Direct-current high-voltage parallel switch fault response method and related equipment

Abstract

The invention provides a direct-current high-voltage parallel switch fault response method and related equipment, and the method comprises the steps: obtaining the sensor data of each direct-current high-voltage switch from a plurality of parallel direct-current high-voltage switches, and extracting the feature vector of each sensor data; inputting the feature vector corresponding to each direct-current high-voltage switch into a pre-trained deep belief network to obtain a fault identification result of each direct-current high-voltage switch; in each direct-current high-voltage switch, when the existence of the fault direct-current high-voltage switch is determined according to each fault identification result, performing fault isolation on the fault direct-current high-voltage switch in combination with the direct-current high-voltage switch related to the fault direct-current high-voltage switch; and after fault isolation, according to the load current of each direct-current high-voltage switch, adjusting the opening speed of each direct-current high-voltage switch by adopting a self-adaptive control algorithm. Therefore, organic combination of switch state sensing, fault early warning and self-adaptive protection is realized, and the operation stability of the direct-current high-voltage parallel switch is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of high-voltage switchgear technology in power systems, and in particular to a fault response method and related equipment for DC high-voltage parallel switches. Background Technology

[0002] DC high-voltage parallel switches are key control devices in power systems, primarily responsible for switching, protecting, and isolating DC circuits. During system operation, the reliability and response speed of these switches directly affect the safety and stability of the power system.

[0003] Currently, most DC high-voltage switches on the market employ mechanical structures or power electronic topologies. Mechanical switches are simple in structure and low in cost, but limited by mechanical inertia, their operating speed is typically tens of milliseconds, making it difficult to meet the requirements of modern DC systems for rapid fault isolation. While power electronic switches can achieve microsecond-level breaking speeds, they suffer from high conduction losses, high costs, and extremely high reliability requirements for the drive circuit. Furthermore, because DC current lacks a natural zero-crossing point, strong arcing is easily generated during the breaking process, leading to contact material ablation and exacerbating operational risks.

[0004] With the rapid development of DC power distribution, rail transit, and new energy power generation, the scale and complexity of DC power systems are constantly increasing, placing higher demands on the reliability and intelligence of switchgear. However, existing DC switches still have shortcomings in terms of state perception, fault early warning, and adaptive protection, which has become a technical bottleneck restricting the safe and stable operation of the system. Summary of the Invention

[0005] The purpose of this application is to at least address one of the aforementioned technical deficiencies, particularly the shortcomings of existing DC switches in terms of state perception, fault warning, and adaptive protection, which have become a technical bottleneck restricting the safe and stable operation of the system.

[0006] In a first aspect, this application provides a fault response method for DC high-voltage parallel switches, the method comprising:

[0007] In multiple DC high-voltage switches connected in parallel, sensor data of each DC high-voltage switch is acquired, and feature vectors of each sensor data are extracted.

[0008] The feature vector corresponding to each DC high-voltage switch is input into a pre-trained deep belief network to obtain the fault identification result of each DC high-voltage switch.

[0009] In each DC high voltage switch, when it is determined that there is a faulty DC high voltage switch based on the fault identification results, the faulty DC high voltage switch is isolated by combining the DC high voltage switches related to the faulty DC high voltage switch.

[0010] After the fault is isolated, the opening speed of each DC high-voltage switch is adjusted according to the load current of each DC high-voltage switch using an adaptive control algorithm.

[0011] In one embodiment, the step of acquiring sensor data for each DC high-voltage switch includes:

[0012] Sensor data for each sensor is acquired via wireless communication from various sensors installed on each DC high-voltage switch. The sensors include at least a temperature sensor installed at the contact point of the DC high-voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high-voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high-voltage switch.

[0013] In one embodiment, the sensor data includes vibration signals, temperature signals, and electric arc signals. The step of extracting feature vectors from each sensor data point includes:

[0014] Wavelet packet decomposition is performed on the vibration signal to calculate the energy distribution of each frequency band and obtain the vibration characteristic components;

[0015] Extract the rising gradient and steady-state value of the temperature signal to obtain the temperature feature components;

[0016] The discharge frequency and intensity distribution of the electric arc signal are statistically analyzed to obtain the characteristic components of the electric arc;

[0017] The vibration characteristic components, temperature characteristic components, and electric arc characteristic components are combined to form a characteristic vector.

[0018] In one embodiment, the training process of a deep belief network includes:

[0019] Determine the DC high voltage switch status training sample set, which includes DC high voltage switch normal status data and DC high voltage switch fault status data;

[0020] The deep belief network was trained unsupervised using a DC high-voltage switch state training sample set.

[0021] Using normal state data and labels of DC high-voltage switches, as well as fault state data and labels of DC high-voltage switches, supervised fine-tuning is performed on the unsupervised trained deep belief network to obtain the trained deep belief network.

[0022] In one embodiment, after fault isolation, the step of adjusting the opening speed of each DC high-voltage switch using an adaptive control algorithm based on the load current of each switch includes:

[0023] After fault isolation, the load current of each DC high-voltage switch is collected in real time, and the state error of each DC high-voltage switch is calculated based on the load current.

[0024] Based on each state error, an adaptive control algorithm generates a tripping speed command for each DC high-voltage switch. The tripping speed command is used to reduce the tripping speed under low current conditions to reduce mechanical shock, and to increase the tripping speed under high current conditions to ensure rapid arc extinguishing.

[0025] In one embodiment, the step of generating the tripping speed command for each DC high-voltage switch by an adaptive control algorithm based on each state error includes:

[0026] Generate the tripping speed command according to the following formula:

[0027]

[0028] in, Let be the opening speed command generated at time t, representing the adjustment amount for the opening speed of each DC high-voltage switch. Let be the state error of each DC high-voltage switch at time t, representing the difference between the actual operating state of the DC high-voltage switch and the expected reference value. , and These represent the proportional, integral, and derivative coefficients for adaptive adjustment, respectively.

[0029] Secondly, this application provides a fault response device for a DC high-voltage parallel switch, the device comprising:

[0030] The sensor data acquisition module is used to acquire sensor data from each of the multiple DC high-voltage switches connected in parallel, and to extract the feature vector of each sensor data.

[0031] The fault identification result acquisition module is used to input the feature vector corresponding to each DC high voltage switch into a pre-trained deep belief network to obtain the fault identification result of each DC high voltage switch.

[0032] The fault isolation module is used to isolate the faulty DC high voltage switch in conjunction with the DC high voltage switches related to the faulty DC high voltage switch when it is determined that there is a faulty DC high voltage switch in each DC high voltage switch based on the fault identification results.

[0033] The tripping speed adjustment module is used to adjust the tripping speed of each DC high-voltage switch according to the load current of each DC high-voltage switch after fault isolation, using an adaptive control algorithm.

[0034] In one embodiment, the sensor data acquisition module includes:

[0035] The sensor data acquisition unit is used to acquire sensor data from each sensor installed on each DC high voltage switch via wireless communication. The sensors include at least a temperature sensor installed at the contact of the DC high voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high voltage switch.

[0036] Thirdly, this application provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of any of the DC high-voltage parallel switch fault response methods described in the above embodiments.

[0037] Fourthly, this application provides a computer device, including: one or more processors, and a memory;

[0038] The memory stores computer-readable instructions, which, when executed by one or more processors, perform the steps of any of the DC high-voltage parallel switch fault response methods described in the above embodiments.

[0039] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0040] The DC high-voltage parallel switch fault response method and related equipment provided in this application acquire sensor data from multiple DC high-voltage switches in parallel in real time and extract their feature vectors, enabling precise modeling of the state of each switch. By inputting the feature vectors into a pre-trained deep belief network, high-accuracy fault identification is achieved, thus quickly locating the faulty switch. For the identified faulty switch, further fault isolation is performed in conjunction with related switches, effectively preventing arc propagation and fault spread, and improving operational safety. After fault isolation, the opening speed is adaptively adjusted using the load current information of each switch, ensuring the reliability of the breaking action while reducing energy loss and contact erosion risk during the breaking process. Therefore, this method not only overcomes the shortcomings of slow response of traditional mechanical switches and high cost and high conduction losses of power electronic switches, but also achieves an organic combination of switch state perception, fault early warning, and adaptive protection, thereby significantly improving the intelligence level of DC high-voltage parallel switches and the overall operational reliability of the system, meeting the requirements of rapid, stable, and safe operation of modern DC power systems. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 A flowchart illustrating the fault response method for DC high-voltage parallel switches provided in this application embodiment;

[0043] Figure 2 This is a schematic diagram of the structure of the DC high-voltage parallel switch fault response device provided in the embodiments of this application;

[0044] Figure 3 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0045] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0046] This application provides a fault response method for DC high-voltage parallel switches. The following embodiments illustrate this method using a computer device as an example. It is understood that the computer device can be any device with data processing capabilities, including but not limited to a single server, server cluster, personal laptop, desktop computer, etc. Figure 1 As shown, the method may include the following steps:

[0047] S101: In multiple DC high-voltage switches connected in parallel, acquire sensor data from each DC high-voltage switch and extract the feature vector of each sensor data.

[0048] In this context, a DC high-voltage parallel switch refers to multiple switching units installed within the same DC power system to perform circuit switching, protection, and fault isolation functions. Each switch is equipped with sensors for condition monitoring. Sensors are electronic components capable of collecting key switch parameters, including but not limited to current, voltage, temperature, and vibration sensors. A feature vector is a multi-dimensional data representation formed by preprocessing, normalizing, and numerically encoding the data collected by sensors, used to characterize the operating state and performance characteristics of each switch.

[0049] In practice, the computer equipment establishes a real-time data transmission channel with multiple parallel DC high-voltage switches, continuously receiving operational data collected by the sensors of each switch. The collected data can include instantaneous and average values ​​of the switch current waveform, voltage fluctuations, real-time changes in contact temperature, and vibration signals of the switch mechanism. These parameters comprehensively reflect the dynamic characteristics of the switch under different load conditions and operating states. The computer equipment can synchronously acquire sensor signals according to a preset sampling frequency and data accuracy to ensure that critical operational status information can be captured during transient faults or sudden load changes.

[0050] The received sensor data is first preprocessed in a computer device to improve data quality and stability. Preprocessing steps may include noise filtering, removing random fluctuations through digital filtering or moving average algorithms; outlier removal, eliminating abnormal points that deviate from the normal range by comparing with historical data ranges or set thresholds; data normalization, mapping sensor data of different dimensions and magnitudes to a unified numerical range to ensure consistent weighting of various data types in subsequent processing; and missing data completion, filling in missing values ​​through interpolation methods or predictions based on historical data to ensure the continuity and completeness of feature vector generation. Through preprocessing, the computer device obtains stable and reliable basic data, providing a sufficient guarantee for high-precision feature extraction.

[0051] After preprocessing, the computer equipment transforms various sensor data into feature vectors based on a preset feature extraction algorithm. Feature extraction can calculate time-domain indicators such as mean, variance, peak value, skewness, and peak-to-peak value of the data using statistical methods. It can also obtain information on periodic changes, harmonic characteristics, and frequency distribution through Fourier transform or wavelet analysis. Furthermore, it can automatically identify key patterns, abnormal trends, or potential correlation features in the data using machine learning methods. Each feature vector comprehensively describes the operating status, load condition, and potential abnormal characteristics of a single switch, providing standardized and quantified data input for subsequent deep learning fault identification, intelligent control, and adaptive protection, enabling unified monitoring and dynamic evaluation of the status of multiple switches.

[0052] It is understandable that acquiring sensor data from each DC high-voltage switch allows for a complete and continuous understanding of the switch's operating status and key parameter changes, avoiding blind spots caused by data omissions or delays. Extracting feature vectors from each sensor's data transforms the complex raw data into a standardized, multi-dimensional state representation, enabling efficient understanding and analysis of switch states and providing reliable input for subsequent fault identification and intelligent control. In this way, unified characterization and dynamic monitoring of multiple switches can be achieved, improving the accuracy and response speed of fault identification and providing a reliable data foundation for the safe operation and stable protection of the system.

[0053] S102: Input the feature vector corresponding to each DC high voltage switch into a pre-trained deep belief network to obtain the fault identification result of each DC high voltage switch.

[0054] Deep belief networks (DPNs) are multi-layer probabilistic generative models pre-trained in computer equipment. They can automatically identify switch states based on input feature vectors and output corresponding fault probabilities or classification results. Fault identification results refer to the determination information of whether each switch is abnormal and the type of abnormality obtained after processing the feature vectors of each switch by computer equipment and inputting them into the DPN.

[0055] In the specific implementation process, the computer equipment first organizes the feature vectors generated by each DC high-voltage switch according to a preset format and order, ensuring that they fully match the input layer dimensions and data structure of the deep belief network. The organized feature vectors can include multi-dimensional information such as time-domain statistical features, frequency-domain features, and pattern features automatically extracted based on machine learning, ensuring that the network can comprehensively perceive the operating status of the switch. The computer equipment can perform inference through the locally stored deep belief network model, or it can transmit the feature vectors to a remote server via the network, where the server performs high-performance computation and returns the recognition result. During the input process, the computer equipment performs standardization processing on the feature vectors, including normalization or zero-mean operations on the data in each dimension, ensuring that the input data is consistent with the feature space during network training, thereby avoiding recognition bias caused by differences in units or data offsets.

[0056] Within a deep belief network, the computer device performs layer-by-layer calculations based on the multi-dimensional state information contained in the feature vector, using weights, biases, and activation functions at each layer to simulate the feature patterns and potential state distributions learned during network training. The network infers the operating state of each switch using a probabilistic generative model, calculates the probability of each possible fault, and comprehensively judges whether the switch is abnormal and the type of abnormality. After acquiring the network output, the computer device can further perform threshold determination or classification processing on the results, transforming continuous probability information into a clear fault identification result. For example, the deep belief network output includes a fault type identification result and an occurrence probability value. When the probability exceeds a warning threshold of 0.85, a warning signal is issued; when it exceeds an action threshold of 0.95, a protective action is immediately initiated.

[0057] By inputting feature vectors into a deep belief network, rapid analysis and intelligent judgment of the status of multiple switches can be achieved. This transforms complex multidimensional sensor data into clear fault identification results, enabling timely detection and location of potential anomalies, and improving fault response speed and accuracy. Therefore, the automatic analysis and reasoning of deep belief networks can effectively overcome the limitations of manual judgment or traditional rule-based methods in processing high-dimensional, multi-source data, providing reliable data support and decision-making basis for intelligent monitoring, rapid fault isolation, and safe and stable operation of DC high-voltage switches.

[0058] S103: When it is determined that there is a faulty DC high-voltage switch among each DC high-voltage switch based on the fault identification results, the faulty DC high-voltage switch is isolated by combining the DC high-voltage switches related to the faulty DC high-voltage switch.

[0059] Among these, the relevant DC high-voltage switch refers to other switches associated with the faulty switch's load, topology, or control logic, whose operating state may affect or be constrained by the faulty switch's state. Fault isolation refers to disconnecting the faulty switch from the circuit by controlling the coordinated action of relevant switches, thereby preventing the fault from spreading and protecting system safety.

[0060] In the specific implementation process, the computer equipment first receives the fault identification results of each DC high-voltage switch after deep belief network analysis, and combines this with pre-stored topology information and real-time load distribution data to analyze the circuit areas that may be affected by each faulty switch. By analyzing the load relationships, interconnection paths, and redundancy configurations between switches, the computer equipment determines the potential risk range of the faulty switch, as well as the related switches and load devices that may be affected.

[0061] Subsequently, the computer equipment, according to preset control strategies and algorithms, identifies the relevant switches requiring joint operation and generates detailed fault isolation control instructions based on fault severity, load balance, and safety priorities. These instructions clearly specify which switches need to be tripped, the tripping sequence, the time interval between each switch operation, and necessary synchronization control measures to ensure that isolating faulty switches does not impact normally operating switches or loads. When generating instructions, the computer equipment can also consider real-time parameters such as current, voltage, and contact temperature to optimize the tripping process and prevent excessive arcing or load fluctuations during switch disconnection.

[0062] Subsequently, the computer equipment sends fault isolation control commands to the control modules of each relevant switch via the communication interface, and the switches execute the opening operation sequentially according to the commands. The switch execution unit adjusts the operating mechanism's speed and sequence of actions based on the received commands to safely isolate the faulty switch, cut off the fault circuit, and prevent arc propagation and fault spread from affecting other normal switches. Throughout the entire operation, the computer equipment monitors the status feedback of each switch in real time to ensure the opening operation is completed smoothly according to the predetermined plan, and can immediately issue adjustment or re-control commands upon detecting abnormalities, ensuring the overall safety and stability of the system.

[0063] By isolating faults through joint operation of related switches, the abnormal switch can be quickly disconnected when a fault occurs, while maintaining the normal operation of other switches, thus avoiding unnecessary impact on the overall load and power supply stability. Therefore, systematic and automated fault isolation control can improve the response speed and accuracy of fault handling, reduce the risk of equipment damage or power outages caused by fault propagation, provide reliable protection for the safe and stable operation of DC high-voltage parallel switch systems, and provide a safe foundation for subsequent adaptive control and load adjustment.

[0064] S104: After fault isolation, the opening speed of each DC high-voltage switch is adjusted according to the load current of each DC high-voltage switch using an adaptive control algorithm.

[0065] Load current refers to the actual current carried by each switch in the circuit under normal or fault isolation conditions, and is an important parameter for measuring the load status of the switches. Adaptive control algorithms are intelligent control programs running in computer equipment that can automatically adjust the switch operating speed based on the load current, topology information, and historical operating data of each switch to optimize the opening process. Opening speed refers to the movement speed of the switch contacts when performing the disconnection action; its control accuracy directly affects the intensity of the arc generation, circuit stability, and equipment safety.

[0066] In the implementation process, the computer equipment first collects the load current data of each DC high-voltage switch in real time after fault isolation, and then performs a comprehensive analysis based on the topological location of each switch, load distribution, and its relationship with other switches. During the analysis, the computer equipment can filter the collected current signals to remove transient noise or interference, and correct for abnormal fluctuations using historical load data and threshold models to ensure the accuracy and reliability of the load current data. Through a comprehensive evaluation of real-time current, switch mechanical characteristics, and system circuit safety constraints, the computer equipment can accurately determine the tripping operation requirements of each switch under the current load conditions.

[0067] The computer equipment dynamically optimizes the opening speed of each switch using an adaptive control algorithm. Specifically, it adjusts the opening speed of each switch appropriately according to the load current, ensuring the action is neither too fast nor too slow. This balances the smoothness of switch operation with system response efficiency, thereby maintaining safety and stability while aiding in overall fault handling and system recovery. During speed optimization, the computer equipment continuously adjusts the speed curve by considering the mechanical response delay of the switches and the inductance characteristics of the circuit, ensuring that the action of each switch meets safety constraints while also meeting the requirements for rapid system recovery.

[0068] Subsequently, the computer equipment sends the calculated tripping speed control command to the operation modules of each switch via the communication interface. Each switch precisely adjusts the action speed and acceleration curve of its operating mechanism according to the received command, and executes the optimized tripping operation. During the operation, the computer equipment continuously monitors the switch contact status, current changes, and operation progress, and fine-tunes the tripping speed based on real-time feedback to ensure that the entire disconnection process is smooth, reliable, and safe.

[0069] After fault isolation, the tripping speed is adjusted using an adaptive control algorithm based on the load current of each DC high-voltage switch. This allows for dynamic determination of the operating rhythm according to the load conditions of different switches, ensuring that the tripping is neither too fast nor too slow, thus balancing the smoothness of switch operation and the system's response efficiency. In this way, while ensuring switch safety and mechanical lifespan, the timeliness of fault disconnection and system recovery capability are improved, reducing the risk of arcing and contact wear, and ensuring that the DC high-voltage parallel switch system can still operate safely, stably, and efficiently under abnormal conditions.

[0070] In the above embodiments, by acquiring sensor data from multiple parallel DC high-voltage switches in real time and extracting their feature vectors, the state of each switch can be accurately modeled. Inputting the feature vectors into a pre-trained deep belief network enables high-accuracy fault identification, thereby quickly locating the faulty switch. For the identified faulty switch, further fault isolation is achieved in conjunction with related switches, effectively preventing arc propagation and fault spread, and improving operational safety. After fault isolation, the opening speed is adaptively adjusted using the load current information of each switch, ensuring the reliability of the breaking action while reducing energy loss and contact erosion risk during the breaking process. Therefore, this method not only overcomes the shortcomings of slow response of traditional mechanical switches and high cost and high conduction losses of power electronic switches, but also achieves an organic combination of switch state perception, fault early warning, and adaptive protection, thereby significantly improving the intelligence level of parallel DC high-voltage switches and the overall operational reliability of the system, meeting the requirements of rapid, stable, and safe operation of modern DC power systems.

[0071] In one embodiment, the step of acquiring sensor data for each DC high-voltage switch includes:

[0072] Sensor data for each sensor is acquired via wireless communication from various sensors installed on each DC high-voltage switch. The sensors include at least a temperature sensor installed at the contact point of the DC high-voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high-voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high-voltage switch.

[0073] Among them, the temperature sensor is a device installed at the switch contacts to detect changes in the working temperature of the contacts. The ultraviolet light sensor is a device installed around the switch's arc-extinguishing chamber to detect the ultraviolet light signal emitted when an electric arc is generated. The vibration acceleration sensor is a device installed on the switch's operating mechanism to measure changes in the vibration acceleration of the operating mechanism during operation.

[0074] In practice, the computer equipment initiates data acquisition requests to the control module of each DC high-voltage switch via a wireless communication interface. Sensors on each switch collect corresponding physical quantities in real time and transmit the sensor data to the computer equipment via wireless signals. Upon receiving the data, the computer equipment performs unified formatting, time synchronization, and preliminary verification of the data from different types of sensors to ensure data integrity and reliability. For example, temperature sensor data reflects the thermal state of the contacts, ultraviolet light sensor data detects potential arcing, and vibration acceleration sensor data monitors the stability and abnormal vibration of the operating mechanism. Through centralized acquisition and management of this sensor data, the computer equipment can monitor the critical operating status of each switch in real time.

[0075] In one example, the DC high-voltage switch includes arc-resistant alloy contacts and a double-break arc-extinguishing structure to ensure reliable breaking operation under DC high-voltage conditions. Monitoring sensors are precisely positioned at key locations on the switch body using a distributed layout. A high-precision PT100 platinum resistance temperature sensor is embedded within the moving contact support at the contact location, and a slip ring lead is used to address the issue of moving contact, enabling real-time monitoring of the contact temperature. A UVTRON ultraviolet light sensor is installed around the arc-extinguishing chamber, enabling arc detection through the chamber's viewing window, with sensitivity down to the single-photon level. An ICP acceleration vibration sensor is positioned at key nodes of the operating mechanism, installed at the output bearing housing, with a frequency response range of 0.5 to 10 kHz, to monitor the mechanism's vibration status. Furthermore, a Rogowski coil enables precise measurement of large currents. All sensor signals are connected to a data acquisition unit via shielded cables. This unit uses a 16-bit ADC chip with a sampling rate of 100 kHz, accurately capturing microsecond-level transient characteristic signals. It also features electromagnetic interference immunity and is adaptable to industrial-grade temperature environments ranging from -40 to 85 degrees Celsius. The communication module supports LoRaWAN protocol and 5G dual-mode transmission to ensure reliable uploading of collected data. The intelligent control unit adopts a multi-core processor architecture, with one core dedicated to real-time signal processing and another core running intelligent diagnostic algorithms to achieve comprehensive monitoring of switch status, fault identification, and intelligent control.

[0076] Data is acquired from various sensors on each DC high-voltage switch via wireless communication, enabling real-time monitoring of switch contact temperature, arc status in the arc-extinguishing chamber, and vibration status of the operating mechanism. This allows the computer equipment to comprehensively understand the key operating parameters and potential anomalies of each switch. Temperature sensors monitor the thermal state of the contacts, ultraviolet sensors detect arc activity, and vibration acceleration sensors assess the operational stability of the operating mechanism. By centrally collecting this data, the computer equipment provides reliable input for feature extraction, fault identification, and intelligent control, achieving real-time monitoring and early warning of switch operating status. This improves the system's response speed and accuracy to abnormal situations, thereby ensuring the safe, stable, and efficient operation of the DC high-voltage parallel switches.

[0077] In one embodiment, the sensor data includes vibration signals, temperature signals, and electric arc signals. The step of extracting feature vectors from each sensor data point includes:

[0078] Wavelet packet decomposition is performed on the vibration signal to calculate the energy distribution of each frequency band and obtain the vibration characteristic components;

[0079] Extract the rising gradient and steady-state value of the temperature signal to obtain the temperature feature components;

[0080] The discharge frequency and intensity distribution of the electric arc signal are statistically analyzed to obtain the characteristic components of the electric arc;

[0081] The vibration characteristic components, temperature characteristic components, and electric arc characteristic components are combined to form a characteristic vector.

[0082] Among them, vibration signal refers to mechanical vibration data collected by an accelerometer installed on the DC high-voltage switch operating mechanism, used to reflect the stability and abnormal vibration state during switch operation. Temperature signal refers to temperature change data collected by a temperature sensor at the contact point, used to reflect the thermal characteristics of the switch contacts under operating or fault conditions. Arc signal refers to arc discharge data collected by an ultraviolet light sensor in the arc-extinguishing chamber, used to reflect arc activity that may occur during switch opening and closing. Vibration characteristic component, temperature characteristic component, and arc characteristic component are key parameters that reflect the switch state extracted from the original signal through specific algorithms.

[0083] In the specific implementation process, the computer equipment first performs wavelet packet decomposition on the vibration signal collected from the vibration acceleration sensor of the operating mechanism, dividing the original signal into multiple sub-signals with different frequency bands, and calculating the energy distribution of each frequency band to quantify the intensity and variation of vibration in different frequency bands. The computer equipment can further normalize the frequency band energy to eliminate the influence of amplitude differences on subsequent analysis, thereby obtaining vibration characteristic components that can comprehensively reflect the stability and potential anomalies of the operating mechanism. These vibration characteristic components can reveal the transient vibration characteristics and possible abnormal modes of the switch during operation, providing reliable input information for subsequent fault identification.

[0084] Subsequently, the computer equipment analyzes the temperature signal collected by the contact temperature sensor. First, it calculates the temperature gradient over time, reflecting the rate of temperature change during energization or disconnection. Simultaneously, it extracts the steady-state temperature value to describe the thermal equilibrium state of the contact under long-term load. The computer equipment can filter the temperature signal and remove outliers to ensure the accuracy of gradient calculation and steady-state value extraction. The obtained temperature characteristic components effectively reflect the thermal response characteristics of the contact under different load conditions and potential faults, providing a basis for judging contact overheating or thermal anomalies.

[0085] For arc signals, the computer equipment first processes the arc discharge data acquired from the ultraviolet light sensor, including noise removal and peak correction. Then, the computer equipment statistically analyzes the arc discharge frequency and the intensity distribution of each discharge, thereby generating arc characteristic components. These characteristic components can quantify potential arc anomalies that may occur during the switch breaking process, revealing the frequency and intensity characteristics of arc generation between contacts, providing a basis for identifying arc anomalies and assessing switch safety.

[0086] Computer equipment extracts key features from three types of signals: vibration, temperature, and electric arc, forming a multi-dimensional feature vector for each switch. This enables a comprehensive description of the switch status and early identification of potential faults, significantly improving the system's responsiveness to abnormal operation and overall reliability.

[0087] It should be noted that wavelet packet decomposition of the vibration signal and calculation of the energy distribution in each frequency band can extract the vibration characteristics of the switch operating mechanism in different frequency ranges, thereby obtaining vibration characteristic components that reflect the mechanical state and potential anomalies. Extracting the rise gradient and steady-state value of the temperature signal can accurately reflect the thermal response of the contacts under different load or fault conditions, thus obtaining the temperature characteristic component. Statistically analyzing the discharge frequency and intensity distribution of the arc signal can quantify the arc activity characteristics of the switch during the breaking process, thus obtaining the arc characteristic component. Combining the vibration characteristic component, temperature characteristic component, and arc characteristic component to form a feature vector can comprehensively reflect the mechanical, thermal, and arc states of the switch, providing multi-dimensional and reliable input data for fault identification and intelligent control, thereby improving the system's accuracy and response capability to abnormal operation, and ensuring the safe, stable, and efficient operation of the DC high-voltage switch.

[0088] In one embodiment, the training process of a deep belief network includes:

[0089] Determine the DC high voltage switch status training sample set, which includes DC high voltage switch normal status data and DC high voltage switch fault status data;

[0090] The deep belief network was trained unsupervised using a DC high-voltage switch state training sample set.

[0091] Using normal state data and labels of DC high-voltage switches, as well as fault state data and labels of DC high-voltage switches, supervised fine-tuning is performed on the unsupervised trained deep belief network to obtain the trained deep belief network.

[0092] The DC high-voltage switch state training sample set refers to the feature vector data collection from each DC high-voltage switch. This set includes normal state data and its corresponding labels representing the switch's normal operation, and fault state data and its corresponding labels representing the switch's abnormal or fault states, used to train the deep belief network. Unsupervised training refers to using computer equipment to enable the deep belief network to learn the latent distribution and patterns of input features without state labels. Fine-tuning refers to adjusting the network weights using labeled normal state data and fault state data based on the already unsupervised trained network, in order to improve the network's accuracy in recognizing different switch states.

[0093] In the specific implementation process, the computer equipment first determines the training sample set of DC high-voltage switch states, and organizes and labels the collected normal state feature vectors and fault state feature vectors to form a complete training dataset. Subsequently, the computer equipment inputs the training sample set into a pre-constructed deep belief network for unsupervised training, enabling the network to automatically learn the inherent distribution characteristics and potential patterns of various feature vectors. Through unsupervised training, the deep belief network can capture the overall characteristics of the switch state data in the initial stage, providing a good weight initialization for subsequent supervised fine-tuning.

[0094] After unsupervised training, the computer equipment further uses labeled normal state data and their labels, and fault state data and their labels, to perform supervised fine-tuning of the deep belief network. During the fine-tuning process, the computer equipment optimizes the network weights using the backpropagation algorithm, enabling the network to more accurately distinguish the feature patterns of normal and fault states. Through this training strategy, a trained deep belief network can be obtained, which can quickly and accurately identify faults in real-time acquired switch feature vectors, achieving intelligent monitoring and early warning of DC high-voltage switch status.

[0095] In one example, a sample database containing typical operating conditions such as normal state, arc fault, mechanical jamming, and contact deterioration is first constructed, and no less than one thousand valid samples are collected. A complete experimental platform is built in the laboratory, using a programmable load to generate fault currents of different amplitudes, a heating device to simulate contact overheating, and a mechanical limit device to simulate jamming faults, thus covering various operating conditions. The collected vibration, temperature, and arc signals undergo feature engineering processing: the vibration signal is decomposed into four layers of wavelet packets and the energy distribution of each frequency band is calculated to extract vibration features; the temperature signal's rising gradient and steady-state value are extracted to obtain temperature features; and the arc signal's discharge frequency and intensity distribution are statistically analyzed to obtain arc features. The vibration, temperature, and arc features are combined to form a multi-dimensional feature vector, which is then input into a deep belief network for training. The deep belief network adopts a five-layer hidden layer structure, with ReLU activation function, an initial learning rate of 0.01, and decays once every fifty rounds. Through a combination of unsupervised pre-training and supervised fine-tuning, the network can accurately learn various fault features and achieve precise identification of fault types. In the engineering application phase, sensor data is collected at fixed intervals, and feature vectors are generated through the same feature extraction process. These vectors are then input into a trained deep belief network model. The model output includes fault type identification results and corresponding occurrence probabilities, enabling real-time monitoring, fault early warning, and intelligent diagnosis of the DC high-voltage switch status.

[0096] It is understandable that defining a training sample set for DC high-voltage switch states provides a diverse dataset covering normal operation and various fault states for the deep belief network (DBN), enabling the network to learn the characteristic distribution of the switch under different operating conditions. Unsupervised training of the pre-built DBN using the training sample set allows the network to automatically capture the latent patterns and internal structure of input features, providing a good initialization of weights for subsequent classification tasks. Subsequently, supervised fine-tuning of the unsupervised-trained network using labeled normal and fault state data further optimizes the network parameters, improving its ability to distinguish between different state features, thus obtaining the trained DBN. Based on this, the DBN can accurately identify and classify the states of DC high-voltage switches, improving the system's ability to perceive and respond to abnormal operations, ensuring switch safety and system reliability.

[0097] In one embodiment, after fault isolation, the step of adjusting the opening speed of each DC high-voltage switch using an adaptive control algorithm based on the load current of each switch includes:

[0098] After fault isolation, the load current of each DC high-voltage switch is collected in real time, and the state error of each DC high-voltage switch is calculated based on the load current.

[0099] Based on each state error, an adaptive control algorithm generates a tripping speed command for each DC high-voltage switch. The tripping speed command is used to reduce the tripping speed under low current conditions to reduce mechanical shock, and to increase the tripping speed under high current conditions to ensure rapid arc extinguishing.

[0100] State error refers to an index calculated by computer equipment based on the deviation between the actual load current and the expected or target current, used to measure the difference between the current state of the switch and the ideal state. Opening speed command refers to a control signal generated by computer equipment and sent to the switch operation module, used to adjust the speed of the switch's opening action to optimize mechanical and electrical characteristics.

[0101] In the implementation process, after fault isolation, the computer equipment collects load current data from each DC high-voltage switch in real time via wired or wireless communication interfaces. The collected data includes the instantaneous current amplitude, average current, and short-term fluctuations of each switch. To ensure the accuracy of the calculated state error, the computer equipment performs multi-level processing on the collected data, including removing random noise, eliminating outliers, smoothing signals, and interpolating missing short-term data to obtain stable and reliable load current information. Subsequently, the computer equipment compares the processed load current data with the preset target current value to calculate the state error of each switch. The state error specifically quantifies the deviation between the actual action and the expected action of the switch under the current load conditions, providing precise input for the adaptive control algorithm.

[0102] The computer equipment inputs the state error of each switch into the adaptive control algorithm module. The algorithm, combined with the switch's mechanical characteristics, contact tolerance, and arc extinguishing behavior, dynamically generates a tripping speed command for each DC high-voltage switch. Under low-current conditions, the tripping speed command appropriately reduces the switch's operating speed, resulting in smoother contact separation, reduced mechanical shock and wear, and extended switch life. Under high-current conditions, the tripping speed command appropriately increases the switch's operating speed to accelerate the disconnection process, ensure rapid arc extinguishing, and prevent arc propagation.

[0103] The computer equipment transmits the generated tripping speed command to the operation modules of each switch in real time via a communication interface. Each switch precisely adjusts its tripping mechanism according to the command to achieve optimized disconnection. During execution, the computer equipment continuously monitors the tripping status and current changes of each switch, and dynamically fine-tunes the tripping speed through closed-loop feedback to ensure a smooth, safe, and reliable tripping process. This method enables intelligent adjustment of the tripping speed under different load conditions, improving the system's rapid response capability, reducing mechanical and electrical losses, and enhancing the operational safety and reliability of the DC high-voltage switch and the entire system.

[0104] In one embodiment, the step of generating the tripping speed command for each DC high-voltage switch by an adaptive control algorithm based on each state error includes:

[0105] Generate the tripping speed command according to the following formula:

[0106]

[0107] in, Let be the opening speed command generated at time t, representing the adjustment amount for the opening speed of each DC high-voltage switch. Let be the state error of each DC high-voltage switch at time t, representing the difference between the actual operating state of the DC high-voltage switch and the expected reference value. , and These represent the proportional, integral, and derivative coefficients for adaptive adjustment, respectively.

[0108] In this embodiment, the adaptive control algorithm generates a tripping speed command based on the state error of each DC high-voltage switch, and uses a control strategy combining proportional, integral, and derivative operations to dynamically adjust the tripping action. The proportional part can adjust the tripping speed in real time according to the current state error, enabling the switch to respond quickly to load deviations; the integral part corrects by accumulating historical errors, which can eliminate the cumulative effect of continuous deviations and ensure the accuracy of the tripping action; the derivative part predicts future trends based on the error change rate and makes forward-looking adjustments to the tripping speed to reduce overshoot and oscillation.

[0109] This method enables computer equipment to perform real-time, continuous, and smooth optimized control of the opening speed of each switch. Under low load current conditions, the opening speed can be automatically reduced, making the disconnection action smoother and reducing mechanical shock and contact wear; under high load current conditions, the opening speed can be automatically increased, ensuring rapid arc extinguishing, shortening the disconnection time, and improving the system's fault response efficiency. Thus, adaptive adjustment of DC high-voltage switches under different load conditions is achieved, improving the safety, reliability, and overall system operational stability of the switches, while extending equipment life and reducing maintenance costs.

[0110] The following describes the DC high-voltage parallel switch fault response device provided in the embodiments of this application. The DC high-voltage parallel switch fault response device described below can be referred to in correspondence with the DC high-voltage parallel switch fault response method described above. Figure 2 As shown, this application provides a DC high-voltage parallel switch fault response device, the device comprising:

[0111] The sensor data acquisition module 201 is used to acquire sensor data of each DC high voltage switch in multiple DC high voltage switches connected in parallel, and extract the feature vector of each sensor data.

[0112] The fault identification result acquisition module 202 is used to input the feature vector corresponding to each DC high voltage switch into a pre-trained deep belief network to obtain the fault identification result of each DC high voltage switch.

[0113] The fault isolation module 203 is used to isolate the faulty DC high voltage switch in conjunction with the DC high voltage switches related to the faulty DC high voltage switch when it is determined that there is a faulty DC high voltage switch in each DC high voltage switch based on the fault identification results.

[0114] The tripping speed adjustment module 204 is used to adjust the tripping speed of each DC high-voltage switch according to the load current of each DC high-voltage switch after fault isolation, using an adaptive control algorithm.

[0115] In one embodiment, the sensor data acquisition module 201 includes:

[0116] The sensor data acquisition unit is used to acquire sensor data from each sensor installed on each DC high voltage switch via wireless communication. The sensors include at least a temperature sensor installed at the contact of the DC high voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high voltage switch.

[0117] In one embodiment, the sensor data includes vibration signals, temperature signals, and arc signals, and the sensor data acquisition module 201 includes:

[0118] The vibration characteristic component determination unit is used to perform wavelet packet decomposition on the vibration signal, calculate the energy distribution of each frequency band, and obtain the vibration characteristic components.

[0119] The temperature feature component determination unit is used to extract the rising gradient and steady-state value of the temperature signal to obtain the temperature feature components.

[0120] The arc characteristic component determination unit is used to statistically analyze the discharge frequency and intensity distribution of the arc signal to obtain the arc characteristic components;

[0121] The eigenvector determination unit is used to combine vibration characteristic components, temperature characteristic components and electric arc characteristic components to form an eigenvector.

[0122] In one embodiment, the fault identification result acquisition module 202 includes:

[0123] The training sample set determination unit is used to determine the DC high voltage switch status training sample set, which includes DC high voltage switch normal status data and DC high voltage switch fault status data.

[0124] The unsupervised training unit for deep belief networks is used to train a pre-built deep belief network in unsupervised manner using a DC high-voltage switch state training sample set.

[0125] Shenzhen Zhixin Network has a supervised fine-tuning unit, which uses normal state data and labels of DC high-voltage switches, as well as fault state data and labels of DC high-voltage switches, to perform supervised fine-tuning on the unsupervised trained deep belief network, thus obtaining the trained deep belief network.

[0126] In one embodiment, the tripping speed adjustment module 204 includes:

[0127] The state error calculation unit is used to collect the load current of each DC high voltage switch in real time after fault isolation, and calculate the state error of each DC high voltage switch based on the load current.

[0128] The tripping speed adjustment unit is used to generate tripping speed commands for each DC high-voltage switch based on each state error using an adaptive control algorithm. The tripping speed commands are used to reduce the tripping speed under low current conditions to reduce mechanical shock, and to increase the tripping speed under high current conditions to ensure rapid arc extinguishing.

[0129] In one embodiment, the step of generating the tripping speed command for each DC high-voltage switch by an adaptive control algorithm based on each state error includes:

[0130] Generate the tripping speed command according to the following formula:

[0131]

[0132] in, Let be the opening speed command generated at time t, representing the adjustment amount for the opening speed of each DC high-voltage switch. Let be the state error of each DC high-voltage switch at time t, representing the difference between the actual operating state of the DC high-voltage switch and the expected reference value. , and These represent the proportional, integral, and derivative coefficients for adaptive adjustment, respectively.

[0133] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the DC high-voltage parallel switch fault response method as described in any of the above embodiments.

[0134] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the DC high-voltage parallel switch fault response method as described in any of the above embodiments.

[0135] Indicatively, such as Figure 3 As shown, Figure 3 This is a schematic diagram of the internal structure of a computer device 300 provided in an embodiment of this application. The computer device 300 can be provided as a server. (Refer to...) Figure 3The computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions, such as application programs, that can be executed by the processing component 302. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 302 is configured to execute instructions to perform the DC high-voltage parallel switch fault response method of any of the above embodiments.

[0136] The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate on an operating system stored in memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0137] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0138] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, "a," "an," "the," "the," and "its" may also include plural forms unless the context clearly indicates otherwise. "Multiple" refers to at least two, such as 2, 3, 5, or 8, etc. "And / or" includes any and all combinations of the related listed items.

[0139] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0140] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fault response method for DC high-voltage parallel switches, characterized in that, The method includes: In a series of DC high-voltage switches connected in parallel, sensor data of each DC high-voltage switch is acquired, and feature vectors of each sensor data are extracted. The feature vector corresponding to each DC high-voltage switch is input into a pre-trained deep belief network to obtain the fault identification result of each DC high-voltage switch. In each of the aforementioned DC high-voltage switches, when it is determined that there is a faulty DC high-voltage switch based on the fault identification results, the faulty DC high-voltage switch is isolated by combining the DC high-voltage switches associated with the faulty DC high-voltage switch. After the fault is isolated, the opening speed of each DC high-voltage switch is adjusted according to the load current of each DC high-voltage switch using an adaptive control algorithm.

2. The fault response method for DC high-voltage parallel switches according to claim 1, characterized in that, The step of acquiring sensor data for each of the DC high-voltage switches includes: Sensor data of each sensor is acquired from each sensor installed on each DC high voltage switch via wireless communication. The sensors include at least a temperature sensor installed at the contact of the DC high voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high voltage switch.

3. The fault response method for DC high-voltage parallel switches according to claim 1, characterized in that, The sensor data includes vibration signals, temperature signals, and electric arc signals. The step of extracting feature vectors from each of the sensor data includes: The vibration signal is decomposed by wavelet packet analysis to calculate the energy distribution of each frequency band and obtain the vibration characteristic components. The rising gradient and steady-state value of the temperature signal are extracted to obtain the temperature feature components; The discharge frequency and intensity distribution of the electric arc signal are statistically analyzed to obtain the characteristic components of the electric arc; The vibration characteristic component, the temperature characteristic component, and the electric arc characteristic component are combined to form the characteristic vector.

4. The fault response method for DC high-voltage parallel switches according to claim 1, characterized in that, The training process of the deep belief network includes: Determine the DC high voltage switch status training sample set, which includes DC high voltage switch normal status data and DC high voltage switch fault status data; The pre-constructed deep belief network is trained unsupervised using the DC high-voltage switch state training sample set. Using the normal state data and labels of the DC high voltage switch, and the fault state data and labels of the DC high voltage switch, the unsupervised deep belief network is fine-tuned in a supervised manner to obtain the trained deep belief network.

5. The fault response method for DC high-voltage parallel switches according to claim 1, characterized in that, The step of adjusting the opening speed of each DC high-voltage switch according to its load current and using an adaptive control algorithm after fault isolation includes: After fault isolation, the load current of each DC high-voltage switch is collected in real time, and the state error of each DC high-voltage switch is calculated based on the load current. Based on each state error, the adaptive control algorithm generates a tripping speed command for each DC high-voltage switch. The tripping speed command is used to reduce the tripping speed under low current conditions to reduce mechanical shock, and to increase the tripping speed under high current conditions to ensure rapid arc extinguishing.

6. The fault response method for DC high-voltage parallel switches according to claim 5, characterized in that, The step of generating the opening speed command corresponding to each DC high-voltage switch by the adaptive control algorithm based on each state error includes: Generate the tripping speed command according to the following formula: in, Let be the opening speed command generated at time t, representing the adjustment amount for the opening speed of each DC high-voltage switch. Let be the state error of each of the DC high-voltage switches at time t, representing the difference between the actual operating state of the DC high-voltage switch and the expected reference value. , and These represent the proportional, integral, and derivative coefficients for adaptive adjustment, respectively.

7. A fault response device for a DC high-voltage parallel switch, characterized in that, The device includes: The sensor data acquisition module is used to acquire sensor data of each of the multiple DC high voltage switches connected in parallel, and extract the feature vector of each sensor data. The fault identification result acquisition module is used to input the feature vector corresponding to each DC high voltage switch into a pre-trained deep belief network to obtain the fault identification result of each DC high voltage switch. The fault isolation module is used to isolate the faulty DC high voltage switch in conjunction with the DC high voltage switches related to the faulty DC high voltage switch when it is determined that there is a faulty DC high voltage switch in each of the aforementioned DC high voltage switches based on the fault identification results. The tripping speed adjustment module is used to adjust the tripping speed of each DC high-voltage switch according to the load current of each DC high-voltage switch after fault isolation, using an adaptive control algorithm.

8. The DC high-voltage parallel switch fault response device according to claim 7, characterized in that, The sensor data acquisition module includes: The sensor data acquisition unit is used to acquire sensor data of each sensor installed on each DC high voltage switch via wireless communication. The sensors include at least a temperature sensor installed at the contact of the DC high voltage switch, an ultraviolet light sensor installed around the arc-extinguishing chamber of the DC high voltage switch, and a vibration acceleration sensor installed on the operating mechanism of the DC high voltage switch.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the DC high-voltage parallel switch fault response method as described in any one of claims 1 to 6.

10. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the DC high-voltage parallel switch fault response method as described in any one of claims 1 to 6.

Images