Solid-state transformer fault early warning and isolation system based on protocol access

Abstract

The invention relates to the technical field of power electronics and transformers, in particular to a solid-state transformer fault early warning and isolation system based on protocol access, which is characterized in that when a solid-state transformer is started, a data frame generated by equipment is collected and stored by a sensor through a local control module; the protocol analysis module receives the equipment data frames and extracts information in the equipment data frames in a dynamic analysis mode; the fault early warning module analyzes the analyzed data based on a fault prediction model, predicts possible transformer faults, and evaluates prediction results; the fault isolation module is used for positioning equipment needing to be isolated according to an evaluation result, and generating related fault information by disconnecting the equipment from other components; the communication interface module reports the fault information to a central server or superior equipment for further analysis and processing of fault data; according to the system, real-time monitoring, fault early warning and intelligent isolation of the solid-state transformer are realized, and the operation safety and stability of equipment are improved.

Description

Technical Field

[0001] This invention belongs to the field of power electronics and transformer technology, specifically relating to a solid-state transformer fault early warning and isolation system based on protocol access. Background Technology

[0002] Traditional fault handling processes typically rely on application-layer protocols for fault diagnosis. However, the lengthy protocol parsing and diagnostic processes lead to delayed fault response, impacting network and device stability. Traditional solutions often depend on a central server for fault diagnosis and response, which not only increases the central server's load but can also create system bottlenecks. This is especially true in large-scale device deployments, where the central server may be unable to handle excessive request volumes. Fault detection within devices such as solid-state drives is often limited, and traditional detection methods struggle to detect faults locally in real time. Furthermore, relying on a central server or upper-layer applications for fault diagnosis results in delays and degraded network performance. Existing fault diagnosis methods typically depend on application-layer protocols, resulting in complex and slow processes. The large amounts of device data require centralized processing on a central server, increasing system load, reducing response speed, and hindering local fault diagnosis and real-time response, ultimately impacting network and device reliability. Summary of the Invention

[0003] The purpose of this invention is to propose a solid-state transformer fault early warning and isolation system based on protocol access. This system acquires data frames generated by sensors deployed in the transformer, parses these data frames dynamically, evaluates the fault level of the devices based on the parsed data frames, implements isolation measures based on the fault level, and generates a control report that is synchronously sent to a central server. This achieves parsing and response, significantly improving the efficiency of fault early warning and isolation for transformers under various operating modes. To achieve the above objective, according to one aspect of the invention, a solid-state transformer fault early warning and isolation system based on protocol access is provided. The system includes a local control module, a protocol parsing module, a fault isolation module, and a communication interface module. Local control module: When the solid-state transformer starts working, it transmits and stores the generated device data frames through sensors; Protocol parsing module: Receives device data frames from the local control module and parses the device data frames dynamically; Fault early warning module: Receives parsed data from the protocol parsing module, predicts transformer faults based on the fault prediction model, and evaluates the prediction results; Fault isolation module: Receives the evaluation results from the fault early warning module, locates the equipment that needs to be isolated, disconnects the equipment from other components, and generates fault information; Communication interface module: Sends fault information to the central server or upstream equipment for fault data reporting and analysis.

[0004] Furthermore, the local control module includes: Sensor Deployment Unit: Deploys sensors to the data interfaces of various devices according to sensor deployment rules and receives device data at the interfaces.

[0005] Data filtering unit: Used to filter out data that does not meet the requirements using filtering rules for abnormal data, so as to ensure data stability.

[0006] Data transmission unit: Securely and losslessly transmits data using encryption and lossless compression algorithms to ensure security during data transmission.

[0007] Data storage unit: Stores the processed data to a distributed database.

[0008] Furthermore, the measurement terminal includes, but is not limited to, any one or more of the following: smart meters, current sensors, voltage sensors, fiber optic sensors, temperature and humidity sensors, environmental monitoring terminals, and data loggers.

[0009] Furthermore, the filtering rules are the Z-score algorithm and the filtering model, and the filtering model is a data filtering model constructed by a multilayer perceptron.

[0010] Furthermore, the step of receiving device data frames from the local control module and parsing the device data frames dynamically includes the following steps: A1. Construct a point-to-point interaction system between the sensor deployment module and the data parsing module, and construct a feature data system within the data acquisition room based on the point-to-point interaction system; A2, when parsing data, constructs data-protocol response features based on the characteristic data system and parsing parameters within the data; A3, calibrating protocol response standards based on data-protocol response characteristics and historical interaction data; A4, dynamically adjusts parsing parameters based on real-time transmitted data and calibration protocol response standards; Furthermore, in A2, the method for constructing the point-to-end interaction system specifically includes: acquiring all protocols supported by the sensor data acquisition terminal to form an acquisition protocol set, wherein the protocol includes corresponding parsing rules; acquiring the receiving protocols supported by the transmission port between the protocol parsing module and the sensor data acquisition module to form a receiving protocol set, wherein the receiving protocol includes corresponding parsing rules; extracting the interaction features of the parsing rules in the acquisition protocol set and the receiving protocol set, and constructing the point-to-end interaction system based on the interaction features.

[0011] Furthermore, the fault early warning module includes: Fault prediction unit: performs fault prediction based on the data according to the fault prediction model; Fault grading unit: Used to grade the current fault situation based on the fault prediction results of the prediction model. Furthermore, the fault isolation module includes: Fault feedback unit: used to locate the faulty equipment based on the specific fault situation and return the fault parameters of the equipment; Equipment control unit; generates equipment control parameters based on equipment fault parameters; Furthermore, the communication interface module includes: Information analysis unit: used to generate equipment control reports based on fault parameters and equipment control parameters; Information sending unit: Used to send the generated control report to the central server.

[0012] The beneficial effects of this invention are as follows: Improved response speed: This invention achieves data acquisition, parsing, and diagnosis by directly connecting the protocol parsing module to the local control module, which greatly improves the fault response speed. At the same time, the protocol parsing module is designed with a dynamic parsing method to ensure high protocol compatibility when facing multiple parsing requests, maximize parsing efficiency, reduce the impact of high latency issues, and ensure the real-time performance of the system.

[0013] Reduce the burden on the central server: The independent fault isolation module reduces the load on the central server, avoids system bottlenecks caused by excessive requests, and improves the overall processing capacity of the system.

[0014] Localized fault diagnosis: Real-time fault detection is performed through local modules, avoiding excessive reliance on central servers, improving the device's self-diagnostic capabilities, and reducing performance degradation caused by network latency. Attached Figure Description

[0015] Figure 1 The diagram shows the structure of a solid-state transformer fault early warning and isolation system based on protocol access. Figure 2 The diagram shows a flowchart of the method by which the protocol parsing module of the present invention receives device data frames from the local controller module and parses the device data frames through dynamic parsing. Detailed Implementation

[0016] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0017] Figure 1 The diagram shows a flowchart of a solid-state transformer fault early warning and isolation system based on protocol access.

[0018] Reference Figure 1 This invention proposes a solid-state transformer fault early warning and isolation system based on protocol access. The system includes a local control module, a protocol parsing module, a fault isolation module, and a communication interface module. Local control module: When the solid-state transformer starts working, it transmits and stores the generated device data frames through sensors; Protocol parsing module: Receives device data frames from the local control module and parses the device data frames dynamically; Fault early warning module: Receives parsed data from the protocol parsing module, predicts transformer faults based on the fault prediction model, and evaluates the prediction results; Fault isolation module: Receives the evaluation results from the fault early warning module, locates the equipment that needs to be isolated, disconnects the equipment from other components, and generates fault information; Communication interface module: Sends fault information to the central server or upstream equipment for fault data reporting and analysis.

[0019] The system adopts a distributed operating system architecture to enable collaborative work between modules and supports unified access for terminals using multiple protocols through a dynamic protocol adaptation mechanism.

[0020] Furthermore, the local control module includes: Sensor deployment unit: Deploys sensors to the data interfaces of various devices according to sensor deployment rules, and receives device data at the interfaces; Furthermore, for different types of data, corresponding sensors are installed at the corresponding data interfaces. For example, current sensors are installed at the current input and output terminals of the equipment, and temperature sensors are installed on the temperature control components of the cooling system and the surface of the transformer casing. For data that can be detected in real time, sensors are directly installed at the ports for data acquisition. The sensors are directly connected through standard industrial communication interfaces. For data that requires remote monitoring, wireless sensors can be used for data acquisition. The data is transmitted to the central data acquisition unit or server for remote monitoring via wireless communication protocols. The equipment data collected by the sensors is connected through a dedicated interface to ensure that each sensor is connected to the data acquisition device through a standard protocol. The collected real-time data is then transmitted to the data filtering unit.

[0021] Furthermore, in one possible embodiment, for parameters that can be directly acquired, such as current parameters, the system installs current sensors at the current input and output terminals of the device, and sets the range of the current sensors to 0 to 100A and the accuracy to 0.1A to ensure that the power consumption or current fluctuations of the device are accurately captured. Every 10 seconds, the current sensor directly transmits the real-time acquired data to the data acquisition unit via the MODBUS RTU protocol. Alternatively, in another possible embodiment, the system installs thermocouple-type temperature sensors on the temperature control components of the device's cooling system and on the surface of the transformer casing, with a range set to -50°C to 150°C and an accuracy of ±0.5°C. Every 10 seconds, the temperature sensor transmits the temperature data to the data acquisition unit via the MODBUS TCP protocol.

[0022] Furthermore, in another possible embodiment, for equipment data requiring remote monitoring, such as vibration parameters, the system installs wireless sensors on the vibration monitoring components of the equipment to collect vibration data in real time. The wireless vibration sensor is set to a monitoring range of 0-10 mm / s and an accuracy of 0.1 mm / s. Simultaneously, the sensor transmits data to the data acquisition unit via the Zigbee protocol every 1 minute. After the sensor transmits the data to the data acquisition unit, the data filtering unit performs preliminary filtering and cleaning on the received real-time data. For example, temperature data undergoes noise filtering to remove unstable or abnormal fluctuations, while current data undergoes smoothing to eliminate short-term current surges.

[0023] Furthermore, dynamic frequency sampling rules are added to the sensor deployment unit to enable adaptive adjustment of the sampling frequency based on different parameters, ensuring that the collected data is representative. For example, the sampling frequency can be dynamically adjusted based on the load condition of the transformer.

[0024] Furthermore, in one possible embodiment, when using a temperature sensor to collect transformer temperature, when the transformer load is below a preset threshold (e.g., 10%), the sampling frequency of the temperature sensor is adjusted to 1 minute / time; when the transformer load is between 20% and 50%, the sampling frequency is adjusted to 30 seconds / time; and when the transformer load is above the preset threshold (e.g., 50%), the sampling frequency is adjusted to 1 second / time. By adjusting the sampling frequency according to the load status and equipment operation, frequent sampling under low load conditions is avoided, saving computing and storage resources.

[0025] Data filtering unit: Used to filter out data that does not meet the requirements using filtering rules for abnormal data, so as to ensure data stability.

[0026] Furthermore, after acquiring the collected data transmitted by the sensor deployment unit, outliers are removed based on the Z-Score detection method. Specifically, the Z-Score calculation formula is as follows: Among them, Z x This represents the outlier score for the x-th data point. The standard deviation of the dataset is represented by the standard deviation of the dataset. This represents the mean of the dataset.

[0027] Furthermore, outlier scores are calculated for all collected data, and data with outlier scores greater than a preset threshold are removed to complete data filtering.

[0028] Furthermore, the above methods are only applicable to datasets that follow a normal distribution. The removal effect will be significantly worse for datasets with other distribution patterns or no distribution pattern. However, the data distribution when a sudden fault occurs inside a solid-state transformer is often irregular.

[0029] Based on this, in another possible implementation, abnormal data is removed by constructing a data filtering model: A data filtering model is constructed using a multilayer perceptron to automatically identify and filter abnormal data. The input layer of the network model receives a vector containing five features: read speed, write speed, I / O latency, health status, and error rate. The hidden layers of the model consist of three fully connected layers: the first layer has 128 neurons, the second layer has 64 neurons, and the third layer has 32 neurons. Each layer uses the ReLU activation function, the mathematical expression of which is: The output layer uses the Sigmoid activation function, outputting a probability value between 0 and 1. The probability of normal data tends to 0, and the probability of abnormal data tends to 1. The loss function is binary cross-entropy, and its mathematical expression is: Where N is the number of samples, yi is the true label of the sample, and pi is the probability value predicted by the model.

[0030] The loss function optimizes the model by minimizing the difference between the predicted value and the actual label. The optimizer uses the Adam optimization algorithm to update the network weights. The Adam algorithm combines the advantages of momentum and adaptive learning rate, and its formula is: Where η is the learning rate, β1 and β2 are the decay rates of the momentum term, and ϵ is a small constant to prevent division by zero errors.

[0031] The filtering model constructed using the above methods can automatically identify and filter abnormal data in solid-state drives, avoiding the limitations of traditional threshold-based filtering methods. By training and learning abnormal patterns in the data, the output results are used to judge the stability of the data, filter out abnormal data, and ensure the data quality and reliability of the system. Compared with traditional methods, the above model can adapt to the performance characteristics and usage environment of different solid-state drives, and has strong versatility and adaptability, thus improving the overall performance of the storage system.

[0032] Data transmission unit: Securely and losslessly transmits data using encryption and lossless compression algorithms to ensure security during data transmission.

[0033] Furthermore, in one possible embodiment, the data transmission unit first compresses the raw data collected by the sensor using the LZ77 compression algorithm. The LZ77 algorithm reduces the data volume by finding duplicate substrings in the data and replacing them with triplets of offset and length. During the compression process, the data is divided into a lookup window and a lookahead window. Duplicate parts are located and replaced using a sliding window to generate compressed data. The compressed data is then encrypted using the AES encryption algorithm. AES is a symmetric key encryption algorithm whose encryption process includes multiple rounds of byte substitution, row shifting, column mixing, and round key XOR operations to ensure the confidentiality and unpredictability of the data. AES supports keys of different lengths (128-bit, 192-bit, or 256-bit). The number of encryption rounds is determined by the selected key length. During the encryption process, each data block (128-bit) is XORed with the key, and a series of transformations are used to make the data difficult to crack. The encrypted data is transmitted through a secure communication channel to ensure that its content is not accessed or tampered with by unauthorized third parties during transmission.

[0034] Data storage unit: Stores the processed data to a distributed database.

[0035] Furthermore, in one possible embodiment, the data storage unit receives processed encrypted and compressed data from the data transmission unit. Upon receiving the encrypted data, the data storage unit first decrypts it using the same AES key to recover the compressed data. Then, it applies the LZ77 decompression algorithm to the decrypted data to restore the original data content. Next, the data storage unit allocates the data to appropriate storage nodes based on the data type and storage requirements, ensuring balanced distribution and redundant backup across multiple nodes. The data is written through a distributed database interface, guaranteeing high availability and fault tolerance. During data storage, the data storage unit utilizes the database's partitioning and indexing mechanisms to optimize data query efficiency and access speed. This combination of LZ77 compression and AES encryption not only efficiently compresses data, reducing storage and bandwidth requirements, but also ensures data security during transmission.

[0036] Furthermore, the measurement terminal includes, but is not limited to, any one or more of the following terminals: smart meters, current sensors, voltage sensors, fiber optic sensors, temperature and humidity sensors, environmental monitoring terminals, and data loggers.

[0037] Furthermore, the protocol parsing module: receives device data frames from the local control module and parses the device data frames using dynamic parsing; Furthermore, the step of receiving device data frames from the local control module and parsing the device data frames dynamically includes the following steps: A1. Construct a point-to-point interaction system between the sensor deployment module and the data parsing module, and construct a feature data system within the data acquisition room based on the point-to-point interaction system; A2, when parsing data, constructs data-protocol response features based on the characteristic data system and parsing parameters within the data; A3, calibrating protocol response standards based on data-protocol response characteristics and historical interaction data; A4 dynamically adjusts the parsing parameters based on the real-time data transmitted and the calibration protocol response standard.

[0038] Further, in A1, the method for constructing a point-to-end interaction system between the sensor deployment module and the data parsing module, and constructing a feature data system within the data acquisition interval based on the point-to-end interaction system, specifically includes: obtaining all supported protocols supported by the sensor data acquisition terminal to form a collection protocol set, wherein for each supported protocol in the collection protocol set, a parsing rule corresponding to the supported protocol is also included; obtaining all supported receiving protocols included in the transmission port between the protocol parsing module and the sensor data acquisition module to form a receiving protocol set, wherein for each receiving protocol in the receiving protocol set, a parsing rule corresponding to the receiving protocol is also included; extracting the interaction features of the parsing rules in the collection protocol set and the receiving protocol set respectively, constructing a point-to-end interaction system based on the interaction features of the parsing rules, mapping the collection protocols and receiving protocols with consistent parsing rules one-to-one in the constructed point-to-end interaction system to form a classification system, obtaining the data transmission format supported by each classification relationship in the classification system, wherein the transmission format includes data transmission structure and verification method, extracting the interaction features between the data transmission formats through a feature extraction network, and constructing a feature data system within the data interval based on the interaction features; Furthermore, in A2, the method for constructing data-protocol response features based on the feature data system and parsing parameters during data parsing specifically includes: when the sensor starts collecting data, acquiring all protocols supported by the sensor acquisition end and the collected data information, the collected data information including the data acquisition time, data transmission format, corresponding sensor identifier, and corresponding protocol identifier; extracting the protocol identifier from the collected data information, converting the protocol identifier into a unique string, creating an index table using the converted unique string and the data; finding the corresponding protocol identifier in the classification system based on the index table; parsing the data using the corresponding protocol identifier and the protocol; acquiring parsing parameters during parsing, the parsing parameters including data intersection rate, data error rate, and data transmission speed; standardizing the parsing parameters using a standardization processing algorithm; generating one-dimensional feature vectors for each of the processed parsing parameters based on a feature extraction network; concatenating multiple one-dimensional feature vectors that have been parsed into a multi-dimensional feature vector; separating and integrating the multi-dimensional feature vector; and using the separated and integrated feature vector as the data-protocol response feature of the corresponding protocol.

[0039] Furthermore, in one possible implementation, when the system starts up, it first automatically detects and acquires all data acquisition protocols supported by the sensor. These protocols not only define the data acquisition method but also how to transmit and parse the data. Simultaneously, the system also collects information related to each data point, including the data acquisition time, transmission format, sensor's unique identifier, and protocol identifier. The system extracts the protocol identifier from this information and converts each protocol identifier into a unique string. This converted string is a standardized form of the protocol identifier, used to ensure consistent protocol recognition within the system. Next, the system uses this unique string along with the corresponding data record to create an index table. Through this index table, the system can find the protocol identifier corresponding to the data record within the classification system and uses this protocol identifier to parse the data. During the parsing process, the system simultaneously acquires information generated during parsing. The parameters to be analyzed include data intersection frequency (the frequency of interaction between data packets), data error rate (the number of errors that occur during data transmission), and data transmission rate (the number of bytes transmitted per second). After obtaining the analyzed parameters, to avoid the influence of certain extreme values, the system first needs to standardize them to ensure that different features are on the same scale, thereby avoiding the impact of differences between different units on subsequent analysis. Different standardization algorithms are used for different analyzed data. For example, for data intersection frequency, due to the different acquisition frequencies of the sensor acquisition terminals and the different operating modes of the system, it is possible that some sensors are working while others are not. This will result in multiple extreme value distributions in the data intersection frequency generated when different protocols are working. Therefore, the Z-Score algorithm can be used to standardize the data intersection frequency, where the Z-Score calculation formula is: Where x represents the x-th data point, represents the standard deviation of the data intersection volume acquired during this period, and μ represents the mean of the data intersection volume acquired during this period.

[0040] Z-Score standardization transforms data into a standard normal distribution with a mean of 0 and a standard deviation of 1. It is suitable for data with a large fluctuation range. In one possible implementation, the data overlap in the acquired parsed data may vary significantly between different sensors due to differences in sensor capabilities or system operating modes. Z-Score standardization converts all data overlaps into a standardized value with the same scale, facilitating subsequent processing. In another possible implementation, since different protocols parse data simultaneously, the byte length distribution transmitted within a certain time is uneven. Therefore, the Min-Max algorithm can be used to standardize the data transmission speed. The Min-Max calculation formula is: in This represents the standardized data. These are the maximum and minimum values ​​of the data, respectively. The Min-Max algorithm scales the data to a specified range, typically [0, 1]. For example, in another possible embodiment, assuming the data transmission rate is in the range of 0 to 1000 bytes / second at a certain moment, the data transmission rate will be mapped to between 0 and 1 through Min-Max normalization, making the scale of different data features consistent.

[0041] After standardization, the system converts each parsing parameter into a one-dimensional feature vector. In another possible implementation, the system calculates a standardized value of [0.25] for data transmission speed, [-0.5] for data error rate, and [1.2] for data intersection rate. Each data parsing parameter is converted into a corresponding one-dimensional feature vector. Subsequently, the system concatenates multiple one-dimensional feature vectors from each data acquisition record to form a multi-dimensional feature vector. For example, the concatenated feature vector might be [1.2, -0.5, 0.25]. This multi-dimensional feature vector integrates the standardization results of each parsing parameter and represents the characteristics of the data across all parsing dimensions.

[0042] Furthermore, in a feasible embodiment, a multidimensional feature vector is obtained for any protocol as follows: Where i and j represent the row and column indices of the elements in the multidimensional eigenvector, respectively. The multidimensional eigenvector A is decomposed into the product of j matrices. The left singular matrix of matrix A is U, the diagonal matrix is ​​∑, and the transpose of the right singular matrix is ​​V. T This allows the multidimensional feature vector to be decomposed into A = U∑V T Projecting the right singular matrix of matrix A onto the first eigenvector of the multidimensional eigenvector obtained in the first column is P1. P1 is taken as the component feature of the multidimensional eigenvector A on the first eigenvector, and Pi is taken as the component feature of the multidimensional eigenvector A on the i-th eigenvector. All component features are concatenated into a new eigenvector, and the new eigenvector is taken as the data-protocol response feature of the protocol.

[0043] Further, in A3, the specific method for calibrating protocol response standards based on data-protocol response characteristics and historical interaction data is as follows: The parsing parameters of each protocol within the last 24 hours, 72 hours, and 148 hours are obtained as historical interaction data. Multidimensional feature vectors of the parsing parameters are extracted using a feature extraction network. These multidimensional feature vectors are then labeled as 24-multidimensional feature vectors, 72-multidimensional feature vectors, and 148-multidimensional feature vectors according to the acquisition time. A response evaluation model is constructed based on the data-protocol response characteristics of the corresponding protocols obtained in step A2. The 24-multidimensional feature vectors, 72-multidimensional feature vectors, and 148-multidimensional feature vectors are then used to output data processing efficiency, data transmission efficiency, and data error correction efficiency through the response evaluation model. The output data processing efficiency, data transmission efficiency, and data error correction efficiency are compared based on preset response standards to obtain the data-protocol response standards for each protocol within the last 24 hours, 72 hours, and 148 hours. The response standards consist of multiple intervals, which are left-closed and right-open intervals.

[0044] Furthermore, in A4, the method for dynamically adjusting the parsing parameters based on the real-time data transmitted and the calibration protocol response standard specifically includes: acquiring parsing parameters for a preset duration at fixed intervals, converting the parsing parameters into multi-dimensional feature vectors using the method in step A2, outputting the status response of each protocol within the current acquisition duration through the response evaluation model, and generating parameter adjustment instructions based on the data-protocol response standards of each protocol in the most recent 24 hours, 72 hours, and 148 hours. Furthermore, the specific method for generating parameter adjustment instructions based on the data-protocol response standards of each protocol within the last 24 hours, 72 hours, and 148 hours is as follows: assign different weights to the data-protocol response standards of each protocol within the last 24 hours, 72 hours, and 148 hours according to the actual situation to obtain a comprehensive standard interval; compare the data-protocol response status of each protocol obtained in step A4 with the comprehensive standard interval; and adjust the parsing parameters based on the comparison results.

[0045] Further, in one possible embodiment, the specific content of comparing the data-protocol response status of each protocol obtained in step A4 with the comprehensive standard interval, and adjusting the parsing parameters based on the comparison results, is as follows: Three intervals are set for each of the most recent 24 hours, 72 hours, and 148 hours, representing the three response statuses of "excellent," "good," and "poor," respectively. The three standard intervals for the 24-hour period are [a1,a2), [a2,a3), [a3,a4); for the 72-hour period, they are [b1,b2), [b2,b3), [b3,b4); and for the 148-hour period, they are [c1,c2), [c2,c3), [c3,c4). Based on the acquisition duration, different weights are assigned to these three sets of intervals to form a comprehensive interval.

[0046] Furthermore, in another possible implementation, the acquisition time is shorter than T1, and weights A, B, and C are assigned to the 24-hour, 72-hour, and 148-hour periods, respectively, where A+B+C=1. The comprehensive interval is: [Aa1+Bb1+Cc1, Aa2+Bb2+Cc2), [Aa2+Bb2+Cc2, Aa3+Bb3+Cc3), [Aa3+Bb3+Cc3, Aa4+Bb4+Cc4). When the protocol's data-protocol response status is in the interval [Aa2+Bb1+Cc1, Aa2+Bb2+Cc2, Aa3+Bb3+Cc3, Aa3+Bb3+Cc3, Aa4+Bb4+Cc4), the acquisition time is determined by the protocol's data-protocol response status. When the data-protocol response status is in the interval [Aa1+Bb1+Cc1,Aa2+Bb2+Cc2), adjust the parsing parameters to make it fall within the interval [Aa3+Bb3+Cc3,Aa4+Bb4+Cc4). When the data-protocol response status is in the interval [Aa3+Bb3+Cc3,Aa4+Bb4+Cc4), adjust the parsing parameters to make it fall within the interval [Aa2+Bb2+Cc2,Aa3+Bb3+Cc3). Adjust the parsing parameters to make it fall within the interval [Aa1+Bb1+Cc1,Aa2+Bb2+Cc2].

[0047] Furthermore, the fault early warning module includes: Fault prediction unit: performs fault prediction based on the data according to the fault prediction model; Furthermore, the method for predicting faults in data based on a preset fault prediction model specifically includes: receiving data parsed by a protocol parsing module, and outputting a fault judgment result from the parsed data through the preset fault prediction model.

[0048] Further, in one possible implementation, a pre-trained LSTM network is used as the fault prediction model. The fault prediction model includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer receives parsed data, and the LSTM layer extracts temporal features from the data transmitted from the input layer. The first layer of the LSTM layer contains 64 neurons, and the second layer contains 32 neurons. The fully connected layer maps the features extracted by the LSTM to multiple categories and contains 128 neurons, using the ReLU function as the activation function. The output layer contains multiple neurons, each corresponding to a fault level, and calculates a probability distribution based on the softmax activation function. The output consists of multiple fault levels and their corresponding probabilities. In one possible implementation, fault levels 1, 2, 3, and 4 are set, and the output is [0.1, 0.2, 0.5, 0.2], representing a probability of 0.1 for fault level 1, 0.2 for fault level 2, 0.5 for fault level 1, and 0.2 for fault level 4.

[0049] Fault grading unit: used to grade the current fault situation based on the fault prediction results of the prediction model; Furthermore, the method for classifying the current fault situation based on the fault prediction results of the prediction model specifically includes: classifying the current fault situation based on the fault prediction results of the fault prediction unit combined with classification rules. The classification rules specifically include: pre-setting a level priority rule, outputting the level with the highest probability as the fault level of the transformer, and when multiple highest probabilities occur, according to the level finite rule, taking the level with the highest priority as the fault level of the transformer.

[0050] Furthermore, in one possible implementation, the preset priority rule is Level 1 > Level 2 > Level 3 > Level 4, and the result output by the prediction module is [0.1, 0.2, 0.5, 0.2]. Then, the fault classification unit outputs the current fault level of the transformer as 3. In another possible implementation, the result output by the prediction module is [0.3, 0.3, 0.2, 0.2]. Then, the fault classification unit outputs the current fault level of the transformer as 1.

[0051] Furthermore, the fault isolation module includes: Fault feedback unit: used to locate the faulty equipment according to the fault level and return the fault parameters of the equipment; Furthermore, the system acquires the collected data within the acquisition period, locates the corresponding sensor device based on the sensor identifier corresponding to the acquired data, marks the device as a faulty device, uses all the collected data within the acquisition period as fault parameters of the faulty device, and returns the fault parameters to the device control unit.

[0052] Furthermore, in one possible implementation, the sensor identifier corresponding to the collected data is Y-01, where Y represents the sensor type and 01 represents the sensor serial number, such as TH-01, which represents the first temperature sensor. The device where the sensor is deployed is marked as a faulty device based on the geographical location of the temperature sensor, and all collected data collected by the first temperature sensor within the collection period is used as the fault parameters of the faulty device.

[0053] Equipment control unit: Generates equipment control parameters based on equipment fault parameters and sends them to the central server for execution.

[0054] Furthermore, by disconnecting the input or output of the inverter or rectifier of the faulty equipment in the transformer, and by disconnecting the isolating switch, the circuit connection on both sides of the faulty equipment can be quickly disconnected, thereby isolating the faulty equipment.

[0055] Furthermore, the communication interface module includes: Information analysis unit: used to generate equipment control reports based on fault parameters and equipment control parameters; Furthermore, the method for generating an equipment control report based on fault parameters and equipment control parameters specifically includes: The fault parameters output by the fault feedback unit are received and possible faults are predicted through pre-trained fault presets. The predicted equipment status or fault diagnosis results are output. Based on the analysis results, relevant information is extracted and an equipment control report is generated according to a preset template format. The report usually includes the current status of the equipment, whether there is a fault, the fault type, the fault prediction, and suggestions or adjustment measures for the equipment control parameters.

[0056] Furthermore, in one possible embodiment, the temperature sensor data received by the system shows that the device's operating temperature is 80°C, while the normal operating temperature range of the device is 60°C to 75°C. Exceeding this range may indicate a risk of overheating. At the same time, the pressure sensor data shows that the pressure fluctuates significantly, and the device's historical fault data indicates that similar temperature and pressure fluctuations often lead to cooling system failures. Based on the above parameters, the system uses a machine learning model to predict the device's status and determine whether the device has a fault or is about to fail. Assuming the model predicts an impending failure in the equipment's cooling system, the equipment's current status is marked as "failure prediction in progress," and the possible failure type is further diagnosed as "cooling system failure." The model also provides a further failure prediction result, predicting that the cooling system may fail within the next 48 hours. Simultaneously, the system extracts relevant information based on the above analysis results and generates an equipment control report according to a preset template format. The report includes the following: Current equipment status: The equipment is running, but the current operating temperature and pressure values ​​are outside the normal range, indicating a potential failure risk; Failure present: The equipment has not yet failed, but there is a risk of overheating, and the cooling system may be about to fail; Failure type: Predicted as "cooling system failure," possibly caused by excessive temperature and pressure fluctuations; Failure prediction: The system predicts that the cooling system will fail within the next 48 hours, therefore preventative maintenance is required; Control parameter recommendations or adjustment measures: It is recommended to lower the equipment temperature by adjusting the operating parameters of the cooling system (such as water pump speed or coolant flow rate) to maintain the equipment within the normal operating temperature range; Simultaneously, it is recommended to check the coolant flow rate and pump operating status in the cooling system to ensure they are working properly and reduce the likelihood of failure.

[0057] Information sending unit: Used to send the generated device control report to the central server.

[0058] Furthermore, in one possible embodiment, the information sending unit connects to the device or server via an Ethernet network interface, and the generated device control report is converted into JSON format and transmitted via the HTTP network communication protocol before being sent.

[0059] Furthermore, the information sending unit is connected to the device or server via a network interface, and the generated device control report is converted into a format suitable for transmission before being sent, and the data is transmitted via a network communication protocol; Furthermore, during the transmission process, the information transmission unit communicates with the target server using the API interface to ensure that the data can be transmitted correctly and received by the target device.

[0060] Furthermore, to ensure data security during transmission, the information sending unit also employs a secure transmission protocol for encryption to protect the information from being stolen or tampered with.

[0061] Furthermore, to improve the reliability of transmission, the information sending unit is also configured with a retry mechanism, and uses a queuing system to handle delays or backlogs in information transmission when the network fluctuates or the server is busy, ensuring that the report can be delivered accurately in the shortest possible time. In one possible embodiment, if the report fails to be sent successfully, the information sending unit will take corresponding actions according to the set strategy, such as resending, saving as a local file, or sending a warning to the administrator, to ensure the timeliness and completeness of the report.

[0062] Furthermore, in one possible embodiment, the information sending unit connects to the device or server via a network interface. The generated device control report is converted into a format suitable for transmission before being sent. The device control report adopts JSON format and includes the device's operating status, predicted fault information, fault type, suggested operating steps, etc. Then, the system sends the report data to the target server or device via the HTTP network communication protocol. The information sending unit communicates with the target server through an API interface and sends the JSON data of the report to the target server.

[0063] In another possible implementation, if a report fails to be transmitted due to network fluctuations or when the target server is busy, the information sending unit will activate a retry mechanism. When the report fails to be sent, the system will automatically try to resend the data. The information sending unit will wait for a predetermined time interval (e.g., 5 seconds) and then try to send the data again until it is successful. If the report still fails to be sent successfully, the information sending unit will save the data to a local file and try to send it again when the network returns to normal, ensuring that the data can eventually be successfully transmitted.

[0064] In another possible implementation, if the system is still unable to send after multiple retries, it may send a warning email to the administrator, reminding them to check the network status or the operating status of the target server. The system will also save the report data in local storage and send it automatically the next time the network is restored or the server becomes available.

[0065] In another possible implementation, assuming the device's control report encounters network fluctuations during transmission, the information sending unit first formats the report as JSON data and sends it to the target server's RESTful API interface via HTTPS. During transmission, if the report fails to reach the target server immediately due to network instability, the information sending unit will resend it according to the retry mechanism, waiting 5 seconds before attempting to send it again. If it still fails after 3 attempts, the report will be saved to the local disk, and an email will be sent to the administrator to notify them of the network connection problem. After receiving the warning, the administrator will manually check the network status and restore the connection. At this time, the information sending unit will continue to transmit the locally saved report through the queue system to ensure that the report can be delivered in a timely manner after the network is restored.

[0066] The protocol-based solid-state transformer fault early warning and isolation system described above can run on computing devices such as desktop minicomputers, laptops, handheld computers, and cloud servers. The system that can run on this protocol-based solid-state transformer fault early warning and isolation system may include, but is not limited to, processors and memory. Those skilled in the art will understand that the example described is merely an illustration of a protocol-based solid-state transformer fault early warning and isolation system and does not constitute a limitation on such a system. It may include more or fewer components, combinations of certain components, or different components. For example, the protocol-based solid-state transformer fault early warning and isolation system may also include input / output devices, network access devices, buses, etc.

[0067] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the protocol-access-based solid-state transformer fault early warning and isolation system, connecting various parts of the system through various interfaces and lines.

[0068] The memory can be used to store the computer program and / or modules. The processor implements various functions of the protocol-access-based solid-state transformer fault early warning and isolation system by running or executing the computer program and / or modules stored in the memory, and by calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0069] Although the invention has been described in considerable detail and particularly with regard to several of the described embodiments, it is not intended to limit itself to any of these details or embodiments or any particular embodiment, thereby effectively covering the intended scope of the invention. Furthermore, the invention has been described above with respect to embodiments foreseeable by the inventors in order to provide a useful description, and non-substantial modifications to the invention that have not yet been foreseen may still represent equivalent modifications.

Claims

1. A solid-state transformer fault early warning and isolation system based on protocol access, characterized in that, It includes a local control module, a protocol parsing module, a fault isolation module, and a communication interface module; Local control module: When the solid-state transformer starts working, it transmits and stores the generated device data frames through sensors; Protocol parsing module: Receives device data frames from the local control module and parses the device data frames dynamically; Fault early warning module: Receives parsed data from the protocol parsing module, predicts transformer faults based on the fault prediction model, and evaluates the prediction results; Fault isolation module: Receives the evaluation results from the fault early warning module, locates the equipment that needs to be isolated, disconnects the equipment from other components, and generates fault information; Communication interface module: Sends fault information to the central server or upstream equipment for fault data reporting and analysis; Communication interface module: Sends fault information to the central server or upstream equipment for fault data reporting and analysis; The aforementioned solid-state transformer fault early warning and isolation system based on protocol access achieves inter-module collaboration based on a distributed operating system architecture. The distributed operating system architecture adopts a distributed communication mechanism to realize data sharing between cross-regional devices and automatic takeover of faulty devices.

2. The solid-state transformer fault early warning and isolation system based on protocol access according to claim 1, characterized in that, The local control module includes a sensor deployment unit, a data filtering unit, a data transmission unit, and a data storage unit; Sensor deployment unit: Deploys sensors to the data interfaces of various devices according to sensor deployment rules, and receives device data at the interfaces; Data filtering unit: Used to filter out data that does not meet the requirements using filtering rules for abnormal data, ensuring data stability; Data transmission unit: Securely and losslessly transmits data using encryption and lossless compression algorithms to ensure security during data transmission; Data storage unit: Stores the processed data to a distributed database.

3. The solid-state transformer fault early warning and isolation system based on protocol access according to claim 2, characterized in that, The data interface consists of various measurement terminals within the transformer, including devices such as smart meters, current sensors, voltage sensors, fiber optic sensors, temperature and humidity sensors, environmental monitoring terminals, and data loggers.

4. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 2, characterized in that, The filtering rules are based on the Z-score algorithm and a filtering model, which is a data filtering model constructed using a multilayer perceptron.

5. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 1, characterized in that, The protocol parsing module receives device data frames from the local control module and parses the device data frames dynamically, including the following steps: A1. Construct a point-to-point interaction system between the sensor deployment module and the data parsing module, and construct a feature data system within the data acquisition room based on the point-to-point interaction system; A2, when parsing data, constructs data-protocol response features based on the characteristic data system and parsing parameters within the data; A3, calibrating protocol response standards based on data-protocol response characteristics and historical interaction data; A4 dynamically adjusts the parsing parameters based on the real-time data transmitted and the calibration protocol response standard.

6. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 5, characterized in that, The method for constructing the point-to-end interaction system specifically includes: acquiring all protocols supported by the sensor data acquisition terminal to form an acquisition protocol set, wherein the protocol includes corresponding parsing rules; acquiring the receiving protocols supported by the transmission port between the protocol parsing module and the sensor data acquisition module to form a receiving protocol set, wherein the receiving protocol includes corresponding parsing rules; extracting the interaction features of the parsing rules in the acquisition protocol set and the receiving protocol set, and constructing the point-to-end interaction system based on the interaction features.

7. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 4, characterized in that, The parsing parameters are the data intersection rate, data error rate, and data transmission speed obtained from the parsed data.

8. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 1, characterized in that, The fault early warning module includes a fault prediction unit and a fault classification unit; Fault prediction unit: performs fault prediction based on the data according to the fault prediction model; Fault grading unit: used to grade the current fault situation based on the fault prediction results of the prediction model.

9. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 1, characterized in that, The fault isolation module includes a fault feedback unit and a device control unit; Fault feedback unit: used to locate the faulty equipment based on the specific fault situation and return the fault parameters of the equipment; Equipment control unit; isolates faulty equipment based on its fault parameters.

10. A solid-state transformer fault early warning and isolation system based on protocol access according to claim 1, characterized in that, The communication interface module includes an information analysis unit and an information transmission unit; Information analysis unit: used to generate equipment control reports based on fault parameters and equipment control parameters; Information sending unit: Used to send the generated control report to the central server.

Images