficient stored in the database (indicating the degree of enhanced non-linear response after the input voltage disturbance is coupled with the high-frequency pulse, the larger the value, the higher the sensitivity to abnormal fluctuations), and μ2 is the cycle modulation coefficient stored in the database (indicating the response sensitivity of the system to the periodic oscillation signal in the voltage fluctuation).

[0069] It should be explained that the specific acquisition steps for the anti-interference gain coefficient μ1 and the cycle modulation coefficient μ2 stored in the database are as follows: For the anti-interference gain coefficient μ1, select multiple distribution nodes with known high-frequency pulse disturbance events, extract the combined input of their input voltage fluctuation value and the peak value of the high-frequency pulse energy, and perform fitting regression with the actual fault level or equipment response situation during this period. The optimal exponential amplification coefficient is obtained through the minimum error fitting; for the cycle modulation coefficient μ2, select node samples with periodic voltage fluctuation characteristics, perform spectral analysis (such as FFT or wavelet decomposition) on the periodic components of the variance of their input voltage fluctuations, and combine the contribution degree of the periodic disturbance to the fault trigger in historical events to construct a periodic disturbance response model, and then fit to obtain the amplitude modulation coefficient that is most sensitive to the periodic oscillation signal.

[0070] In this implementation scheme, by constructing a voltage disturbance coupling time series index, the coupling modeling between the node input voltage fluctuation and the high-frequency disturbance signal is realized, effectively breaking through the traditional voltage stability assessment method that only relies on the mean value or single-point voltage deviation, improving the system's response ability to complex phenomena such as non-power frequency disturbance, hidden oscillation, and voltage interruption. The system first extracts the fluctuation index of the node input voltage and combines it with the high-frequency disturbance energy value within the set area to characterize the linkage risk between external disturbance and node voltage abnormality. At the same time, an anti-disturbance gain coefficient and a cycle modulation coefficient are introduced to realize the non-linear amplification adjustment of the coupling strength and periodic fluctuation. Among them, the anti-disturbance gain coefficient is automatically obtained by regression fitting the fluctuation-fault relationship in historical events, and the cycle modulation coefficient extracts the periodic response characteristics by combining frequency domain analysis, enabling the model to adapt to strong mutation disturbances and identify periodic hidden abnormalities, thereby enhancing the voltage abnormality perception accuracy and adaptability of the system in a complex operating environment.

[0071] Specifically, the grounding operation time series data includes the node grounding resistance values at several time points. The specific steps to obtain the grounding risk time series index of each distribution node in the distribution network are as follows: comprehensively analyze the node grounding resistance values at several time points of each distribution node in the distribution network to obtain the node grounding resistance mean value and the node grounding resistance change rate of each distribution node in the distribution network; obtain the rated output current operation index of the node output end of each distribution node in the distribution network, and comprehensively analyze it in combination with the node grounding resistance mean value, the node grounding resistance change rate, and the output current operation index of the node output end of the corresponding distribution node to obtain the grounding risk time series index of each distribution node in the distribution network.

[0072] Among them, the node grounding resistance value can be measured and obtained through a grounding resistance on-line detection module installed in the grounding device of the distribution node, such as a low-frequency current injection method grounding resistance sensor or a loop method grounding detector.

[0073] The rated output current operation index of the node output end is used to evaluate the proportion of the current actual output current of the distribution node relative to its designed rated output capacity. The acquisition steps are as follows: first, the operation current value of the node output end is collected in real time through a current transformer (CT) installed at the node branch, and the sliding mean value is extracted by setting a time window; then, the rated output current value of the device (such as a transformer, a feeder loop, or a circuit breaker) corresponding to the node output end is queried, and finally, the operation index is calculated as the ratio of the two.

[0074] Among them, the specific formula for calculating the voltage disturbance coupling time series index of a certain distribution node in the distribution network is as follows: Among them, JdF is the grounding risk time series index of a certain distribution node in the distribution network, JdJ is the average value of the node grounding resistance of a certain distribution node in the distribution network, JdB is the change rate of the node grounding resistance of a certain distribution node in the distribution network, ω is the grounding resistance influence coefficient stored in the database, CdL is the output current operation index of the node output end of a certain distribution node in the distribution network, and EcD is the rated output current operation index of the node output end of a certain distribution node in the distribution network.

[0075] It should be explained that the specific acquisition steps of the grounding resistance influence coefficient ω stored in the database are as follows: Select multiple distribution node samples with historical grounding deterioration or grounding fault events, and collect multi-dimensional data such as the historical average grounding resistance, change rate, current, voltage, fault type and level during the fault occurrence period in these nodes; Secondly, construct a mapping relationship model between abnormal grounding behavior and fault response (which can adopt multiple linear regression or logistic regression method), use node grounding characteristics (such as average grounding resistance, change rate) as input, and actual fault risk index or alarm level as output for fitting training; Finally, extract the standardized weight of the grounding variable in the fitting model as the grounding resistance influence coefficient of this node type under different operating backgrounds, and store it in the database for subsequent model structure call.

[0076] In this implementation plan, by constructing the grounding risk time series index, the average value of the grounding resistance, change rate and current load operation state of the distribution node are systematically integrated, realizing the dynamic perception and hierarchical judgment of the deterioration risk of the grounding system. Different from the existing technology that only uses the absolute value of the grounding resistance exceeding the limit as the alarm basis, by introducing the change rate as the trend index of the grounding state, and by combining the output current operation index to reflect the risk amplification effect under the high-load state, it is possible to identify those hidden grounding defects that have not reached the alarm value but have shown a deteriorating trend. At the same time, by analyzing historical grounding fault samples to construct the grounding resistance influence coefficient, the grounding characteristic risk weight under different operating scenarios is determined in a data-driven manner, making the model more in line with the actual operating logic on site and having good generalization and adaptation capabilities. This mechanism not only improves the response accuracy to the precursors of grounding failure, but also helps to intervene in nodes with potential safety hazards in advance, providing more scientific and controllable risk management support for the safe and stable operation of the distribution network.

[0077] Specifically, such as Figure 6As shown in the figure, the sheath operation timing data includes the leakage current values of the node cables of the sheath at several time points. The specific steps to obtain the sheath abnormal regulation timing index of each distribution node in the distribution network are as follows: Based on the leakage current values of the node cables of the sheath at several time points of each distribution node in the distribution network, analyze the maximum leakage current value and the average leakage current of the node cable sheath of each distribution node in the distribution network respectively; comprehensively analyze the maximum leakage current value and the average leakage current of the node cable sheath of each distribution node in the distribution network respectively to obtain the sheath abnormal regulation timing index of each distribution node in the distribution network.

[0078] Among them, the leakage current value of the node cable sheath refers to the abnormal weak leakage current generated by the outer sheath or shielding layer of a certain distribution node cable in the distribution network during operation due to insulation aging, moisture ingress, mechanical damage or the influence of stray electric fields; this value can reflect the change of the cable insulation state and is an important physical parameter for identifying potential hidden faults such as insulation degradation or humidity erosion. It can be measured and obtained through highly sensitive leakage current sensors (such as microampere-level leakage current transformers or high-frequency current probes) installed at the end of the distribution node cable or the grounding lead-out point of the sheath.

[0079] Among them, the specific formula for calculating the sheath abnormal regulation timing index of a certain distribution node in the distribution network is as follows: Among them, HyT is the sheath abnormal regulation timing index of a certain distribution node in the distribution network, e is the natural constant, which takes the value of 2.71 in this embodiment, LdZ is the maximum leakage current value of the node cable sheath of a certain distribution node in the distribution network, LdL is the average leakage current of the node cable sheath of a certain distribution node in the distribution network, θ is the leakage current adjustment factor stored in the database to prevent the denominator from being zero, ψ1 is the leakage current fluctuation offset influence coefficient stored in the database, and ψ2 is the average leakage current influence coefficient stored in the database.

[0080] It should be explained that the specific steps to obtain the leakage current fluctuation offset influence coefficient ψ1 and the average leakage current influence coefficient ψ2 stored in the database are as follows: For the leakage current fluctuation offset influence coefficient, select typical operation periods with mild sheath insulation degradation or precursors of partial discharge in multiple distribution nodes, calculate the ratio of the difference between the maximum sheath leakage current and the average value in each period, and then perform fitting modeling on the subsequent insulation degradation degree or operation and maintenance response level corresponding to these samples, and extract the slope of the response function between the peak and the fault as the sensitivity coefficient. After normalization, it is ψ1; for the average leakage current influence coefficient, collect the average sheath leakage current of a large number of distribution nodes in the non-peak stage and the corresponding insulation health levels (such as cable insulation resistance, on-site infrared thermal imaging results, etc.), use polynomial fitting or piecewise regression analysis to extract the response intensity coefficient between the average sheath leakage current and the insulation risk, and take its standardized coefficient as ψ2 and store it in the database for dynamic call during model calculation.

[0081] In this embodiment, by constructing a sheath anomaly adjustment time series index, a dynamic recognition model based on the maximum and average leakage currents is established, which can effectively characterize the deviation trend of the cable sheath's operating state under various hidden deterioration conditions such as insulation aging, moisture ingress, and mechanical damage. Compared with the single current threshold judgment method in the prior art, this index model combines the relative deviation degree and overall strength level of the maximum leakage current and the average leakage current, and realizes the non-linear adjustment of the fault risk through the fluctuation deviation influence coefficient and the average leakage current influence coefficient stored in the database, thereby realizing the multi-stage perception of the entire process of insulation degradation. All parameters used in the system can be obtained online through conventional leakage current sensors, and the calculation process is simple, easy to deploy at cable joints or terminal lead-out points, and has good engineering feasibility. Especially when the early sheath anomaly has not yet developed to the breakdown stage, this index can identify the operating state that, although not severe but deviates from stability, significantly improves the ability to advance the fault warning and the resolution granularity, and helps to implement data-driven proactive inspection and predictive maintenance strategies.

[0082] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0083] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. A distribution network fault perception system driven by big data, characterized in that, It includes the following steps: A time-series data acquisition unit for continuously acquiring the time-series operation status data of each distribution node in the distribution network and performing preprocessing. The operation status time-series data includes load operation time-series data, voltage operation time-series data, grounding operation time-series data, and sheath operation time-series data; A time-series data analysis unit for comprehensively analyzing the operation time-series data of each distribution node in the preprocessed distribution network respectively to obtain the operation status time-series index set of each distribution node in the distribution network. The operation status time-series index set includes a thermal load offset time-series index, a voltage disturbance coupling time-series index, a grounding risk time-series index, and a sheath anomaly regulation time-series index; A fault perception analysis unit for comprehensively analyzing the operation status time-series index set of each distribution node in the distribution network to obtain the comprehensive fault perception index of the distribution network; A fault perception judgment unit for judging and analyzing the comprehensive fault perception index of the distribution network with a preset fault perception interval, and regarding the distribution network as an operation fault and sending a fault alarm when the comprehensive fault perception index of the distribution network is outside the preset fault perception interval.

2. The big data-driven distribution network fault perception system according to claim 1, characterized in that The specific steps for obtaining the comprehensive fault perception index of the distribution network are as follows: Obtain the sheath discrimination threshold of each distribution node in the distribution network. The sheath discrimination threshold includes a sheath disturbance sensitivity threshold and a sheath anomaly activation threshold; Input the sheath discrimination threshold and the operation status time-series index set of each distribution node in the distribution network into the fault perception analysis model respectively to obtain the fault perception index of each distribution node in the distribution network; Comprehensively analyze the fault perception index of each distribution node in the distribution network to obtain the comprehensive fault perception index of the distribution network.

3. The fault perception system of the distribution network driven by big data according to claim 2, characterized in that The specific fault perception analysis model is as follows: Among them, GzG, RfZ, DyR, JdF, HyT, RmG, and YcJ are the fault perception index, thermal load offset time-series index, voltage disturbance coupling time-series index, grounding risk time-series index, sheath anomaly regulation time-series index, sheath disturbance sensitivity threshold, and sheath anomaly activation threshold of a certain distribution node in the distribution network respectively. α and χ are the load offset influence coefficient and sheath anomaly influence coefficient stored in the database respectively.

4. The big data-driven distribution network fault perception system according to claim 1, wherein The load operation time-series data includes the input current value at the node input end, the output current value at the node output end, and the operating temperature value of the node cable joint at several time points. The specific steps for obtaining the thermal load offset time-series index of each distribution node in the distribution network are as follows: Comprehensively analyze the load operation time-series data of each distribution node in the distribution network respectively to obtain the load operation status set of each distribution node in the distribution network. The load operation status set includes the input current operation index at the node input end, the output current operation index at the node output end, and the operating temperature index of the node cable joint; Obtain the operating temperature safety index of the node cable joint of each distribution node in the distribution network, and comprehensively analyze it in combination with the load operation status set of the corresponding distribution node to obtain the thermal load offset time-series index of each distribution node in the distribution network.

5. The fault perception system of the distribution network based on big data drive according to claim 4, wherein The specific steps for obtaining the load operation status set of each distribution node in the distribution network are as follows: Read the input current values at the node input ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the input current operation index at the node input ends of each distribution node in the distribution network; Read the output current values at the node output ends of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the output current operation index at the node output ends of each distribution node in the distribution network; Read the operating temperature values of the node cable joints of each distribution node in the distribution network at several time points, and conduct comprehensive analysis to obtain the operating temperature index of the node cable joints of each distribution node in the distribution network.

6. The fault perception system of the distribution network driven by big data according to claim 4, wherein, The specific formula for calculating the thermal load offset time series index of a certain distribution node in the distribution network is as follows: Among them, RfZ, RdL, CdL, YwZ, and YwA are the thermal load offset time series index, the input current operation index at the node input end, the output current operation index at the node output end, the operating temperature index of the node cable joint, and the operating temperature safety index of the node cable joint of a certain distribution node in the distribution network, respectively. β, δ1, and δ2 are the operating current adjustment factor, current influence coefficient, and temperature influence coefficient stored in the database, respectively.

7. The fault perception system for a distribution network driven by big data according to claim 1, wherein The voltage operation time series data includes the input voltage values of the node busbars at several time points. The specific steps for obtaining the voltage disturbance coupling time series index of each distribution node in the distribution network are as follows: Based on the voltage operation time series data of each distribution node in the distribution network, analyze the input voltage fluctuation index of the node busbars of each distribution node in the distribution network; Obtain the high-frequency disturbance energy values within the node setting areas of each distribution node in the distribution network, and conduct comprehensive analysis by combining with the input voltage fluctuation index of the node busbars of the corresponding distribution nodes respectively, to obtain the voltage disturbance coupling time series index of each distribution node in the distribution network.

8. The big data-driven distribution network fault perception system according to claim 7, wherein, The specific steps for analyzing the input voltage fluctuation index of the node busbars of each distribution node in the distribution network are as follows: Conduct comprehensive analysis on the input voltage values of the node busbars of each distribution node in the distribution network at several time points respectively, to obtain the input voltage mean value of the node busbars of each distribution node in the distribution network; Combine the input voltage values of the node busbars of each distribution node in the distribution network at several time points with the input voltage mean value of the node busbars of the corresponding distribution nodes respectively, and conduct comprehensive analysis to obtain the input voltage fluctuation index of the node busbars of each distribution node in the distribution network.

9. The fault perception system of the distribution network based on big data drive according to claim 4, characterized in that The grounding operation time series data includes the node grounding resistance values at several time points. The specific steps for obtaining the grounding risk time series index of each distribution node in the distribution network are as follows: Conduct comprehensive analysis on the node grounding resistance values of each distribution node in the distribution network at several time points respectively, to obtain the node grounding resistance mean value and the node grounding resistance change rate of each distribution node in the distribution network; Obtain the rated output current operation index at the node output end of each distribution node in the distribution network, and conduct comprehensive analysis by combining with the node grounding resistance mean value, the node grounding resistance change rate, and the output current operation index at the node output end of the corresponding distribution node respectively, to obtain the grounding risk time series index of each distribution node in the distribution network.

10. The fault perception system of the distribution network based on big data drive according to claim 1, characterized in that The sheath operation timing data includes the leakage current values of the node cables in the sheath at several time points. The specific steps to obtain the sheath abnormal adjustment timing index for each distribution node in the distribution network are as follows: Based on the leakage current values of the node cables in the sheath at several time points for each distribution node in the distribution network, analyze the maximum leakage current value and the average leakage current value of the node cables in the sheath for each distribution node in the distribution network respectively; Comprehensively analyze the maximum leakage current value and the average leakage current value of the node cables in the sheath for each distribution node in the distribution network respectively, and obtain the sheath abnormal adjustment timing index for each distribution node in the distribution network.

Images