A multi-stage cooperative fault detection method and system based on intelligent air switching

Abstract

The application relates to the technical field of electrical safety, and discloses a multi-stage cooperative fault detection method and system based on intelligent air switches, which comprises the following steps: deploying intelligent air switch terminals at each level of a power distribution network to synchronously collect multi-physical quantity time sequence signals; the intelligent air switch terminals calculate dynamic correlation entropy of the collected current signals, and perform local fault diagnosis according to the dynamic correlation entropy; each terminal performs cooperative communication, collects dynamic correlation entropy, constructs an entropy migration spectrum, and cooperatively traces the fault source; finally, the results of local diagnosis and cooperative tracing are comprehensively considered to perform corresponding hierarchical cooperative protection actions. The system comprises a collection module, an entropy calculation module, a local diagnosis module, a communication module, a cooperative tracing module and a protection execution module. Through multi-stage cooperative fault detection of intelligent air switches, the application realizes accurate identification, rapid response and state early warning of hidden faults, and improves the safety and reliability of a power distribution system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical safety, in particular to a multi-level cooperative fault detection method and system based on intelligent air switch. BACKGROUND

[0002] In modern power systems, air switch (referred to as air switch) is the most widely used basic protection electrical appliance, which undertakes the core function of cutting off overload and short-circuit current, protecting the safety of lines and equipment. With the promotion of smart grid construction and the popularity of industrial internet of things, the access of a large number of distributed new energy and high-frequency power electronic devices makes the operation environment and fault mode of low-voltage distribution system increasingly complex, which also puts forward higher technical requirements for the accuracy, rapidity and foresight of fault protection.

[0003] In the prior art, the traditional air switch mainly relies on the built-in thermal magnetic tripping mechanism to realize protection. The mechanism uses the principle of thermal bending of bimetallic strip due to overcurrent to deal with long-term overload, or uses the principle of strong magnetic attraction of electromagnet due to short-circuit large current to cut off the circuit. This protection mode based on a single physical quantity of current amplitude for passive response plays an important role in the past simple radial power supply network.

[0004] However, in the current complex power grid environment, the inherent defects of this traditional protection mechanism are increasingly prominent. First of all, this mechanism relies on a single current threshold as a criterion, and for hidden faults such as series arc and poor contact that do not cause significant overcurrent, its detection sensitivity is insufficient, it is difficult to identify effectively, and it is easily disturbed by harmonic current to cause misoperation or refusal to operate. First of all, the mechanism relies on mechanical structure action, and its response speed is usually tens of milliseconds, which lags behind the formation time of the first peak of short-circuit current (usually within 10 milliseconds), so it cannot effectively suppress the energy impact in the early stage of fault. First of all, due to the lack of sensing function of temperature and other physical quantities, the mechanism cannot monitor the gradual degradation process caused by contact aging, virtual connection and other factors, and does not have the ability of state early warning, which may lead to the existence of electrical fire hazards.

[0005] Therefore, the present application proposes a multi-level cooperative fault detection method and system based on intelligent air switch to solve the problems of the prior art. SUMMARY

[0006] The purpose of the present application is to provide a multi-level cooperative fault detection method and system based on intelligent air switch, which solves the problems that the existing air switch has a single protection criterion, slow response speed and lacks a cooperative mechanism, so it is difficult to effectively identify hidden faults such as series arc, cannot quickly cut off the early fault current, and does not have the ability of state early warning.

[0007] To achieve the above object, the present application is implemented by the following technical solutions:

[0008] The first aspect of the present application provides a multi-level cooperative fault detection method based on intelligent air switch, which comprises:

[0009] Step one: synchronously collecting multi-physical quantity time series by the intelligent air switch terminal deployed at each level of the power distribution network.Step two: reconstructing phase space by the intelligent air switch terminal based on the collected current time series, and calculating the dynamic correlation entropy reflecting the dynamic behavior of the power circuit.Step three: performing local fault diagnosis by the intelligent air switch terminal based on the dynamic correlation entropy.Step four: performing cooperative communication between the intelligent air switch terminals, constructing entropy migration spectrum based on the dynamic correlation entropy of each air switch terminal, and cooperatively tracing the fault according to the entropy migration spectrum.Step five: performing hierarchical protection actions according to the results of the local fault diagnosis and the cooperative tracing.

[0010] Preferably, in step one, after completing time synchronization by network time protocol or GPS module, the intelligent air switch terminal deployed at each level of the power distribution network collects the multi-physical quantity time series, wherein the multi-physical quantity time series includes current time series, voltage time series, contact temperature sequence, and bus temperature sequence.

[0011] In one specific embodiment, the step of reconstructing phase space by the intelligent air switch terminal based on the collected current time series and calculating the dynamic correlation entropy in step two comprises:

[0012] Reconstructing the current time series I(i) into a state vector Y in m-dimensional phase space by time delay embedding method i to construct an attractor capable of representing the dynamic behavior of the power circuit:

[0013] Y i =[I(i),I(i+τ),…,I(i+(m-1)τ)];

[0014] In the formula, I(i) represents the current sampling value at discrete time index i; m is the embedding dimension; τ is the time delay;

[0015] Calculating the correlation integral based on the state vector Y i , and then determining the dynamic correlation entropy as a quantitative indicator of the complexity of the dynamic behavior of the power circuit according to the rate of change of the correlation integral with the embedding dimension m.

[0016] Preferably, the step of performing local fault diagnosis by the intelligent air switch terminal based on the dynamic correlation entropy in step three comprises performing at least one of the following diagnostic logics:

[0017] 1) calculating the first order difference of the dynamic correlation entropy, and determining a short circuit or parallel arc fault when the first order difference exceeds a first preset threshold value;

[0018] 2) judging the amplitude of the dynamic correlation entropy, and determining a series arc fault when the amplitude continuously exceeds a second preset threshold value;

[0019] 3) combining the dynamic correlation entropy and the contact temperature sequence and the bus temperature sequence obtained in the synchronous sampling of the multi-physical quantity time sequence signal to determine a gradual fault when both the dynamic correlation entropy and the differential temperature obtained by calculating the difference between the contact temperature sequence and the bus temperature sequence continuously increase.

[0020] In one specific embodiment, the step of constructing the entropy migration spectrum in step four includes:

[0021] When any of the intelligent air switch terminals detects a preset change in the dynamic correlation entropy thereof, the intelligent air switch terminal generates an entropy migration event, and broadcasts event information including a time stamp of the event and an amplitude of the entropy to other intelligent air switch terminals in the power distribution network through the communication module of the intelligent air switch terminal;

[0022] The superior intelligent air switch terminal receives and collects a series of entropy migration events caused by the same physical disturbance to form the entropy migration spectrum.

[0023] Further, the step of cooperatively tracing the fault according to the entropy migration spectrum in step four includes:

[0024] calculating a time delay spectrum and an attenuation factor spectrum based on the time stamp and the amplitude of the entropy value of each entropy migration event in the entropy migration spectrum;

[0025] and determining the fault type from a preset fault feature spectrum library according to the combination characteristics of the time delay spectrum and the attenuation factor spectrum.

[0026] Specifically, the calculation method of the attenuation factor spectrum is:

[0027]

[0028] wherein, A ems is the attenuation factor spectrum; K 2,s is the amplitude of the entropy value of the fault source node; K 2,s-1 to K 2,1 are the amplitudes of the entropy values detected by the intelligent air switch terminals at each level from the fault source node to the upper level in the power transmission direction, respectively;

[0029] and determining a fault type according to characteristics of the attenuation factor spectrum:

[0030] when the attenuation factor spectrum presents a rapid attenuation characteristic, determining a series type fault;

[0031] when the attenuation factor spectrum presents a slight attenuation characteristic, determining a parallel type fault.

[0032] In one specific embodiment, the step of performing a hierarchical protection action in step five includes:

[0033] when the result of the local fault diagnosis is a short circuit or parallel arc fault, directly driving a power electronic switch built in the intelligent air switch terminal making the diagnosis to perform a rapid cut-off of the circuit;

[0034] when the result of the collaborative tracing is a parallel type fault, issuing a pre-blocking instruction by the upper level intelligent air switch terminal that has collected the entropy migration spectrum to a higher level intelligent air switch terminal to which the upper level intelligent air switch terminal belongs, so as to make the power electronic switch of the higher level intelligent air switch terminal enter a standby state and perform collaborative backup protection;

[0035] and, after the power electronic switch performs the rapid cut-off, driving an electromagnetic tripping mechanism built in the intelligent air switch terminal that has performed the cut-off action in a delayed manner to complete a final physical disconnection.

[0036] Preferably, the method further includes:

[0037] after performing the hierarchical protection action, or when it is determined according to the local fault diagnosis that the fault is a gradual change type fault, uploading fault related data including the result of the local fault diagnosis, the result of the collaborative tracing, the entropy migration spectrum and the multi-physical quantity time sequence signal to a cloud platform for archiving.

[0038] The second aspect of the present application provides a multi-level collaborative fault detection system based on intelligent air switch, which includes a plurality of intelligent air switch terminals deployed at each level of a power distribution network, and each functional module of the intelligent air switch terminals includes:

[0039] an entropy calculation module for phase space reconstruction on the current time sequence collected by the acquisition module and calculating a dynamic correlation entropy reflecting dynamic behavior of the power circuit;

[0040] a local diagnosis module for performing local fault diagnosis based on the dynamic correlation entropy;

[0041] A communication module is configured to enable the intelligent air circuit breaker terminal to communicate with other intelligent air circuit breaker terminals;

[0042] A cooperative traceability module is configured to construct an entropy migration spectrum based on the dynamic correlation entropy of each terminal through the cooperative communication, and to cooperatively trace the fault according to the entropy migration spectrum;

[0043] A protection execution module is configured to execute a hierarchical protection action according to the diagnosis result of the local diagnosis module and the traceability result of the cooperative traceability module.

[0044] In summary, the present application has at least one of the following beneficial technical effects:

[0045] 1. The present application introduces dynamic correlation entropy as the core index of fault diagnosis, and performs phase space reconstruction and quantitative analysis on the collected current time series, which can accurately capture fault characteristics from the dimension of signal complexity. Compared with the traditional method of relying only on current amplitude threshold, the present application can effectively identify and distinguish hidden non-overcurrent faults such as series arc and poor contact from normal load fluctuations and harmonic interference, fundamentally solving the problem of fault missed detection and misoperation caused by single criterion of traditional circuit breaker, and significantly improving the accuracy and breadth of fault identification.

[0046] 2. The present application realizes rapid response and accurate isolation of faults by constructing an entropy migration spectrum for multi-level cooperative traceability and executing protection actions by power electronic switches. Using the group perception ability of multiple intelligent air circuit breaker terminals, the method can accurately locate the fault source and determine its root type, and then guide the power electronic switch to perform high-speed cutting at microsecond level. This method not only effectively suppresses the impact of the first peak of short-circuit current, but also avoids fault expansion and step-out tripping. Compared with the traditional mechanical release mechanism protection with slow response and no selectivity, the safety and power supply reliability of the system are greatly enhanced.

[0047] 3. The present application fuses multi-physical quantity time series such as contact temperature, and uses dynamic correlation entropy to quantify the subtle changes of system state, so as to realize the forward-looking warning ability of circuit breaker. By continuously tracking and analyzing the gradual change trend of entropy value and the abnormal rise of temperature, early warning can be performed before the gradual fault such as equipment insulation aging and virtual contact of contact point evolves into serious electrical fire. This makes the intelligent air circuit breaker upgrade from a pure post-fault protection device to an intelligent terminal capable of system health state monitoring and risk prediction, significantly improving the operation efficiency and intrinsic safety level of the power distribution system. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 It is a flowchart of the multi-level cooperative fault detection method of the present application;

[0049] Figure 2 Fig. 1 is a schematic diagram of the system architecture of the present application;

[0050] Figure 3 Fig. 2 is a schematic diagram of the hardware structure of the intelligent air switch terminal of the present application;

[0051] Figure 4 Fig. 3 is a functional module block diagram of the intelligent air switch terminal of the present application.

[0052] In the present application, 100 is the intelligent air switch terminal; 101 is the microcontroller unit; 102 is the multi-physical quantity sensing unit; 103 is the protection execution unit; 104 is the communication unit; 105 is the storage unit; 110 is the acquisition module; 120 is the entropy calculation module; 130 is the local diagnosis module; 140 is the communication module; 150 is the collaborative traceability module; 160 is the protection execution module; 170 is the data management module; and 200 is the cloud platform. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments herein are only used to explain the present application, and do not limit the present application.

[0054] The specific embodiments of the present application will be further described below with reference to the accompanying drawings. Figure 1 - the accompanying drawings Figure 4 The present application will be further described in detail.

[0055] Reference is made to the accompanying drawings Figures 2-4 The present application provides a multi-level collaborative fault detection system based on intelligent air switch. The system is applied to a power distribution network, for example, a three-level power distribution structure including a general power distribution box, a distribution power distribution box and a terminal user power distribution box. The system includes a plurality of intelligent air switch terminals 100 deployed at each level of the power distribution network, and an optional cloud platform 200 for data archiving and analysis. Each intelligent air switch terminal 100 exchanges data with each other and with the cloud platform 200 through a communication network.

[0056] Each intelligent air switch terminal 100 in the embodiment of the present application is an independent physical device, which integrates hardware circuit and software module running thereon to perform the functions of fault detection and protection. The hardware structure of the intelligent air switch terminal 100 includes:

[0057] The microcontroller unit (MCU) 101, the multi-physical quantity sensing unit 102, the protection execution unit 103, the communication unit 104 and the storage unit 105. The MCU 101 as the control core is electrically connected and data-interacts with the sensing unit 102, the protection execution unit 103, the communication unit 104 and the storage unit 105 respectively.

[0058] The storage unit 105 stores program instructions, and the MCU 101 implements the coordinated work of multiple functional modules by executing the program instructions.

[0059] The intelligent air switch terminal 100 specifically includes the following functional modules:

[0060] The acquisition module 110 is configured to control the multi-physical quantity sensing unit 102 to synchronously acquire the physical quantities of the circuit. The multi-physical quantity sensing unit 102 includes a current sensor, a voltage sampling circuit, and a temperature sensor. The acquisition module 110 synchronously acquires the current time series, the voltage time series, and the temperature sequence installed on the circuit breaker contact and the busbar at a preset sampling frequency.

[0061] The entropy calculation module 120 is configured to receive the current time series acquired by the acquisition module 110. The module performs a phase space reconstruction operation on the current time series, and calculates the correlation integral based on the reconstructed state vector to determine the dynamic correlation entropy. The dynamic correlation entropy is used to quantify the complexity of the current power circuit dynamic behavior.

[0062] The local diagnosis module 130 is configured to receive and analyze the dynamic correlation entropy output by the entropy calculation module 120. The module determines the fault type of the local circuit by calculating the first-order difference of the dynamic correlation entropy, judging its amplitude, and combining the temperature sequence acquired by the acquisition module 110. The determination result includes short circuit, parallel arc fault, series arc fault, or gradual fault.

[0063] The communication module 140 is composed of the communication unit 104 and its driver program. The module is configured to establish a communication link between the intelligent air switch terminal 100 and other intelligent air switch terminals 100, and between the intelligent air switch terminal 100 and the cloud platform 200. It is responsible for broadcasting the entropy migration event, issuing and receiving the pre-blocking instruction, and uploading the fault-related data.

[0064] The collaborative tracing module 150 is configured to receive the entropy migration events broadcast by other intelligent air switch terminals in the power distribution network through the communication module 140. The module collects a series of entropy migration events from the same physical disturbance, constructs an entropy migration spectrum, and calculates the time delay spectrum and the attenuation factor spectrum based on the spectrum. Finally, by matching the preset fault characteristic spectrum library, the source location and specific type of the fault are determined.

[0065] The protection execution module 160 is configured to issue control instructions to the protection execution unit 103 according to the diagnosis result of the local diagnosis module 130 or the tracing result of the collaborative tracing module 150. The protection execution unit 103 internally includes a power electronic switch and an electromagnetic tripping mechanism. The module can drive the power electronic switch to perform microsecond-level fast cutting, or drive the electromagnetic tripping mechanism to perform millisecond-level physical breaking.

[0066] In addition, the intelligent air switch terminal 100 in the embodiment further comprises a data management module 170. The module is used for packaging fault data including local fault diagnosis results, collaborative traceability results, entropy migration spectrum and multi-physical quantity time sequence signals after the end of the fault handling process, and uploading the fault data to the cloud platform 200 for archiving through the communication module 140.

[0067] Referring to the accompanying drawings Figure 1 The method is collaboratively executed by a plurality of intelligent air switch terminals 100 deployed in the power distribution network, and specifically can include the following steps:

[0068] S100, the intelligent air switch terminals deployed at each level of the power distribution network perform synchronous acquisition of multi-physical quantity time sequence signals.

[0069] S200, the intelligent air switch terminal reconstructs the current time sequence acquired by the intelligent air switch terminal in phase space, and calculates a dynamic correlation entropy reflecting the dynamic behavior of the power circuit.

[0070] S300, the intelligent air switch terminal performs local fault diagnosis based on the dynamic correlation entropy.

[0071] S400, the intelligent air switch terminals perform collaborative communication, construct an entropy migration spectrum based on the dynamic correlation entropy of each air switch terminal, and perform collaborative traceability of the fault according to the entropy migration spectrum.

[0072] S500, according to the results of local fault diagnosis and collaborative traceability, a hierarchical protection action is performed.

[0073] The method further includes:

[0074] S600, fault data archiving. After performing the hierarchical protection action, or when it is determined that the fault is a gradual change, the data management module 170 packages the local fault diagnosis results, collaborative traceability results, entropy migration spectrum and multi-physical quantity time sequence signals related to the event, and uploads the data to the cloud platform 200 for archiving through the communication module 140.

[0075] The starting step of the multi-level collaborative fault detection method of the embodiment is the synchronous acquisition of multi-physical quantity time sequence signals in the intelligent air switch terminal 100 deployed at each level of the power distribution network (S100). This step is the data basis for all subsequent fault diagnosis and collaborative traceability, and the accuracy and synchronization of its execution directly determines the performance of the entire system. This step is executed by the acquisition module 110 inside each intelligent air switch terminal 100, and the communication module 140 provides time synchronization support, and the specific implementation process is as follows.

[0076] First, to ensure that the data collected by multiple smart air switch terminals 100 distributed in different physical locations have a unified time reference, the collection module 110 will trigger the communication module 140 to perform high-precision time synchronization before data collection. In an embodiment, the communication module 140 periodically communicates with a superior gateway deployed in the local network or a time server on the public Internet through the Network Time Protocol (NTP) to obtain a standard timestamp and calibrate the local real-time clock (RTC) of its internal microcontroller (MCU) 101, ensuring that the time error is controlled within milliseconds. In scenarios with higher requirements for time accuracy, the communication module 104 can integrate a Global Positioning System (GPS) receiver. By receiving GPS satellite signals, especially the PPS (Pulse Per Second) signal contained therein, the synchronization error of the local clock can be reduced to the microsecond level. High-precision time synchronization is a prerequisite for accurately locating the fault source by calculating the time delay of the entropy migration event in the subsequent collaborative traceability step S400.

[0077] After completing time synchronization, the collection module 110 synchronously samples multiple key physical quantities through the multi-physical quantity sensing unit 102 at a fixed, pre-set sampling frequency (e.g., 6.4 kHz). The multi-physical quantity time series specifically include: current time series, voltage time series, contact temperature series, and bus temperature series. Synchronous sampling is adopted to ensure that at any sampling instant, the four types of data accurately correspond to the same circuit state at the same time.

[0078] Specifically, for the collection of current time series, the multi-physical quantity sensing unit 102 uses high-bandwidth Hall effect sensors or high-precision shunt resistors to continuously measure the current flowing through the main circuit of the circuit breaker, and the output analog signal is converted to a digital sequence via an analog-to-digital converter (ADC). This current time series is the core input for calculating dynamic correlation entropy.

[0079] For the collection of voltage time series, a resistive voltage dividing circuit is used to proportionally reduce the 220V or 380V line voltage to the safe input range (e.g., 0-3.3V) of the internal ADC of the microcontroller (MCU) 101, thereby realizing real-time sampling of the voltage waveform. The voltage sequence is mainly used for auxiliary judgment, such as distinguishing between voltage sag on the power grid side and load failure on the user side.

[0080] For the collection of contact temperature sequence and busbar temperature sequence, high-precision negative temperature coefficient (NTC) thermistor or PT100 platinum thermistor is closely attached to the vicinity of the moving and static contacts inside the intelligent air switch terminal 100 and the main incoming line busbar through a heat-conducting medium, respectively. The acquisition module 110 converts the corresponding temperature by measuring the resistance value or the voltage division value. The acquisition of contact temperature is to monitor the abnormal temperature rise caused by arc ablation or contact pressure drop, and the acquisition of busbar temperature is to provide a reference. By calculating the difference between the contact temperature and the busbar temperature, the interference of overall environmental temperature rise caused by load current change can be excluded, so as to realize accurate diagnosis of gradual faults.

[0081] The acquisition module 110 binds the four data points of current, voltage, contact temperature and busbar temperature obtained by each round of synchronous sampling with the current high-precision time stamp to form a data frame, and stores it in the buffer area of the storage unit 105 for subsequent calling by the entropy calculation module 120 and the local diagnosis module 130.

[0082] After the collection of multi-physical quantity time sequence signals is completed in step S100, the process enters step S200, and the entropy calculation module 120 deployed in each intelligent air switch terminal 100 processes the collected current time sequence to calculate the dynamic correlation entropy. This step is the core of the fault diagnosis method of the present application, and its purpose is to convert the one-dimensional and seemingly random current signal into an index that can accurately quantify the complexity of the dynamic behavior of the power system. The specific implementation process includes the following two stages.

[0083] Phase space reconstruction of current time sequence, the entropy calculation module 120 first reads the current time sequence collected and stored by the acquisition module 110 from the storage unit 105. The reason for choosing the current time sequence as the analysis object is that the current signal directly reflects the behavior of the load and the state of the line, and is most sensitive to disturbances such as short circuit, arc, poor contact and other faults.

[0084] To reveal the system dynamics characteristics hidden behind the one-dimensional time sequence, the entropy calculation module 120 uses the time delay embedding method to reconstruct the phase space of the current time sequence. This method expands the one-dimensional time sequence into a high-dimensional phase space to form an attractor that can represent the dynamic behavior of the power circuit. The reconstruction process is performed by the following formula: i = [I(i), I(i+τ), …, I(i+(m-1)τ)];

[0085] In the formula, I(i) refers to the current sampling value at discrete time index i; m is the embedding dimension; τ is the time delay, which refers to the time interval between consecutive components when constructing the state vector; Y iIt refers to a state vector in the reconstructed m-dimensional phase space at time index i, and the set of all state vectors {Y}. i These constitute the phase space attractor of the system.

[0086] The selection of the embedding dimension *m* and the time delay *τ* is crucial for the effectiveness of phase space reconstruction. In this embodiment, these two parameters are determined as follows: the time delay *τ* is determined by calculating the autocorrelation function or average mutual information function of the current time series. Typically, the time delay corresponding to the first minimum point of the function is selected to ensure that the linear independence or nonlinear correlation of the components in the state vector is minimized. The embedding dimension *m* is determined using the False Nearest Neighbors (FNN) method. By continuously increasing the value of *m* and calculating the proportion of false nearest neighbors in the phase space, the value of *m* corresponding to the first drop to zero or a sufficiently small threshold is the optimal embedding dimension, ensuring that the attractor is fully unfolded in the reconstruction space without false overlap.

[0087] The calculation of dynamic correlation entropy is performed after the phase space is reconstructed and the set of state vectors {Y} is obtained. i After that, the entropy calculation module 120 performs subsequent calculations. Specifically, this module is based on the state vector Y. i The correlation integral is calculated, and then the dynamic correlation entropy is determined based on the rate of change of the correlation integral with the embedding dimension m, so as to serve as a quantitative indicator of the complexity of the dynamic behavior of the power circuit.

[0088] First, calculate the correlation integral C. m (r). For a given embedding dimension m and a small distance radius r, the correlation integral C m (r) is defined as all pairs of state vectors (Y, Y) in the m-dimensional phase space. i ,Y j The probability that the distance between () and () is less than r. The formula for calculating this is:

[0089]

[0090] In the formula, N is the total number of state vectors; ||·|| represents the distance between two vectors, usually using the Euclidean norm; Θ(·) is the Heaviside step function, which has a value of 1 when its independent variable is greater than or equal to 0, and 0 otherwise.

[0091] Subsequently, the dynamic correlation entropy, or K2 entropy, referred to in this invention, is determined by analyzing the rate of change of the correlation integral with the embedding dimension m. For a deterministic chaotic system, its correlation integral exhibits the following relationship as the embedding dimension increases from m to m+1:

[0092]

[0093] where h2 is the correlation entropy of the system. The formula of dynamic correlation entropy can be derived as follows:

[0094]

[0095] The entropy calculation module 120 calculates the correlation integral C m (r) and C m+1 (r) of the same set of data respectively when the embedding dimension is m and m+1, and then substitutes them into the above formula, so as to obtain the dynamic correlation entropy K2. The K2 entropy value is a scalar, which quantifies the rate of orbit divergence in the phase space of the system, i.e. the complexity of the dynamic behavior of the system. After the calculation is completed, the dynamic correlation entropy value is output to the local diagnosis module 130 for subsequent fault diagnosis.

[0096] After the dynamic correlation entropy is calculated in step S200, the process enters step S300, in which the local diagnosis module 130 in each intelligent air switch terminal 100 performs real-time fault diagnosis on the dynamic correlation entropy sequence output by the entropy calculation module 120. The purpose of this step is to make a quick and accurate preliminary judgment on the electrical state of the current circuit, and to provide decision basis for subsequent protection actions or cooperative tracing. The specific diagnosis logic is as follows.

[0097] The local diagnosis module 130 receives a continuous dynamic correlation entropy (K2) time sequence, and performs at least one of the following diagnosis logics in parallel.

[0098] The first diagnosis logic is for short circuit or parallel arc fault. The physical feature of such fault is that the line current suddenly changes dramatically in a very short time, and the amplitude quickly rises. This mutation is reflected in the current time sequence, which will instantly and sharply increase the complexity and uncertainty of the dynamic behavior, thereby causing a large and pulsed jump in the dynamic correlation entropy K2 value output by the entropy calculation module 120. In order to capture this feature, the local diagnosis module 130 performs first-order difference calculation on the received K2 sequence, i.e. ΔK2(t) = K2(t) - K2(t-1). When the absolute value of the first-order difference |ΔK2(t)| exceeds a first preset threshold value preset in the storage unit 105, it is determined that a short circuit or parallel arc fault has occurred. The setting of the first preset threshold value is based on the experimental data analysis of the load switching (such as motor starting) under normal working conditions and the entropy value change rate under real fault working conditions, to ensure the reliability of the diagnosis.

[0099] The second diagnostic logic is for series arc fault. The physical feature of series arc fault is that it creates an unstable, repeatedly arcing and extinguishing plasma channel in the circuit. This process is highly random and nonlinear, introducing a large amount of broadband chaotic noise into the current waveform, but the total current amplitude does not necessarily exceed the rated value. This persistent chaotic state keeps the complexity of the current time series at an abnormally high level. Therefore, the amplitude of its dynamic correlation entropy K2 deviates from the low-level stable state under normal load and instead fluctuates within a higher numerical interval. The local diagnostic module 130 determines that a series arc fault has occurred when the amplitude of K2 is always higher than a second preset threshold value for a certain duration (e.g., 50 milliseconds). The duration judgment is set to distinguish it from transient disturbances that are not persistent.

[0100] The third diagnostic logic is for gradual faults, such as loose electrical connections or aging contacts. The physical feature of such faults is that the contact resistance of the fault point slowly increases over time. On the one hand, the increased contact resistance will generate additional heat at the contact point according to Joule's law (P = I 2 R), causing the contact temperature to abnormally and persistently rise compared to the surrounding environment (e.g., busbar). On the other hand, the nonlinear change in contact resistance also causes weak and gradually increasing distortion to the current waveform, slowly increasing its dynamic complexity. To diagnose such faults, the local diagnostic module 130 combines entropy analysis and temperature analysis. The module first receives the contact temperature sequence and busbar temperature sequence provided by the acquisition module 110, and obtains the differential temperature sequence by calculating the difference between the two. Then, the module simultaneously performs trend analysis on the dynamic correlation entropy sequence and the differential temperature sequence. When both the dynamic correlation entropy and the differential temperature show a persistent and monotonically increasing trend, it is determined that a gradual fault has occurred. The application of this double criterion effectively eliminates the temperature rise interference caused by normal changes in load current, significantly improving the accuracy of gradual fault diagnosis.

[0101] After performing local fault diagnosis at step S300, or when a serious fault that may affect the upper circuit is diagnosed, the process proceeds to step S400 to perform collaborative tracing based on entropy migration spectrum. The core purpose of this step is to use the group perception capability of multiple intelligent air switch terminals 100 in the power distribution network to accurately locate and identify the source location and root type of the fault, rather than being limited to a single node judgment. The specific implementation process is as follows.

[0102] Construction of entropy migration spectrum

[0103] The first step of this phase is to define the generation and broadcasting of entropy migration events. When the local diagnostic module 130 of any smart air switch terminal 100 detects a preset change in its dynamic correlation entropy (for example, in the diagnostic logic of S300, the first-order difference of the entropy value exceeds the first preset threshold or the amplitude of the entropy value continuously exceeds the second preset threshold), the terminal generates an entropy migration event. The event is encapsulated into a standard data packet and broadcast to other smart air switch terminals in the power distribution network through the communication module 140 of the smart air switch terminal. The content of the data packet at least includes: high-precision timestamp of the event occurrence, unique device ID of the smart air switch terminal generating the event, and the amplitude of the dynamic correlation entropy value measured at that moment.

[0104] Subsequently, the smart air switch terminal 100 deployed at the upper-level power distribution node (such as the distribution box or the main distribution box) is responsible for receiving and collecting a series of entropy migration events caused by the same physical disturbance through the collaborative tracing module 150. The module sets a time window (for example, 50 milliseconds), and all entropy migration events received within this time window and from different terminals downstream are grouped together to form an entropy migration spectrum. The entropy migration spectrum is a data set that records the entropy response of a single fault disturbance at different nodes of the power distribution network, specifically in the form of a structure body containing a list of {device ID, timestamp, entropy value amplitude}.

[0105] After the entropy migration spectrum is constructed, the collaborative tracing module 150 of the upper-level smart air switch terminal performs calculations and analyses based on the spectrum to collaboratively trace the fault.

[0106] Firstly, the module calculates the time delay spectrum and the attenuation factor spectrum based on the timestamps and entropy value amplitudes of each entropy migration event in the entropy migration spectrum. The calculation method of the time delay spectrum is as follows: in the entropy migration spectrum, the event node with the earliest timestamp is identified as the initial fault source node, and the time delay of the timestamps of all other events in the spectrum relative to the timestamp of the source node is calculated. The time delay spectrum reveals the propagation path and sequence of the fault disturbance along the electrical line.

[0107] At the same time, the module calculates the attenuation factor spectrum. The collaborative tracing module 150 determines the fault source node based on the time delay spectrum (its entropy value amplitude is denoted as K 2,s ), and the entropy value amplitudes (denoted as K 2,s-1 to K 2,1 ) detected by the smart air switch terminals at each level from the source in the direction of power transmission, and calculates the attenuation factor spectrum, whose formula is:

[0108]

[0109] where A emsThe attenuation factor spectrum is a vector composed of multiple ratios, which characterizes the attenuation characteristics of the fault disturbance energy during its upward propagation.

[0110] Finally, the collaborative tracing module 150 determines the fault type according to the characteristics of the attenuation factor spectrum. The physical basis of this determination lies in the significant differences in energy propagation patterns between different types of faults. When the attenuation factor spectrum exhibits rapid attenuation characteristics (i.e., each ratio in the spectrum is much smaller than 1), it is determined to be a series fault. This is because the energy of faults such as series arcs is mainly consumed at the fault point itself, and the disturbance propagating upward is greatly weakened. When the attenuation factor spectrum exhibits slight attenuation characteristics (i.e., each ratio in the spectrum is close to 1), it is determined to be a parallel fault. This is because short circuits and other parallel faults provide new low-impedance discharge channels for the system, and the fault current will propagate upward along the transmission path with minimal attenuation.

[0111] Finally, the collaborative tracing module 150 combines the fault propagation path determined by the time delay spectrum with the fault type determined by the attenuation factor spectrum to form a combined feature. According to the combined feature of the time delay spectrum and the attenuation factor spectrum, the module matches and determines the fault type from the fault feature spectrum library pre-stored in the storage unit 105. The final tracing result obtained by this matching process will be passed to the protection execution module 160 for the execution of more targeted collaborative protection actions.

[0112] After completing local fault diagnosis and collaborative tracing in steps S300 and S400, respectively, the flow proceeds to step S500 to execute hierarchical collaborative protection actions. The purpose of this step is to execute the most appropriate circuit shutdown strategy based on the accurate diagnosis and tracing results obtained in the previous steps, in order to achieve rapid response, precise isolation, and systematic protection of the fault. This step is completed by the protection execution module 160 in each intelligent air switch terminal 100 by controlling its protection execution unit 103.

[0113] The protection execution unit 103 in this embodiment includes a power electronic switch (such as an IGBT or SiC-MOSFET module) and a traditional electromagnetic tripping mechanism. The former is used to perform microsecond-level fast shutdown, and the latter is used to achieve millisecond-level permanent physical isolation. The protection execution module 160 executes hierarchical protection actions based on the results of local fault diagnosis and collaborative tracing, which specifically includes the following cases:

[0114] The first case is to perform fast tripping for local serious faults. When the result of local fault diagnosis is short circuit or parallel arc fault, the result is directly transmitted by the local diagnosis module 130 to the protection execution module 160. The power electronic switch built-in the intelligent air switch is directly driven by the intelligent air switch terminal making the diagnosis to perform fast tripping of the circuit. Specifically, the protection execution module 160 immediately sends a shutdown control signal to the gate drive circuit of the power electronic switch. Since the response time of the power electronic switch is in the order of microseconds, this action can complete the tripping of the circuit before the short circuit current rises to its first peak (usually within 10 ms of the power frequency cycle), thereby effectively suppressing the release and impact of fault energy.

[0115] The second case is to perform backup protection for the result of cooperative tracing. When the result of cooperative tracing is parallel fault, the result is obtained by the cooperative tracing module 150 of the upper intelligent air switch terminal which collects the entropy migration spectrum. At this time, the upper intelligent air switch terminal which collects the entropy migration spectrum issues a pre-blocking instruction to the intelligent air switch terminal of the next level to which the upper intelligent air switch terminal itself belongs. The instruction is sent through the communication module 140. The protection execution module 160 of the intelligent air switch terminal of the next level receives the instruction and analyzes the instruction to make the power electronic switch of the intelligent air switch terminal of the next level enter a standby state to perform cooperative backup protection. The standby state means that the shutdown threshold of the power electronic switch is temporarily lowered or the drive circuit thereof is pre-charged, and once the current of the current level exceeds a certain threshold or a trip confirmation instruction of the next level is received, the power electronic switch can be instantly actuated, thereby constituting fast and reliable backup support for the protection of the next level.

[0116] The third case is to perform final physical disconnection to ensure safety. After the power electronic switch performs fast tripping, whether based on local diagnosis or cooperative instruction, it only completes the electrical shutdown of the circuit, and the physical contacts are not disconnected. To achieve permanent electrical isolation, the electromagnetic tripping mechanism built-in the intelligent air switch is time-delay driven by the intelligent air switch to complete the final physical disconnection. Specifically, after the power electronic switch is turned off, the protection execution module 160 starts a preset delay (for example, 20 ms). After the delay ends, the module drives a coil to generate sufficient electromagnetic force to actuate the mechanical tripping mechanism to physically separate the moving and static contacts and form a visible break point. The two-stage disconnection mode of "electronic first and mechanical second" combines the speed of electronic switch and the isolation reliability of mechanical switch.

[0117] Referring to the accompanying drawings Figure 1 The multi-level cooperative fault detection method of the embodiment of the present application further includes the step S600 of archiving and analyzing fault data. This step is optional, but plays an important role in improving the operation and maintenance level of the entire power distribution system and continuously optimizing the fault diagnosis model.

[0118] There are two execution occasions for this step: after the hierarchical protection action is executed, that is, when a complete fault event has been disposed of; or when a gradual fault is determined according to the local fault diagnosis, that is, when the system issues a warning but has not yet executed a trip action.

[0119] After any of the above cases occurs, the data management module 170 inside the intelligent air switch terminal 100 is activated. This module is responsible for collecting all data related to this event and organizing it into a structured fault data record. The fault data, including the results of local fault diagnosis, collaborative traceability, entropy migration spectrum, and multi-physical quantity time sequence signals, are uploaded to the cloud platform for archiving.

[0120] Specifically, the data packet content collected by the data management module 170 includes:

[0121] Event metadata: including event unique ID, intelligent air switch terminal ID where the event occurred, high-precision timestamp of event occurrence, and final disposal action (such as trip, warning).

[0122] Diagnosis and traceability results: local fault diagnosis results output by the local diagnosis module 130 (for example, determining a series arc), and collaborative traceability results output by the collaborative traceability module 150 (for example, determining that the fault source is a certain device downstream, and the type is a parallel fault).

[0123] Entropy migration spectrum data: if the event triggers collaborative traceability, it contains the complete entropy migration spectrum collected by the upper node, that is, a data structure containing a list of {device ID, timestamp, entropy value amplitude}.

[0124] Original time sequence signals: containing all multi-physical quantity time sequence signals collected by the acquisition module 110 within a certain period of time before and after the fault occurs (for example, 200 milliseconds before the fault point to 100 milliseconds after the fault point), that is, high-resolution current time series, voltage time series, contact temperature sequence, and bus temperature sequence.

[0125] The data management module 170 packages and compresses all the above data, and then uploads the data packet to the preset cloud platform 200 via the Internet through the communication module 140. After receiving the data, the cloud platform 200 parses and persistently stores it, forming a fault database that can be queried and analyzed for a long time. This database not only provides complete on-site data for post-fault manual analysis, but also provides a data basis for using machine learning algorithms to deeply mine massive historical fault data, continuously optimizing threshold parameters and fault feature spectrum libraries in the local diagnosis module 130 and the collaborative traceability module 150.

[0126] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A multi-stage collaborative fault detection method based on intelligent air switching, characterized in that, The application relates to a power distribution network fault diagnosis method and system. The method comprises the following steps: S1, synchronously collecting multi-physical quantity time series signals by intelligent air break switch terminals deployed at each level of the power distribution network; S2, reconstructing phase space of the current time series collected by the intelligent air break switch terminals, and calculating dynamic correlation entropy reflecting dynamic behavior of the power loop; S3, performing local fault diagnosis by the intelligent air break switch terminals based on the dynamic correlation entropy; S4, performing collaborative communication among the intelligent air break switch terminals, constructing entropy migration spectrum based on the dynamic correlation entropy of each air break switch terminal, and collaboratively tracing the fault according to the entropy migration spectrum; S5, performing hierarchical protection actions according to the results of the local fault diagnosis and the collaborative tracing; by time-delay-embedding method, the current time series is reconstructed into a state vector in a phase space to construct an attractor capable of representing the dynamic behavior of the power circuit under study: ； wherein denotes the current sample value at discrete time index ; is the embedding dimension; is the time delay; based on the state vector computing a correlation integral, and determining the dynamic correlation entropy as a quantitative index of the dynamic behavior complexity of the power loop under study, according to a rate of change of the correlation integral with the embedding dimension computing a correlation integral, and determining the dynamic correlation entropy as a quantitative index of the dynamic behavior complexity of the power loop under study, according to a rate of change of the correlation integral with the embedding dimension S2 further comprises the following steps: S21, reconstructing phase space of the current time series collected by the intelligent air break switch terminals, and calculating dynamic correlation entropy reflecting dynamic behavior of the power loop; S4 further comprises the following steps:

2. The multi-stage collaborative fault detection method based on intelligent air switching according to claim 1, characterized in that, When any of the intelligent air break switch terminals detects that the dynamic correlation entropy thereof has a preset change, the intelligent air break switch terminal generates an entropy migration event, and broadcasts event information containing a time stamp of the event and an amplitude of the entropy to other intelligent air break switch terminals in the power distribution network through the intelligent air break switch terminal communication module; The superior intelligent air break switch terminal receives and collects a series of the entropy migration events caused by the same physical disturbance to form the entropy migration spectrum.

3. The multi-level collaborative fault detection method based on intelligent air switching according to claim 1, characterized in that, S1 further comprises the following steps: After completing time synchronization through a network time protocol or a GPS module, each intelligent air break switch terminal deployed at each level of the power distribution network collects the multi-physical quantity time series signals, wherein the multi-physical quantity time series signals comprise current time series, voltage time series, contact temperature sequence and bus temperature sequence. S3 further comprises the following steps: At least one of the following diagnostic logics is executed: calculating a first-order difference of the dynamic correlation entropy, and determining a short circuit or parallel arc fault when the first-order difference exceeds a first preset threshold; 4. The multi-level collaborative fault detection method based on intelligent air switching according to claim 1, characterized in that, judging an amplitude of the dynamic correlation entropy, and determining a series arc fault when the amplitude continuously exceeds a second preset threshold; combining the dynamic correlation entropy, the contact temperature sequence and the bus temperature sequence obtained in the synchronous collection of the multi-physical quantity time series signals to determine a gradual fault when both the dynamic correlation entropy and the differential temperature continuously increase. S4 further comprises the following steps:

5. The multi-stage collaborative fault detection method based on intelligent air switching according to claim 4, characterized in that, Based on the time stamp and the amplitude of the entropy of each entropy migration event in the entropy migration spectrum, a time delay spectrum and an attenuation factor spectrum are calculated; ； In the formula, is an attenuation factor spectrum; is an entropy value amplitude of a fault source node; to is an entropy value amplitude detected by the intelligent air switch terminal at each level successively upwards from the fault source along the power transmission direction, respectively. and according to the combination characteristics of the time delay spectrum and the attenuation factor spectrum, a fault type is matched and determined from a preset fault feature spectrum library. The attenuation factor spectrum is calculated in the following manner: and the fault type is determined according to the characteristics of the attenuation factor spectrum: When the attenuation factor spectrum exhibits rapid attenuation characteristics, it is determined to be a series fault; When the attenuation factor spectrum exhibits a slight attenuation characteristic, it is determined to be a parallel fault.

6. The multi-level collaborative fault detection method based on intelligent air switching according to claim 1, characterized in that, The step of performing graded protection actions based on the results of the local fault diagnosis and the collaborative tracing includes: When the result of the local fault diagnosis is a short circuit or parallel arc fault, the intelligent circuit breaker terminal that made the diagnosis directly drives the power electronic switch built into the intelligent circuit breaker terminal to quickly disconnect the circuit. When the result of the collaborative tracing is a parallel fault, the upper-level intelligent circuit breaker terminal that has collected the entropy migration spectrum sends a pre-blocking command to the higher-level intelligent circuit breaker terminal to which the upper-level intelligent circuit breaker terminal belongs, so that the power electronic switch of the higher-level intelligent circuit breaker terminal enters the standby state and performs collaborative backup protection. Furthermore, after the power electronic switch performs the rapid disconnection, the intelligent circuit breaker terminal delays and drives the electromagnetic tripping mechanism built into the intelligent circuit breaker terminal to complete the final physical disconnection.

7. The multi-stage collaborative fault detection method based on intelligent air switching according to claim 1, characterized in that, The method further includes: After performing the graded protection action, or when the fault is determined to be a gradual fault based on the local fault diagnosis, the fault data, including the local fault diagnosis result, the collaborative tracing result, the entropy migration spectrum, and the multi-physical quantity time series signal, is uploaded to the cloud platform for archiving.

8. A multi-stage collaborative fault detection system based on intelligent air gap, applying a multi-stage collaborative fault detection method based on intelligent air gap as claimed in any one of claims 1-7, characterized in that, It includes multiple intelligent circuit breaker terminals deployed at various levels of the power distribution network, and each of the intelligent circuit breaker terminals includes the following functional modules: The acquisition module is used for the synchronous acquisition of time-series signals of multiple physical quantities; The entropy calculation module is used to reconstruct the phase space of the current time series acquired by the acquisition module and calculate the dynamic correlation entropy that reflects the dynamic behavior of the power circuit. The local diagnostic module is used to perform local fault diagnosis based on the dynamic correlation entropy; The communication module enables the intelligent circuit breaker terminal to communicate collaboratively with other intelligent circuit breaker terminals. The collaborative tracing module is used to construct an entropy migration spectrum based on the dynamic correlation entropy of each terminal through the collaborative communication, and to perform collaborative tracing of faults according to the entropy migration spectrum; The protection execution module is used to perform hierarchical protection actions based on the diagnostic results of the local diagnostic module and the traceability results of the collaborative traceability module.

Images