Fault self-healing switching control device for power distribution room based on state monitoring and health assessment

Abstract

The application relates to the technical field of power system distribution automation, in particular to a distribution room fault self-healing switching control device based on state monitoring and health evaluation, which comprises a state acquisition module, which acquires a topological structure and collects electrical data containing voltage, current and contact state, switch mechanism operation data and vibration data; a working condition compensation module, which extracts disturbance characteristics based on the vibration data, corrects the electrical data based on the disturbance characteristics and generates compensation electrical data. In the application, vibration posture data of a mobile carrier and electrical operation states are deeply fused, disturbance characteristics are extracted to dynamically filter electrical signals, and fault confidence is corrected in combination with the time sequence coincidence degree of abnormal data and impact peaks, so that transient interference caused by physical jolt and real power grid short-circuit abnormality are effectively distinguished.

Description

Technical Field

[0001] This invention relates to the field of power system distribution automation technology, and in particular to a fault self-healing switching control device for substations based on condition monitoring and health assessment. Background Technology

[0002] With the increasing demand for power grid emergency protection and special operations, mobile power distribution rooms equipped with fault self-healing switching control devices are widely used in mobile carriers such as emergency power vehicles, ships, and rail vehicles. These devices mainly rely on real-time monitoring of electrical parameters such as bus voltage, main circuit current, and switch auxiliary contact positions to achieve rapid identification of power grid anomalies and automatic reconfiguration and restoration of power supply paths.

[0003] Existing power distribution room monitoring thresholds and self-healing logic are usually designed for static locations. When applied to complex mobile conditions, the mechanical vibration and impact of the carrier can easily cause transient spikes in the sensor's sampling signal and cause brief physical jitter in the switch auxiliary contacts. Because the existing system lacks the ability to synchronously perceive and distinguish mechanical environmental conditions, it is easy to misjudge these signal fluctuations caused by physical forced vibrations as real short-circuit anomalies in the power grid, thereby triggering false alarms and unnecessary power outages. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a power distribution room fault self-healing switching control device based on condition monitoring and health assessment. It aims to improve the existing technology's lack of synchronous perception and identification of mechanical environmental conditions, which easily leads to misjudging signal fluctuations caused by physical forced vibrations as real short-circuit anomalies in the power grid.

[0005] This invention provides the following technical solution: a power distribution room fault self-healing switching control device based on condition monitoring and health assessment, comprising: The status acquisition module obtains the topology and collects electrical data including voltage, current and contact status, switching mechanism operation data and vibration data; The operating condition compensation module extracts disturbance features based on the vibration data, corrects the electrical data based on the disturbance features, and generates compensated electrical data. The fault assessment module extracts abnormal features based on the compensated electrical data, calculates the temporal overlap and topology propagation relationship of the abnormal features based on the disturbance features and the topology, and generates the fault location and type. The health assessment module extracts the corresponding operating data of the switching mechanism based on the fault location and type, and generates a probability value of equipment failure to operate. The switching decision module plans a switching path based on the fault location, type, equipment failure to operate risk probability value and the topology, determines the switching window based on the vibration data below the threshold interval, and generates a switching command. The execution verification module performs actions based on the switching command within the switching window. Based on the electrical data verification results after the action, if an anomaly occurs, the topology is updated and the switching decision module is triggered to regenerate the switching command.

[0006] Preferably, in the status acquisition module, the step of acquiring the topology and collecting electrical data including voltage, current, and contact status, switching mechanism operation data, and vibration data specifically includes the following steps: Analyze the static topology node matrix of the distribution network, establish the connection relationships of each switching device, and obtain the topology structure; By using multi-channel synchronous sampling, the transient bus voltage, main circuit current waveform and auxiliary contact level status of each switching device are acquired under the same time reference as electrical data. Vibration data are constructed by acquiring three-dimensional acceleration, impact envelope, and spatial tilt angle through a multi-axis inertial measurement unit. Extract the current waveform characteristics of the closing and opening coils of each switching device and the historical action time to construct the operating data of the switching mechanism.

[0007] Preferably, in the operating condition compensation module, the generation of compensation electrical data specifically includes the following steps: Time-frequency domain analysis was performed on the vibration data to extract the transient energy peak value and frequency band distribution characterizing the carrier's turbulence and mechanical impact, as disturbance features; Construct a dynamic filtering threshold and anti-shake time constant that are piecewise linearly positively correlated with the transient energy peak; Using the dynamic filtering threshold and anti-jitter time constant, the electrical data within the same time window is masked to filter out high-frequency jitter and voltage / current glitches caused directly by severe mechanical vibration, and the compensated electrical data is output.

[0008] Preferably, in the fault assessment module, generating the fault location and type specifically includes the following steps: Detect transient out-of-limit mutations in the compensation electrical data, mark them as abnormal features, and record the time of occurrence of the mutations; Calculate the time difference between the time of the mutation and the time of the peak impact energy in the disturbance feature. If the time difference is less than a preset threshold and is not continuous, it is determined that the timing is highly overlapping, thereby reducing the electrical fault confidence of the abnormal feature. The abnormal features are mapped to the topology, and the overcurrent directionality and voltage drop consistency of the upstream and downstream nodes are checked along the feeder direction. If the topology propagation logic of the upstream node detecting overcurrent and the adjacent downstream node not detecting overcurrent is met, and the electrical fault confidence level is up to standard, then it is confirmed as a real electrical fault and the fault location and type are locked.

[0009] Preferably, reducing the electrical fault confidence of the abnormal characteristics specifically includes the following steps: Extract the acceleration amplitude corresponding to the peak moment of the impact energy from the disturbance features; The time difference and the acceleration amplitude are input into a preset two-dimensional penalty function, and the confidence decay coefficient is calculated based on an algorithm model that includes a negative exponential term for the time difference and a proportional term for the acceleration amplitude; wherein, the time difference is positively correlated with the confidence decay coefficient, and the acceleration amplitude is negatively correlated with the confidence decay coefficient; Multiply the preset initial fault confidence level by the confidence attenuation coefficient to output the reduced electrical fault confidence level.

[0010] Preferably, in the health assessment module, generating the device refusal-to-operate risk probability value specifically includes the following steps: Based on the location and type of the fault, identify the set of target switches to be involved in the isolation and recovery operations; Extract the operating data of the switching mechanism corresponding to the target switch set; Analyze the coil current integral area change rate and historical action time drift trend in the operating data of the switching mechanism; The integral area change rate of the coil current and the drift trend of the historical action time are respectively substituted into the preset Weibull distribution lifetime model as independent variables to map the electrical aging failure probability and the mechanical jamming failure probability respectively. Based on a preset multi-risk coupling formula, the electrical aging failure probability and the mechanical jamming failure probability are combined to generate a device failure risk probability value.

[0011] Preferably, in the switching decision module, generating the switching instruction specifically includes the following steps: Based on the fault location and the topology, a graph theory shortest path algorithm is used to traverse the backup power recovery channel, and during the traversal, switch nodes whose equipment failure risk probability value is higher than the safety threshold are eliminated to generate target isolation and backup power supply path. Using a sliding time window combined with an autoregressive moving average algorithm, the future fluctuation trend of the vibration data is predicted over time. The duration of the impact energy trough and the acceleration being lower than the seismic resistance limit of the switching mechanism is located and marked as the switching window. The action sequence of the target isolation and backup power supply path is packaged with the trigger timestamp of the switching window to generate a switching command.

[0012] Preferably, the steps of the execution verification module specifically include: When the switching window timestamp is reached, the operating mechanism of the corresponding switch is driven to open or close. After the theoretical operation time has elapsed, the current electrical data is re-acquired, and cross-logic judgment is performed: if the auxiliary contact status indicator has been activated, but the main circuit current and bus voltage have not undergone corresponding on / off step changes, then the corresponding switch is determined to be malfunctioning. The corresponding switch that is determined to be abnormal is marked as a deadlock node in the topology, the topology is updated, and the execution failure interrupt is fed back to the switching decision module to trigger the regeneration of the switching instruction.

[0013] The present invention has the following beneficial effects: 1. In this invention, the vibration attitude data of the mobile carrier is deeply integrated with the electrical operating status. The electrical signal is dynamically filtered by extracting disturbance features, and the fault confidence is corrected by combining the temporal overlap of abnormal data and impact peak values. This effectively distinguishes between transient interference caused by physical turbulence and real power grid short circuit anomalies, reducing the probability of false alarms and false trips in mobile power distribution rooms under harsh operating conditions.

[0014] 2. In this invention, by extracting the characteristics of switch action, a quantitative probability index representing the degree of degradation is calculated. During the backup power supply path planning stage, switch nodes with a high risk of failure to operate are actively eliminated. The time series prediction algorithm is used to find the optimal location of the stable time period with weak vibration energy to issue action commands. This avoids physical jamming of the operating mechanism under severe forced vibration and improves the execution rate and success rate of reconfiguration switching operations.

[0015] 3. In this invention, after the switch is activated, it no longer relies solely on the auxiliary contact signal, which is easily distorted by vibration. Instead, it synchronously collects the voltage difference at the switch break and cross-compares it with the main circuit current. When the voltage difference at the break and the circuit current are found to be inconsistent, the abnormal operation is promptly determined, and the corresponding switch is automatically marked as deadlocked and the path reconstruction is retried, thus ensuring the reliability of power supply restoration when the power distribution system encounters accidental failure of the switch to operate. Attached Figure Description

[0016] Figure 1 This is an architecture diagram of the power distribution room fault self-healing switching control device based on condition monitoring and health assessment proposed in this invention. Figure 2 This is a flowchart illustrating the working logic of the working condition compensation module of the present invention. Figure 3This is a flowchart illustrating the working logic of the switching decision module of the present invention. Detailed Implementation

[0017] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention provides a power distribution room fault self-healing switching control device based on condition monitoring and health assessment, such as... Figure 1 As shown, it includes: The status acquisition module obtains the topology and collects electrical data including voltage, current and contact status, switching mechanism operation data and vibration data; Furthermore, the status acquisition module acquires the topology and collects electrical data including voltage, current, and contact status, as well as switching mechanism operation data and vibration data, specifically including the following steps: Analyze the static topology node matrix of the distribution network, establish the connection relationships of each switching device, and obtain the topology structure; By using multi-channel synchronous sampling, the transient bus voltage, main circuit current waveform and auxiliary contact level status of each switching device are acquired under the same time reference as electrical data. Vibration data are constructed by acquiring three-dimensional acceleration, impact envelope, and spatial tilt angle through a multi-axis inertial measurement unit. Extract the current waveform characteristics of the closing and opening coils of each switching device and the historical action time to construct the operating data of the switching mechanism.

[0019] Specifically, the status acquisition module is deployed on the bottom control board of the self-healing switching control device in the power distribution room. It communicates with external sensors and switching equipment in real time via a field data bus and performs data acquisition. The system first reads the preset static topology node matrix of the power distribution network from the memory. Assume the power distribution network contains N switching nodes, and the static topology node matrix is ​​an N x N square matrix. The values ​​of the matrix elements are determined by the physical cable connection status between nodes; when node i and node j are directly connected, the corresponding element is assigned a value of one; when they are not directly physically connected, the value is zero. The system iterates through the matrix to extract all node pairs with a value of one, establishing the electrical connection relationships between each switching device, thereby obtaining the initial topology of the power distribution network.

[0020] After acquiring the topology, the system uses synchronous sampling technology to obtain the electrical operating status of each node. The underlying hardware uses a shared external high-precision active crystal oscillator as the global clock source, driving a multi-channel analog-to-digital converter chip to perform high-frequency synchronous sampling of each switching device at a fixed reference sampling frequency of 10000 Hz. Under an absolutely consistent time reference, the system acquires the transient bus voltage waveform and main circuit current waveform of each switching device in real time, and simultaneously reads the high and low level states of auxiliary contacts through an optocoupler isolation circuit. The system timestamps the transient bus voltage waveform, main circuit current waveform, and auxiliary contact level states together, merges and packages them as the electrical data of the switching device, and stores them in a memory buffer.

[0021] For mobile power distribution room environments, the system acquires attitude information through a multi-axis inertial measurement unit (IMU) deployed on the rigid structural components of the carrier chassis and power distribution cabinet. The IMU acquires three-dimensional acceleration signals and spatial tilt signals in real time at a frequency of 100 Hz. The spatial tilt angle specifically includes the current roll, pitch, and yaw angles of the cabinet. For the three-dimensional acceleration signals, assuming the variable t is the current sampling time, the system first extracts the instantaneous acceleration components along the three orthogonal axes. , and And calculate the spatial composite acceleration modulus. The system then uses the Hilbert transform algorithm to extract the impact envelope data caused by the mechanical impact. Let the spatial composite acceleration modulus be... The result after Hilbert transformation is Then the impact envelope at time t The calculation formula is: The system integrates three-dimensional acceleration, calculated impact envelope, and spatial tilt angle to construct vibration data characterizing the current mechanical physical environment. Under normal stable driving conditions, the basic constant threshold of the impact envelope is usually set at 0.2g; if this fixed value is exceeded, the system determines that there is an external mechanical disturbance.

[0022] The system further monitors the health status parameters of the internal operating mechanisms of each switchgear. When the switch performs a closing or opening operation, the monitoring circuit acquires the drive current of the actuator coil at a high frequency of 1000 Hz, and the time interval constant between two adjacent discrete sampling points. The time is constant at 0.001 seconds. Let the time when the trigger command for the operating mechanism is issued be... The moment when the actual displacement feedback of the switch auxiliary contact occurs is During this period, a total of M discrete sampling points were acquired, where k is the independent variable representing the sampling point number, and the transient current value of the k-th sampling point is... The system calculates the integral area of ​​the coil current. The discrete summation formula is: In terms of physical dimensions, this integral area Let be the total charge flowing through the coil during operation. Under the condition of a distribution network with a constant operating voltage, this total charge is directly proportional to the electromagnetic drive energy consumed by the operating mechanism in a single operation. Therefore, the system uses it as a characteristic parameter to characterize the drive energy consumption. The system also considers the time difference. The historical operation time is recorded. For the operating mechanism of a standard 10kV vacuum circuit breaker, the typical value of the operation time constant under normal factory conditions is set to 40 milliseconds. The system combines the calculated coil current integral area with the historical operation time to construct the switching mechanism operating data and stores it in the equipment's lifecycle archive.

[0023] The above-mentioned status acquisition steps effectively construct a foundational data system that truly reflects the operating conditions of mobile power distribution rooms by aligning and fusing static power grid topology, transient electrical waveforms, external physical motion posture, and underlying mechanical operation energy consumption in multiple dimensions. This provides an accurate data source for subsequent seismic resistance and fault identification to prevent misjudgment.

[0024] like Figure 2 As shown, the working condition compensation module extracts disturbance features based on vibration data, corrects electrical data based on disturbance features, and generates compensated electrical data. Furthermore, in the operating condition compensation module, generating compensation electrical data specifically includes the following steps: Time-frequency domain analysis was performed on the vibration data to extract the transient energy peak and frequency band distribution that characterize the turbulence and mechanical impact of the carrier, as disturbance features; Construct a dynamic filtering threshold and anti-shake time constant that are piecewise linearly positively correlated with the transient energy peak; By using a dynamic filtering threshold and a debouncing time constant, electrical data within the same time window is masked to filter out high-frequency jitter and voltage / current glitches caused directly by severe mechanical vibration, and then outputs compensated electrical data.

[0025] Specifically, the operating condition compensation module receives vibration data and electrical data synchronously output by the status acquisition module. The system first uses short-time Fourier transform to perform time-frequency domain analysis on the vibration data to extract disturbance characteristics. To match the 100 Hz vibration sampling frequency, the system calculates the power spectral density of the three-dimensional synthetic acceleration signal using a 2.56-second sliding data window. The power spectrum is accumulated from low to high frequency, and the start and end frequency intervals corresponding to when the accumulated energy first reaches 80% of the total spectral energy are extracted as the frequency band distribution. Simultaneously, the system calculates the actual impact energy within a 0.2-second short sliding window. Let n be the current vibration sampling point number, and the short sliding window contain L consecutive sampling points. As an implementation parameter, L is set to 20, and the vibration sampling time interval is 0.01 seconds. The transient energy peak value within the short sliding window ending at the nth sampling point is calculated. The calculation formula is: ,in Let be the spatial composite acceleration modulus at the corresponding moment. The physical dimensions of the calculated transient energy peak are: The transient energy peak value and frequency band distribution together constitute the disturbance characteristics characterizing the chassis vibration and cabinet mechanical impact of the mobile power distribution room.

[0026] The system constructs a dynamic filtering threshold and anti-shake time constant based on the extracted transient energy peak value. As an implementation parameter, a preset safe vibration energy threshold value of 0.01 is set. The critical value for severe impact energy is 0.1. When the transient energy peak is less than or equal to 0.01 At this time, the system outputs a basic anti-jitter time constant of 5 milliseconds, a basic voltage surge threshold of 0.05 times the rated bus voltage, and a basic current surge threshold of 0.05 times the rated main circuit current. When the transient energy peak is between 0.01 and 0.1... When the interval is between, the system introduces a proportionality coefficient. , and Using the linear mapping formula Calculate the image stabilization time constant ,use Calculate the voltage change threshold ,use Calculate the current surge threshold .in The rated voltage of the busbar The rated current of the main circuit will be used as an implementation parameter. The value is set to 111 milliseconds per second. ,Will The value is set at 1.33 times the rated bus voltage per ,Will The value is set to 1.33 times the rated current of the main circuit per When the transient energy peak value is greater than 0.1 At that time, the saturation limit of the anti-shake time constant is a preset upper limit of 15 milliseconds. This upper limit is strictly less than the standard action time of the switching mechanism to avoid masking the true displacement.

[0027] After obtaining the dynamic parameters, the system performs masking processing on the electrical data within the same time window. Let k be the current electrical sampling point number. For the auxiliary contact level state, the system is based on the anti-jitter time constant. Generate state mask When the duration of the level transition of the auxiliary contact is less than The mask is set to 0 to maintain the state of the previous valid contact when the duration is greater than or equal to 0. The mask is set to 1 to accept the current contact state. For transient bus voltage waveforms and main circuit current waveforms, the following settings are used: and These are the voltage and current amplitudes at the current sampling point, respectively. and Given the amplitude of the previous sampling point, the system calculates the amplitude difference between adjacent sampling points. and When the frequency band distribution within the current time window overlaps with the preset 15Hz to 35Hz mechanical resonant frequency band used as an implementation parameter, the system performs an independent channel determination. If the voltage amplitude difference... Less than or equal to If the voltage channel has a sampling glitch, then it is determined that there is a sampling glitch; if the current amplitude difference is... Less than or equal to If so, it is determined that there is a sampling glitch in the current channel.

[0028] The system employs a moving average smoothing algorithm to correct waveforms in independent channels identified as having sampling glitches, while retaining normal data that does not trigger glitches. Let the number of sampling points in the sliding window for voltage and current be... and The implementation parameter is set to 5. For electrical sampling point k where a glitch is determined to exist, the compensation voltage value is... The calculation formula is: Compensation current value The calculation formula is: The system integrates the contact status after masking with the analog waveform after on-demand correction, and outputs compensated electrical data. These steps enable the filter parameters to adaptively adjust according to the chassis vibration state, providing underlying data support for the system to distinguish between transient data anomalies caused by mechanical shocks and actual electrical faults.

[0029] The fault assessment module extracts abnormal features based on compensated electrical data, calculates the temporal overlap and topology propagation relationship of abnormal features based on disturbance features and topology structure, and generates fault location and type. Furthermore, the fault assessment module generates the fault location and type by including the following steps: Detect transient out-of-limit mutations in compensation electrical data, mark them as anomalous features, and record the time of occurrence of the mutations; Calculate the time difference between the time of the sudden change and the time of the peak impact energy in the disturbance feature. If the time difference is less than the preset threshold and is not continuous, it is determined that the timing is highly overlapping, thus reducing the electrical fault confidence of the abnormal feature. The abnormal features are mapped to the topology, and the overcurrent directionality and voltage drop consistency of the upstream and downstream nodes are checked along the feeder direction. If the topology propagation logic of the upstream node detecting overcurrent and the adjacent downstream node not detecting overcurrent is met and the electrical fault confidence level is up to standard, then it is confirmed as a real electrical fault and the fault location and type are locked.

[0030] Furthermore, reducing the confidence level of electrical faults with abnormal characteristics specifically includes the following steps: Extract the acceleration amplitude corresponding to the peak moment of the impact energy from the disturbance features; The time difference and acceleration amplitude are input into a preset two-dimensional penalty function, and the confidence decay coefficient is calculated based on an algorithm model that includes a negative exponential term for the time difference and a proportional term for the acceleration amplitude. Among them, the time difference is positively correlated with the confidence decay coefficient, and the acceleration amplitude is negatively correlated with the confidence decay coefficient. Multiply the preset initial fault confidence level by the confidence attenuation coefficient to output the reduced electrical fault confidence level.

[0031] Specifically, the fault assessment module receives compensated electrical data, disturbance characteristics, and topology data from the operating condition compensation module, as well as data from the status acquisition module. The compensated electrical data includes corrected three-phase bus voltage waveforms, three-phase main circuit current waveforms, and auxiliary contact status. The disturbance characteristics include transient energy peaks and their corresponding frequency band distributions. Each data point carries a timestamp generated based on the same time base to facilitate time-series correlation analysis between electrical anomalies and mechanical disturbances.

[0032] The system calculates the effective value of the three-phase main circuit current from the compensated electrical data. Assuming the electrical data sampling frequency is 10000 Hz and the power frequency of the distribution network is 50 Hz, the number of sampling points corresponding to one power frequency cycle is... Let it be 200. This is the current electrical sampling point number. For the distinct independent variables and Pick , or , For the first Phase in The compensation current value at each sampling point Let be the index variable within the cyclic sliding window, then the... Phase in The effective value of the current corresponding to each sampling point According to the formula calculate.

[0033] The system compares the current cycle current RMS value with the previous cycle current RMS value. Let... For the first The effective value of the current of the phase at the corresponding moment of the previous power frequency cycle, the first Phase current abrupt change According to the formula Calculation. When the effective value of the current in at least one phase... Greater than or equal to 1.5 times the rated current of the main circuit, and the corresponding current surge. When the current exceeds or equals 0.5 times the rated current of the main circuit, the system marks the corresponding transient over-limit change as an abnormal feature, and records the global sampling timestamp that first meets the above conditions as the time of the change. Among them, 1.5 times the rated current of the main circuit and 0.5 times the rated current of the main circuit are both feasible implementation parameters.

[0034] The system records the duration for which abnormal characteristics continuously meet transient limit-breaking conditions. As an implementation parameter, the duration threshold is set to 0.02 seconds. When the duration... If the abnormal characteristic recovers to outside the transient over-limit condition within 0.02 seconds, and no protection action command or switch auxiliary contact change is detected within the corresponding time period, the system determines that the abnormal characteristic is not persistent. By adding protection action and auxiliary contact status verification, the system avoids directly classifying real faults that are quickly cleared by the protection device as non-persistent abnormalities.

[0035] At the moment the system mutation occurs Vibration and impact characteristics are searched within a 0.5-second time interval before and after the vibration. Let the set of vibration sampling point indices mapped to this time interval be denoted as . , The independent variable is the vibration sampling point number. The first step obtained from the compensation step of the previous working condition The transient energy index corresponding to each vibration sampling point, and the sampling point number corresponding to the peak transient energy. According to the formula Determined. Let the first... The global timestamp of each vibration sampling point is The time difference between the moment of the mutation and the moment of the peak impact energy is... According to the formula Calculation. As an implementation parameter, the timing overlap threshold is set to 0.1 seconds. When If the time interval is less than 0.1 seconds and the abnormal characteristics are not persistent, the system determines that the electrical abnormality and the mechanical impact timing are highly overlapping and triggers the electrical fault confidence decay calculation.

[0036] The system extracts the dynamic acceleration amplitude from the three-dimensional acceleration components of the vibration data. Since the raw three-dimensional acceleration acquired by the inertial measurement unit is based on standard gravitational acceleration... To avoid the static gravity component from participating in the calculation, the system performs high-pass filtering on the acceleration components of the three orthogonal axes with a cutoff frequency of 0.5 Hz as the implementation parameter, resulting in [the desired value]. Dynamic acceleration components in units , and . No. Dynamic acceleration amplitude corresponding to each vibration sampling point According to the formula Calculation. Let's assume... The set of sampling points corresponding to the impact search window centered at [center] and spanning 0.1 seconds is: Peak dynamic acceleration According to the formula calculate.

[0037] The system will use the time difference and peak dynamic acceleration Input a two-dimensional penalty function and calculate the confidence decay coefficient. The confidence decay coefficient is calculated according to the formula. Calculation. Among them, It is a natural constant. The acceleration proportionality coefficient is set to 0.5 per [unit value] as an implementation parameter. ; The time decay constant is set to 20 per second as an implementation parameter. In this physical logic, It is a dimensionless amplification factor. The dimensionless decay exponent, the overall output The dynamics are limited to dimensionless values ​​ranging from 0.1 to 1. An increase in the absolute time difference causes the negative exponential term to decay rapidly, thus leading to... An increase in the peak dynamic acceleration causes the subtrahend term to be amplified overall, thus leading to Decrease.

[0038] Let the initial fault confidence level be... As an implementation parameter Set to 1.0, the reduced electrical fault confidence level According to the formula Calculation. When the time difference If the time interval is greater than or equal to 0.1 seconds, or if the abnormal characteristic is persistent, the system skips the confidence decay and directly assigns a value. The value is 1, thus the fault confidence level decreases only under severe operating conditions where mechanical shock and short-term electrical anomaly timing overlap.

[0039] The system maps abnormal features to the topology and determines candidate upstream nodes based on the feeder connection relationship, starting from the power node. and its adjacent downstream nodes To eliminate the influence of asymmetrical short circuits on direction determination, the system employs a discrete Fourier transform of one power frequency cycle to extract the positive-sequence fundamental voltage phasor and positive-sequence fundamental current phasor of each node, and calculates the node... Positive sequence directional power Directional power is calculated according to the formula. Calculation. Among them, Number the switch nodes in the topology. For nodes The positive-sequence fundamental voltage phasor. For nodes The conjugate complex number of the positive-sequence fundamental current phasor. This indicates taking the real part of a complex number. The positive direction is defined as the direction from the power source side to the load side. When the value is greater than 0, the system determines the node. A positive current was detected.

[0040] The system simultaneously calculates candidate upstream nodes and adjacent downstream nodes The effective values ​​of the three-phase phase voltages. As implementation parameters, the phase voltage sag threshold is set to 0.8 times the rated phase voltage of the bus, and the voltage sag time difference threshold is set to 0.02 seconds. When the candidate upstream node... Forward overcurrent detected, adjacent downstream node When the effective values ​​of the three-phase currents do not reach 1.5 times the rated current of the main circuit, and the effective value of the phase voltage of at least one phase at each of the two nodes is lower than 0.8 times the rated phase voltage of the bus, and the time difference of the voltage drop does not exceed 0.02 seconds, the system determines that the abnormal characteristics conform to the topology propagation logic that the upstream node detects overcurrent and the adjacent downstream node does not detect overcurrent.

[0041] The system determines the specific fault type based on current and voltage characteristics. Zero-sequence current RMS value. According to the formula Calculation. Among them, , and These are the three-phase fundamental current phasors of the node where the abnormality occurred. When the effective value of any two phase currents at that node is greater than or equal to 1.5 times the rated current of the main circuit, the corresponding line voltage is lower than 0.8 times the rated line voltage of the bus, and the effective value of the zero-sequence current is less than 0.1 times the rated current of the main circuit (preset as an implementation parameter), the system determines it as a phase-to-phase short circuit; when the effective values ​​of all three phase currents are greater than or equal to 1.5 times the rated current of the main circuit and all three line voltages are lower than 0.8 times the rated line voltage of the bus, it is determined as a three-phase short circuit; when the effective value of the zero-sequence current is greater than or equal to 0.1 times the rated current of the main circuit and at least one phase voltage is lower than 0.8 times the rated phase voltage of the bus, it is determined as a ground fault.

[0042] The system ultimately extracts the confidence level of electrical faults. The preset threshold for compliance as an implementation parameter is 0.6. When the candidate fault location satisfies the topology propagation logic, the electrical characteristics meet the judgment conditions for the corresponding fault type, and the electrical fault confidence level is... When the value is greater than or equal to 0.6, the system confirms it as a genuine electrical fault and identifies a candidate upstream node. With adjacent downstream nodes The fault location is identified within the line section between the faults, and corresponding fault type data is generated. This process utilizes short-circuit transient characteristics combined with a seismic penalty model and a positive-sequence directional power algorithm to provide the device system with fault assessment results exhibiting extremely high disturbance immunity.

[0043] The health assessment module extracts the corresponding switching mechanism operation data based on the fault location and type, and generates a probability value of equipment failure to operate. Furthermore, the health assessment module generates the probability value of equipment failure to operate, specifically including the following steps: Based on the location and type of the fault, identify the set of target switches to be involved in the isolation and recovery operations; Extract the operating data of the switching mechanisms corresponding to the target set of switches; Analyze the rate of change of coil current integral area and the drift trend of historical action time in the operating data of the switching mechanism; The integral area change rate of coil current and the drift trend of historical action time are substituted as independent variables into the preset Weibull distribution lifetime model to map the electrical aging failure probability and the mechanical jamming failure probability, respectively. Based on a pre-defined multi-risk coupling formula, the probability of electrical aging failure and the probability of mechanical jamming failure are combined to generate a probability value of equipment failure to operate.

[0044] Specifically, the health assessment module receives fault location and fault type data output by the fault assessment module. Based on the candidate upstream nodes and adjacent downstream nodes determined by the fault location, the system identifies the circuit breakers at both ends directly connected to the faulty section as the switches to be isolated. Simultaneously, based on the initial topology, the system traverses along the non-faulty power supply path to identify tie switches capable of restoring power to the non-faulty outage area as the switches to be restored. The system then merges the switches to be isolated and restored, locking them into a target set of switches to participate in the isolation and restoration operations.

[0045] For each individual switch in the target switch set, the system extracts its corresponding switch mechanism operation data from the equipment lifecycle file maintained by the status acquisition module. Specifically, the switch mechanism operation data includes the coil current integral area and operation time collected during each switch operation. The system reads the first set of normal coil current integral areas recorded when the switch is initially put into operation from this file and uses it as the reference coil current integral area. And extract the integral area of ​​the most recently measured coil current. The system calculates the rate of change of the coil current integral area. Its calculation formula is Similarly, the system extracts the factory baseline action time of 40 milliseconds from the preset parameters as the implementation parameter. Extract the action time from the most recent actual measurement. The system calculates the drift trend of historical action time. Its calculation formula is The aforementioned rates of change and drift trends are dimensionless values ​​characterizing the degree of equipment degradation. If the calculated result is less than zero, it is forcibly set to zero to indicate that no degradation or deterioration has occurred.

[0046] The system employs a mathematical mapping method for equivalent lifetime degradation, integrating the coil current with the rate of change of the area. Trends in historical action time drift Each variable is represented as an independent pseudo-lifetime variable and substituted into the pre-defined Weibull distribution lifetime model. The system calculates the probability of electrical aging failure. Its calculation formula is .in, It is a natural constant. The shape parameter, used to characterize the electrical degradation rate, is set to 2.0 as an implementation parameter; The dimensional parameter characterizing the electrical withstand limit is set to 0.5 as an implementation parameter. The system simultaneously calculates the probability of mechanical jamming failure. Its calculation formula is .in, The shape parameter, used to characterize the rate of mechanical degradation, is set to 2.5 as an implementation parameter; The scale parameter, used to characterize the mechanical endurance limit, is set to 0.3 as the implementation parameter. Using the aforementioned degradation mapping model, the system outputs independent failure probabilities ranging from 0 to 1.

[0047] After obtaining two independent failure probabilities, the system performs joint probability synthesis based on a preset multi-risk coupling formula. Since any failure, whether due to coil electrical aging or mechanical jamming, will lead to overall switch failure, the system employs a multi-risk coupling formula that conforms to independent series failure logic. Let the final equipment failure probability of this switch be... The formula for multiple risk coupling is: The system performs joint probability synthesis using this formula, outputting a probability value for equipment failure risk between 0 and 1. .

[0048] The implicit mechanical aging and electrical degradation characteristics of switches are quantified into unified probability values, providing equipment health constraints for the subsequent switching decision module when traversing the backup power supply channel. This enables the system to proactively eliminate switch nodes with high failure risk, thereby ensuring the success rate of grid reconfiguration.

[0049] like Figure 3 As shown, the switching decision module plans a switching path based on the fault location, type, equipment failure to operate probability value and topology, determines the switching window based on the vibration data below the threshold, and generates a switching command. Furthermore, the process of generating a switching instruction within the switching decision module specifically includes the following steps: Based on the fault location and topology, a graph theory shortest path algorithm is used to traverse the backup power recovery channel. During the traversal, switch nodes with a failure-to-operate risk probability value higher than the safety threshold are removed to generate target isolation and backup power supply paths. By using a sliding time window combined with an autoregressive moving average algorithm, the future fluctuation trend of vibration data is predicted in time series. The duration of the impact energy trough and the acceleration is lower than the seismic resistance limit of the switching mechanism is located and marked as the switching window. The action sequence of the target isolation and backup power supply path is packaged with the trigger timestamp of the switching window to generate a switching command.

[0050] Specifically, the switching decision module receives the equipment failure risk probability value generated by the health assessment module and retrieves the initial distribution network topology obtained during the status acquisition phase and the fault location data generated during the fault assessment phase. The system maps the initial topology to an undirected graph theory model, defining each switching device in the distribution network as a node in the graph theory model and the physical lines connecting the switches as edges. Based on the fault location, the system first locates the directly connected nodes that must be tripped to form an isolation boundary. On this basis, the system initiates a graph theory shortest path algorithm to search for a recovery path from the backup power node to the non-faulty power loss area.

[0051] During the traversal of the backup power restoration channel, the system performs a failure-to-operate risk check on each candidate switch node. The system extracts the equipment failure-to-operate risk probability value of each candidate switch node and sets a safety threshold of 0.15 as an implementation parameter. When a candidate switch node with an equipment failure-to-operate risk probability value greater than 0.15 is encountered, the system sets the edge weight of the corresponding node to infinity in the graph theory model; for normal switch nodes with an equipment failure-to-operate risk probability value less than or equal to 0.15, the system uniformly sets the edge weight of the corresponding node to 1. Based on the graph theory model after the above edge weight assignment, the system uses Dijkstra's algorithm to calculate the shortest path between the backup power node and the connected nodes in the power failure area. This path is the path with the fewest hops to avoid high-risk nodes. The system generates a target isolation and backup power supply path containing a specific opening and closing action sequence based on the nodes traversed by the shortest path.

[0052] To ensure that the switching action avoids periods of severe mechanical vibration, the system uses an autoregressive algorithm to predict the future fluctuation trend of vibration data over time. The system extracts the continuous historical dynamic acceleration amplitude sequence obtained during the state acquisition phase and uses it as an implementation parameter to set the vibration sampling time interval to 0.01 seconds. Let the vibration sampling point number at the current moment be... The corresponding dynamic acceleration amplitude is The system extracts amplitude data containing the current moment and the two previous historical moments. , and The system first calculates the first-order difference sequence to achieve time series stabilization. The formula for calculating the difference value at the current time step is: The formula for calculating the difference value at the previous time step is: .

[0053] The system constructs a second-order autoregressive prediction model. Let... Let be a positive integer variable representing the prediction step size, and be used to cover the mechanism's action time window. The upper limit of the total prediction step size is set to 50 as an implementation parameter. (Looking towards the future...) Prediction difference for each prediction step The recursive formula is .in, and The dimensionless autoregressive coefficients are set to 0.65 and 0.25 respectively as implementation parameters. When calculating the first step prediction... When equal to 1, the required value on the right side of the formula is... and Use the known actual difference values ​​directly and Substitute the values ​​into the calculation. After obtaining the difference prediction value, the system performs an accumulation and restoration calculation, projecting the result onto the future... Predicted dynamic acceleration amplitude for each prediction step The formula is .when When equal to 1, the right side of the formula Use the actual dynamic acceleration amplitude directly Substitute. The system uses the above two sets of formulas in... The values ​​are taken from 1 to 50 and iteratively calculated to generate a predictive dynamic acceleration sequence that includes the future waveform evolution trend.

[0054] The system locates the impact energy trough in the predicted dynamic acceleration sequence. As implementation parameters, the system sets the seismic resistance limit of the switching mechanism to 0.5 times the standard gravitational acceleration and the safety time margin for mechanism operation to 0.1 seconds. Since the sampling interval is 0.01 seconds, this time margin corresponds to 10 consecutive prediction steps. The system searches backward along the time axis in the predicted dynamic acceleration sequence, and when it finds the predicted dynamic acceleration amplitude within 10 consecutive prediction steps... When all values ​​are strictly less than 0.5 times the standard gravitational acceleration, the mechanical environment in that range is considered stable. The system extracts the time node corresponding to the first step among these 10 prediction steps and marks it as the trigger timestamp for the switching window.

[0055] After obtaining the action path and execution timing, the system performs instruction packetization. The system merges the switch equipment numbers and corresponding opening / closing status words contained in the target isolation and backup power supply path with the trigger timestamp of the switching window into the same data frame according to a preset communication protocol format, generating a switching instruction with strict timing execution constraints.

[0056] By visualizing the health characteristics of the switch as edge weight penalties in a graph theory algorithm, and using a dimensionless autoregressive mathematical model to predict future vibration trajectories, an execution instruction that balances electrical reliability and mechanical safety is provided for self-healing switching operations.

[0057] The execution verification module performs actions based on the switching command within the switching window, verifies the electrical data after the action, updates the topology structure and triggers the switching decision module to regenerate the switching command in case of an anomaly.

[0058] Furthermore, the specific steps for executing the verification module include: When the switching window timestamp is reached, the operating mechanism of the corresponding switch is driven to open or close. After the theoretical operation time has elapsed, the current electrical data is re-acquired, and cross-logic judgment is performed: if the auxiliary contact status indicator has been activated, but the main circuit current and bus voltage have not undergone corresponding on / off step changes, then the corresponding switch is judged to be abnormal. The corresponding switch that is determined to be abnormal is marked as a deadlock node in the topology, the topology is updated, and the execution failure interrupt is fed back to the switching decision module to trigger the regeneration of the switching command.

[0059] Specifically, the execution verification module receives a switching instruction with strict timing constraints from the switching decision module. The system parses the data frame containing the switching instruction, extracting the device number of the target switch to be operated, the target action status word, and the trigger timestamp of the switching window. The system calls the high-precision real-time clock of the underlying hardware, setting the local current time to [missing information]. The obtained trigger timestamp is... The system compares the timeline in real time, and when the local current time... With trigger timestamp When they are completely equal, the control unit sends a circuit breaker control signal to the operating mechanism of the corresponding switch through the drive circuit, which drives the mechanical mechanism of the corresponding switch to perform physical actions.

[0060] The system starts an action verification timer after sending a control electrical signal. Assume the theoretical timeout for the switch mechanism to operate is... The implementation parameter is set to 40 milliseconds; the safety margin constant considering the contact mechanical bounce and arc extinction transition process is set to... The implementation parameter is set to 60 milliseconds. The system sets the total verification delay time. Its calculation formula is When the elapsed time since the trigger moment reaches the total verification delay time. Afterwards, the system re-collects the current compensation electrical data. This applies to equipment numbered... For the corresponding switches, the system extracts the effective value of the first port voltage of the switch at the current steady-state power frequency cycle at the verification time. , Effective value of voltage at the second port of the switch and the effective value of the main circuit current The system calculates the differential pressure at the break point after the switch is activated. Its calculation formula is .

[0061] The system performs cross-logic judgment based on the re-acquired auxiliary contact status data. As an implementation parameter, the system sets a pressure determination threshold. Set the break-point voltage difference threshold to 0.8 times the rated bus voltage. Set the current conduction threshold to 0.1 times the rated bus voltage. Set the current cutoff threshold to 0.05 times the rated current of the main circuit. The current is 0.02 times the rated current of the main circuit. If the target action status word is a closing command, and the auxiliary contact status indicator is closed after the action, the system verifies the voltage difference at the break point and the circuit current. When the voltage difference at the break point... Greater than or equal to the fracture pressure difference threshold And the effective value of the main circuit current Strictly less than the current conduction threshold If this occurs, it indicates that there is no electrical connection between the two sides of the switch and no current flows through it. The system determines that the auxiliary contact has a loose connection, and the corresponding switch closing action is abnormal.

[0062] If the target action status word is a tripping command, and the auxiliary contact status indicator has been opened after the action, the system verifies the current cutoff status. When the main circuit current effective value... Greater than or equal to the current cutoff threshold This indicates that the main circuit failed to successfully extinguish the arc and cut off the current. Furthermore, since the power flow direction may reverse during the self-healing reconfiguration process, the system extracts the maximum value from the effective voltage values ​​on both sides of the switch. If the effective value of the main circuit current is... Less than the current cutoff threshold However, the maximum voltage on both sides of the switch Greater than or equal to the pressure detection threshold And the pressure difference at the fracture surface Strictly less than the fracture pressure difference threshold If the switch remains physically connected and at the same potential even when a live power source is present on either side, the system determines that the mechanism is mechanically jammed or the contacts are stuck. If any of these abnormal conditions are met, the system determines that the corresponding switch tripping action is abnormal.

[0063] When a corresponding switch is determined to be malfunctioning, the system marks the malfunctioning switch as a deadlock node in the distribution network's undirected graph theory model. The system then forcibly overwrites the edge weights of all edges associated with this deadlock node in the graph theory model to infinity, thus completing a topology degradation update. After the topology update is complete, the system sends an execution failure interrupt signal containing the device number of the deadlock node to the switching decision module via the internal bus. Upon receiving this interrupt signal, the switching decision module re-triggers the shortest path search and autoregressive time series prediction algorithm based on the updated graph theory model to regenerate a new round of switching instructions that does not include the deadlock node.

[0064] By leveraging the cross-physical properties of the break-point pressure difference and the loop current, the system overcomes the defect of single auxiliary contact feedback distortion in mobile environments, completely decouples the interference of bidirectional power flow reversal on the verification logic, and ensures the system's self-healing and recovery capability when encountering switch failure through a closed-loop topology reconstruction mechanism.

[0065] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power distribution room fault self-healing switching control device based on condition monitoring and health assessment, characterized in that, include: The status acquisition module obtains the topology and collects electrical data including voltage, current and contact status, switching mechanism operation data and vibration data; The operating condition compensation module extracts disturbance features based on the vibration data, corrects the electrical data based on the disturbance features, and generates compensated electrical data. The fault assessment module extracts abnormal features based on the compensated electrical data, calculates the temporal overlap and topology propagation relationship of the abnormal features based on the disturbance features and the topology, and generates the fault location and type. The health assessment module extracts the corresponding operating data of the switching mechanism based on the fault location and type, and generates a probability value of equipment failure to operate. The switching decision module plans a switching path based on the fault location, type, equipment failure to operate risk probability value and the topology, determines the switching window based on the vibration data below the threshold interval, and generates a switching command. The execution verification module performs actions based on the switching command within the switching window. Based on the electrical data verification results after the action, if an anomaly occurs, the topology is updated and the switching decision module is triggered to regenerate the switching command.

2. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, In the status acquisition module, the acquisition of the topology and the collection of electrical data including voltage, current and contact status, switching mechanism operation data and vibration data specifically include the following steps: Analyze the static topology node matrix of the distribution network, establish the connection relationships of each switching device, and obtain the topology structure; By using multi-channel synchronous sampling, the transient bus voltage, main circuit current waveform and auxiliary contact level status of each switching device are acquired under the same time reference as electrical data. Vibration data are constructed by acquiring three-dimensional acceleration, impact envelope, and spatial tilt angle through a multi-axis inertial measurement unit. Extract the current waveform characteristics of the closing and opening coils of each switching device and the historical action time to construct the operating data of the switching mechanism.

3. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, In the operating condition compensation module, generating the compensation electrical data specifically includes the following steps: Time-frequency domain analysis was performed on the vibration data to extract the transient energy peak value and frequency band distribution characterizing the carrier's turbulence and mechanical impact, as disturbance features; Construct a dynamic filtering threshold and anti-shake time constant that are piecewise linearly positively correlated with the transient energy peak; Using the dynamic filtering threshold and anti-jitter time constant, the electrical data within the same time window is masked to filter out high-frequency jitter and voltage / current glitches caused directly by severe mechanical vibration, and the compensated electrical data is output.

4. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, In the fault assessment module, generating the fault location and type specifically includes the following steps: Detect transient out-of-limit mutations in the compensation electrical data, mark them as abnormal features, and record the time of occurrence of the mutations; Calculate the time difference between the time of the mutation and the time of the peak impact energy in the disturbance feature. If the time difference is less than a preset threshold and is not continuous, it is determined that the timing is highly overlapping, thereby reducing the electrical fault confidence of the abnormal feature. The abnormal features are mapped to the topology, and the overcurrent directionality and voltage drop consistency of the upstream and downstream nodes are checked along the feeder direction. If the topology propagation logic of the upstream node detecting overcurrent and the adjacent downstream node not detecting overcurrent is met, and the electrical fault confidence level is up to standard, then it is confirmed as a real electrical fault and the fault location and type are locked.

5. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 4, characterized in that, The steps to reduce the electrical fault confidence of the abnormal characteristics specifically include: Extract the acceleration amplitude corresponding to the peak moment of the impact energy from the disturbance features; The time difference and the acceleration amplitude are input into a preset two-dimensional penalty function, and the confidence decay coefficient is calculated based on an algorithm model that includes a negative exponential term for the time difference and a proportional term for the acceleration amplitude; wherein, the time difference is positively correlated with the confidence decay coefficient, and the acceleration amplitude is negatively correlated with the confidence decay coefficient; Multiply the preset initial fault confidence level by the confidence attenuation coefficient to output the reduced electrical fault confidence level.

6. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, In the health assessment module, generating the probability value of device refusal to operate specifically includes the following steps: Based on the location and type of the fault, identify the set of target switches to be involved in the isolation and recovery operations; Extract the operating data of the switching mechanism corresponding to the target switch set; Analyze the coil current integral area change rate and historical action time drift trend in the operating data of the switching mechanism; The integral area change rate of the coil current and the drift trend of the historical action time are respectively substituted into the preset Weibull distribution lifetime model as independent variables to map the electrical aging failure probability and the mechanical jamming failure probability respectively. Based on a preset multi-risk coupling formula, the electrical aging failure probability and the mechanical jamming failure probability are combined to generate a device failure risk probability value.

7. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, In the handover decision module, generating the handover instruction specifically includes the following steps: Based on the fault location and the topology, a graph theory shortest path algorithm is used to traverse the backup power recovery channel, and during the traversal, switch nodes whose equipment failure risk probability value is higher than the safety threshold are eliminated to generate target isolation and backup power supply path. Using a sliding time window combined with an autoregressive moving average algorithm, the future fluctuation trend of the vibration data is predicted over time. The duration of the impact energy trough and the acceleration being lower than the seismic resistance limit of the switching mechanism is located and marked as the switching window. The action sequence of the target isolation and backup power supply path is packaged with the trigger timestamp of the switching window to generate a switching command.

8. The power distribution room fault self-healing switching control device based on condition monitoring and health assessment according to claim 1, characterized in that, The specific steps of executing the verification module include: When the switching window timestamp is reached, the operating mechanism of the corresponding switch is driven to open or close. After the theoretical operation time has elapsed, the current electrical data is re-acquired, and cross-logic judgment is performed: if the auxiliary contact status indicator has been activated, but the main circuit current and bus voltage have not undergone corresponding on / off step changes, then the corresponding switch is determined to be malfunctioning. The corresponding switch that is determined to be abnormal is marked as a deadlock node in the topology, the topology is updated, and the execution failure interrupt is fed back to the switching decision module to trigger the regeneration of the switching instruction.

Images