Intelligent alarm and digital push advanced inspection system

Abstract

The invention relates to the technical field of power distribution network fault monitoring and prevention, and discloses an intelligent alarm and digital push advanced inspection system, which comprises the following steps: S1, receiving cable operation state data sent by data acquisition equipment; s2, sending the zero sequence grounding alarm signal to a server; s3, receiving alarm information sent by the server, wherein the alarm information is generated after the server encodes an analysis result into an XML (Extensible Markup Language) or JSON (JavaScript Object Notation) format according to a standardized structure and processes the analysis result through a lossless compression and encryption technology; and S4, decoding the alarm information and displaying the fault type, the position and the defect degree. According to the method, a database containing various partial discharge types and degrees is constructed through field experiment simulation, and the types, positions and defect degrees of zero-sequence grounding faults caused by distribution cable partial discharge are analyzed in combination with a frequency spectrum feature extraction algorithm and a pattern recognition algorithm. The problem that the fault type and degree are difficult to accurately judge in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of power distribution network fault monitoring and prevention technology, and in particular to an intelligent alarm and digital push proactive inspection system. Background Technology

[0002] In existing technologies, monitoring of zero-sequence grounding faults caused by partial discharge in power distribution cables mainly relies on sensor devices installed at cable nodes and a back-end analysis system. Sensors collect current, voltage, and discharge pulse signals to generate real-time operating status data, which is then transmitted to the monitoring center via wired or wireless communication. The back-end system typically uses threshold comparison or simple signal processing methods, combined with preset rules, to analyze the data to identify the fault type and location. Some systems utilize databases to store historical fault data and generate alarm information manually or semi-automatically, pushing it to maintenance personnel terminals to assist in inspection and maintenance work.

[0003] However, existing fault analysis methods are mainly based on fixed thresholds or simple feature matching, making it difficult to accurately identify the type, location, and severity of partial discharges. Existing databases are mostly general-purpose, lacking systematic experimental data for different discharge types (e.g., corona, surface, internal discharges) and severity (e.g., minor, moderate, severe), resulting in insufficient accuracy in the analysis results. Furthermore, existing methods typically rely on human experience or basic statistical models, lacking advanced spectral feature extraction and pattern recognition algorithms, thus failing to achieve advanced fault prediction and refined classification. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides an intelligent alarm and digital push-based advanced inspection system, which aims to improve the problem that existing fault analysis methods are mainly based on fixed thresholds or simple feature matching, making it difficult to accurately identify the type, location, and degree of partial discharge.

[0005] In a first aspect, the present invention provides the following technical solution: an intelligent alarm and digital push-based proactive inspection method, applied to a maintenance personnel terminal, for monitoring and preventing zero-sequence grounding faults caused by partial discharge of power distribution cables, comprising the following steps:

[0006] S1. Receive cable operating status data sent by the data acquisition device, wherein the data acquisition device monitors the zero-sequence grounding alarm signal caused by partial discharge in real time;

[0007] S2. Send the zero-sequence grounding alarm signal to the server. The alarm signal is used to trigger the server to generate analysis results of fault type, location and defect degree based on a pre-built partial discharge type and degree database, through spectrum feature extraction algorithm and pattern recognition algorithm.

[0008] S3. Receive alarm information sent by the server. The alarm information is generated by the server after encoding the analysis results into XML or JSON format according to a standardized structure and processing them with lossless compression and encryption technology.

[0009] S4. Decode the alarm information and display the fault type, location, and degree of defect.

[0010] Using the above technical solution, step S1 collects cable operating status data using a high-frequency current transformer and a partial discharge sensor (sampling frequency 1MHz to 10MHz), and transmits it to the terminal via RS485 or LoRa. Step S2 sends the alarm signal to the cloud server in binary data packet form via 4G / 5G or Wi-Fi. Step S3 receives alarm information in XML / JSON format pushed by the server (data packet 1KB to 10KB, AES-256 encryption, WebSocket transmission). Step S4 decodes the fault information on the Android / iOS terminal and displays it in an HTML5 / CSS3 interface, supporting multiple languages. A database containing various partial discharge types and degrees is constructed through field experimental simulation. Combining spectral feature extraction algorithms and pattern recognition algorithms, the type, location, and degree of defect caused by zero-sequence grounding faults due to partial discharge in distribution cables are analyzed.

[0011] Preferably, the partial discharge type and degree database is constructed by the server through the following steps:

[0012] S101. Data on different types and degrees of partial discharge are generated based on field experimental simulations;

[0013] S102. Acquire spectral characteristic signals containing partial discharge type, degree, and cable parameters;

[0014] S103. Store the data in a relational database.

[0015] Preferably, the spectral feature extraction algorithm is executed by a server and includes the following steps:

[0016] S201. Preprocess the zero-sequence grounding alarm signal, filter out noise interference, and generate a preprocessed signal x(t), where t represents time;

[0017] S202. Apply Fast Fourier Transform to convert the preprocessed signal x(t) into the frequency domain signal X(f), the formula is:

[0018] X(f)=∫x(t)e^(-j2πft)dt

[0019] Where f represents frequency, j represents the imaginary unit, e^(-j2πft) represents the complex exponential function, and ∫ represents the integral from negative infinity to positive infinity;

[0020] S203. Calculate the spectral characteristic parameters, including peak frequency f_p, harmonic components H_k, and energy distribution E, using the following formula:

[0021] The peak frequency f_p = argmax|X(f)|, where |X(f)| represents the amplitude of the frequency domain signal, and argmax represents the frequency corresponding to the maximum value.

[0022] The harmonic component H_k = |X(kf_0)|, where k represents the harmonic order and f_0 represents the fundamental frequency;

[0023] The energy distribution E = ∫|X(f)|^2df, where ∫ represents the integral over a specified frequency range;

[0024] S204 stores the peak frequency f_p, harmonic components H_k, and energy distribution E as a feature vector V=[f_p,H_1,H_2,...,H_n,E], where n represents the number of harmonic components.

[0025] Preferably, the pattern recognition algorithm is executed by a server and includes the following steps:

[0026] S301. Compare the feature vector V with the partial discharge feature template T_i in the database, where T_i = [f_p,i,H_1,i,H_2,i,...,H_n,i,E_i], and i represents the template number;

[0027] S302. Using the Support Vector Machine (SVM) algorithm, the feature vector V is mapped to a high-dimensional space through the kernel function K(V,T_i) to construct a classification hyperplane. The kernel function formula is:

[0028] K(V,T_i)=exp(-γ||V-T_i||^2),

[0029] Where γ represents the kernel function parameter, and ||V-T_i||^2 represents the squared Euclidean distance between the feature vector V and the template T_i;

[0030] S303. Calculate the classification result C based on the classification hyperplane, using the following formula:

[0031] C=sign(∑α_iy_iK(V,T_i)+b),

[0032] Where α_i represents the Lagrange multiplier, y_i represents the class label of template T_i, b represents the bias, sign represents the sign function, and ∑ represents the summation over all support vectors;

[0033] S304. Determine the fault type based on the classification result C, and estimate the degree of defect by combining the amplitude |V| of the feature vector V and the distribution characteristics E, and generate the analysis results of fault type, location and degree of defect.

[0034] Preferably, the alarm information is encoded by the server into a standardized structure containing the following fields: fault type, fault location, urgency level, occurrence time, cable parameters, and historical records; and is processed by the server using the Gzip lossless compression algorithm and the AES-256 encryption algorithm.

[0035] Preferably, the alarm information sent by the receiving server includes:

[0036] S401. Receive alarm information transmitted via an adaptive communication protocol, wherein the adaptive communication protocol is selected by the server according to the network environment, including using a UDP-based protocol in a local area network or using a TCP-based HTTP / HTTPS protocol in a wide area network.

[0037] S402. Receive alarm information pushed in real time via WebSocket technology, wherein the WebSocket push is performed by the server through message queue and load balancing technology.

[0038] Preferably, the step of decoding and displaying the alarm information includes:

[0039] S501. Decode alarm information in XML or JSON format on the terminal of the operation and maintenance personnel;

[0040] S502: Displays the fault type, location, and urgency level in a visual manner through an interface adapted to iOS, Android, or Windows operating systems.

[0041] Secondly, this invention provides the following technical solution: an intelligent alarm and digital push-based proactive inspection system, comprising:

[0042] The data acquisition and monitoring module is configured to acquire cable operating status data in real time and monitor zero-sequence grounding alarm signals caused by partial discharge;

[0043] The fault analysis and identification module is configured to analyze the fault type, location, and degree of defect based on a database of partial discharge types and degrees, using spectral feature extraction algorithms and pattern recognition algorithms.

[0044] The information encoding and processing module is configured to encode the analysis results into alarm information in XML or JSON format, and process the data through lossless compression and encryption technology;

[0045] The communication and push module is configured to transmit alarm information to the maintenance personnel's terminal through an adaptive communication protocol and real-time push technology.

[0046] The client receiving and display module is configured to decode and display alarm information on the terminals of operation and maintenance personnel.

[0047] Thirdly, the invention provides the following technical solution: a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned automated protection alarm and digital push advanced inspection method.

[0048] Fourthly, the present invention provides the following technical solution: a readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the above-mentioned automated protection alarm and digital push advanced inspection method is implemented.

[0049] The present invention has the following beneficial effects:

[0050] 1. In this invention, a database containing various partial discharge types and degrees is constructed through on-site experimental simulation. Combining spectral feature extraction algorithms and pattern recognition algorithms, the type, location, and degree of zero-sequence grounding faults caused by partial discharge in power distribution cables are analyzed. The database stores discharge type, discharge quantity, and cable parameters. The spectral feature extraction uses Fast Fourier Transform to convert the time-domain signal into frequency-domain features. The Support Vector Machine algorithm generates fault analysis results through kernel functions and classification hyperplanes, solving the problem of accurately identifying fault types and degrees in existing technologies. The combination of the database and the algorithm provides a brand-new analysis method for power inspection.

[0051] 2. In this invention, the communication mechanism dynamically selects the transmission protocol according to the network environment. In the local area network, the CoAP protocol based on UDP is used to reduce latency, and in the wide area network, the HTTP / HTTPS protocol based on TCP is used to ensure data integrity. Protocol switching is achieved by detecting network status through heartbeat packets. This adaptive selection mechanism can adapt to various scenarios such as urban power distribution networks and rural power grids, optimize the transmission process of alarm information, and provide a flexible communication method through dynamic protocol switching.

[0052] 3. In this invention, WebSocket technology is used to realize the real-time push of alarm information. The server side handles high-concurrency push tasks through message queue management and load balancing technology, while the client maintains connection stability through heartbeat detection and automatic reconnection mechanism. This push mechanism ensures that alarm information can be quickly transmitted to the terminal of maintenance personnel, which is suitable for the timeliness requirements of power inspection. Attached Figure Description

[0053] Figure 1 This is a flowchart of a method for intelligent alarm and digital push-based proactive inspection proposed in this invention.

[0054] Figure 2 This is a system framework diagram of an intelligent alarm and digital push proactive inspection system proposed in this invention. Detailed Implementation

[0055] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] Reference Figure 1 In the first embodiment of the present invention, an intelligent alarm and digital push proactive inspection system is provided, applied to the terminal of maintenance personnel, for monitoring and preventing zero-sequence grounding faults caused by partial discharge of power distribution cables, including the following steps:

[0058] S1. Receive cable operating status data sent by the data acquisition equipment. The data acquisition equipment monitors the zero-sequence grounding alarm signal caused by partial discharge in real time.

[0059] S2. Send a zero-sequence grounding alarm signal to the server. The alarm signal is used to trigger the server to generate analysis results of fault type, location and defect degree based on a pre-built partial discharge type and degree database, through spectrum feature extraction algorithm and pattern recognition algorithm.

[0060] S3. Receive alarm information sent by the server. The alarm information is generated by the server after encoding the analysis results into XML or JSON format according to a standardized structure and processing them with lossless compression and encryption technology.

[0061] S4. Decode alarm information and display the fault type, location, and degree of defect.

[0062] Specifically, in step S1, the data acquisition equipment includes high-frequency current transformers (CTs) and partial discharge sensors installed at cable nodes, with a sampling frequency of 1MHz to 10MHz. The acquired operating status data includes current waveforms, voltage waveforms, and partial discharge pulse signals. The data is transmitted to the maintenance personnel's terminal via RS485 or LoRa protocols. In step S2, the maintenance personnel's terminal sends alarm signals to the server via 4G / 5G or Wi-Fi networks. The signal format is a binary data packet containing a timestamp and device ID. The server is a distributed cloud server deployed at the power dispatch center. In step S3, the alarm information encoded by the server is pushed via the WebSocket protocol. The data packet size is 1KB to 10KB, encrypted, and digitally signed. In step S4, the maintenance personnel's terminal runs an Android or iOS system. After decoding, the fault information is displayed through a graphical interface developed using HTML5 and CSS3, supporting multi-language switching. Possible alternatives include: the data acquisition equipment can use the ZigBee protocol for transmission; the server can be deployed on edge computing nodes; alarm information can be pushed via the MQTT protocol; and the terminal interface can be developed based on the Qt framework. This embodiment supports multiple application scenarios, such as urban power distribution networks, rural power grids, or industrial park power systems.

[0063] The database of partial discharge types and degrees is built by the server through the following steps:

[0064] S101. Data on different types and degrees of partial discharge are generated based on field experimental simulations;

[0065] S102. Acquire spectral characteristic signals containing partial discharge type, degree, and cable parameters;

[0066] S103. Store the data in a relational database.

[0067] Specifically, the construction of a database of partial discharge types and degrees is the foundation of fault analysis. In step S101, the field experiment simulation is conducted in a laboratory or actual distribution network environment. The simulation scenarios include corona discharge, surface discharge, and internal discharge. The discharge degree is divided into slight (discharge amount <10pC), moderate (10pC to 100pC), and severe (>100pC). 10kV to 35kV cable samples are used, and the experimental equipment includes a high-voltage generator and a partial discharge tester. In step S102, the spectral characteristic signal is acquired through an oscilloscope with a sampling rate of 20MS / s. The characteristics include pulse peak value, duration, and repetition rate. Cable parameters include insulation material (XLPE or EPR), length, and service life. In step S103, the relational database uses MySQL or PostgreSQL. The table structure includes a discharge type table, a degree table, and a parameter table. The primary key is the experiment ID, and SQL queries and index optimization are supported. Alternative solutions include: experiments can be assisted by simulation software (such as COMSOL); characteristic signals can be acquired using a digital signal processor (DSP); and the database can be stored using NoSQL (such as MongoDB). This embodiment ensures that the database covers multiple discharge scenarios and supports rapid feature matching.

[0068] The spectral feature extraction algorithm is executed by the server and includes the following steps:

[0069] S201. Preprocess the zero-sequence grounding alarm signal, filter out noise interference, and generate a preprocessed signal x(t), where t represents time;

[0070] S202. Apply Fast Fourier Transform to convert the preprocessed signal x(t) into the frequency domain signal X(f), the formula is:

[0071] X(f)=∫x(t)e^(-j2πft)dt

[0072] Where f represents frequency, j represents the imaginary unit, e^(-j2πft) represents the complex exponential function, and ∫ represents the integral from negative infinity to positive infinity;

[0073] S203. Calculate the spectral characteristic parameters, including peak frequency f_p, harmonic components H_k, and energy distribution E, using the following formula:

[0074] The peak frequency f_p = argmax|X(f)|, where |X(f)| represents the amplitude of the frequency domain signal, and argmax represents the frequency corresponding to the maximum value.

[0075] The harmonic component H_k = |X(kf_0)|, where k represents the harmonic order and f_0 represents the fundamental frequency;

[0076] The energy distribution E = ∫|X(f)|^2df, where ∫ represents the integral over a specified frequency range;

[0077] S204 stores the peak frequency f_p, harmonic components H_k, and energy distribution E as a feature vector V=[f_p,H_1,H_2,...,H_n,E], where n represents the number of harmonic components.

[0078] Specifically, the spectral feature extraction algorithm is executed by the server to extract fault features from the zero-sequence grounding alarm signal. In step S201, preprocessing uses wavelet transform for denoising, with the mother wavelet being Daubechies (db4), a decomposition level of 5, and the signal x(t) being a time-domain current or voltage waveform with a length of 1024 sampling points. In step S202, the Fast Fourier Transform (FFT) is implemented using the Cooley-Tukey algorithm, with the input signal x(t) divided into even and odd sequences, a computational complexity of O(NlogN), where N is the number of sampling points, and the frequency resolution of the output frequency domain signal X(f) is 10Hz. In step S203, the peak frequency f_p is determined by comparing the absolute value of |X(f)|, the harmonic components H_k are calculated for the first 5 harmonics (k = 1 to 5), the fundamental frequency f_0 is typically 50Hz or 60Hz, and the energy distribution E is calculated through frequency domain integration, with an integration range of 0 to 500kHz. In step S204, the feature vector V = [f_p, H_1, H_2, H_3, H_4, H_5, E] is stored as a floating-point array of length 7. Alternative solutions include: Kalman filtering for denoising; replacing FFT with Discrete Cosine Transform (DCT); and extending the feature vector to 10 dimensions by incorporating phase information. This embodiment supports high-precision feature extraction and is suitable for different cable types.

[0079] The pattern recognition algorithm is executed by the server and includes the following steps:

[0080] S301. Compare the feature vector V with the partial discharge feature template T_i in the database, where T_i = [f_p,i,H_1,i,H_2,i,...,H_n,i,E_i], and i represents the template number;

[0081] S302. Using the Support Vector Machine (SVM) algorithm, the feature vector V is mapped to a high-dimensional space through the kernel function K(V,T_i) to construct a classification hyperplane. The kernel function formula is:

[0082] K(V,T_i)=exp(-γ||V-T_i||^2),

[0083] Where γ represents the kernel function parameter, and ||V-T_i||^2 represents the squared Euclidean distance between the feature vector V and the template T_i;

[0084] S303. Calculate the classification result C based on the classification hyperplane, using the following formula:

[0085] C=sign(∑α_iy_iK(V,T_i)+b),

[0086] Where α_i represents the Lagrange multiplier, y_i represents the class label of template T_i, b represents the bias, sign represents the sign function, and ∑ represents the summation over all support vectors;

[0087] S304. Determine the fault type based on the classification result C, and estimate the degree of defect by combining the amplitude |V| of the feature vector V and the distribution characteristics E, and generate the analysis results of fault type, location and degree of defect.

[0088] Specifically, the pattern recognition algorithm is executed by the server and used for fault classification and defect severity estimation. In step S301, the comparison between the feature vector V and the template T_i is calculated using Euclidean distance, with the formula d = √(∑(V_j - T_i,j)^2), where j is the vector component index, and the template T_i is retrieved from the database, with a quantity ranging from 100 to 1000. In step S302, the Support Vector Machine (SVM) uses a radial basis function (RBF) kernel, with the kernel function K(V,T_i) = exp(-γ||V-T_i||^2), where γ ranges from 0.01 to 1, and is optimized through grid search. The training dataset contains 500 labeled samples. In step S303, the classification hyperplane is determined by the optimization problem min(1 / 2||w||^2 + C∑ξ_i), where w is the normal vector, C is the penalty parameter (ranging from 1 to 10), ξ_i is the relaxation variable, and the classification result C represents the fault type (such as corona discharge or surface discharge). In step S304, the degree of defect is estimated using the feature vector magnitude |V|=√(∑V_j^2) and energy distribution E, with the magnitude range normalized to 0 to 1. This embodiment supports multi-class classification. Alternative solutions include: cosine similarity can be used for comparison; SVM can be replaced with random forest; and the training data can be expanded to 1000 samples.

[0089] Alarm information is encoded by the server into a standardized structure containing the following fields: fault type, fault location, urgency level, occurrence time, cable parameters, and historical records; and is processed by the server using Gzip lossless compression algorithm and AES-256 encryption algorithm.

[0090] Specifically, alarm information encoding and processing are handled by the server to ensure data structure and security. Encoded fields include: fault type (string, 20 bytes), fault location (latitude and longitude, 8 bytes), urgency level (enumeration value, 1 byte), occurrence time (UNIX timestamp, 8 bytes), cable parameters (JSON object, 100 bytes), and history (array, 500 bytes). XML encoding uses the UTF-8 character set, and JSON encoding follows the ECMA-404 standard. Gzip compression is set to compression level 6 with a compression ratio of 2:1 to 5:1. AES-256 encryption uses CBC mode, with a 256-bit key length and a 16-byte initialization vector (IV). An HMAC-SHA256 signature is added before encryption to verify integrity. Data processing is performed in the server's Nginx environment, with a single processing time of less than 10ms. Alternative solutions include: encoding using Protobuf format; compression using the Zstandard algorithm; and encryption using RSA-2048. This embodiment supports high compatibility and security and is suitable for cross-platform data exchange.

[0091] The alarm information received from the server includes:

[0092] S401. Receive alarm information transmitted via an adaptive communication protocol. The adaptive communication protocol is selected by the server according to the network environment, including using a UDP-based protocol in a local area network or an HTTP / HTTPS protocol based on TCP in a wide area network.

[0093] S402. Receive alarm information pushed in real time via WebSocket technology. WebSocket push is performed by the server through message queue and load balancing technology.

[0094] Specifically, alarm information reception involves communication protocols and push mechanisms. In step S401, the adaptive communication protocol is selected by the server based on network bandwidth and latency. For the local area network (LAN), UDP-based CoAP protocol is used, port 5683, and the data packet size is 128 bytes; for the wide area network (WAN), HTTP / HTTPS protocol is used, TLS version 1.3, and certificate ECC-256. Protocol switching is achieved by detecting network status via heartbeat packets, with a switching time of less than 100ms. In step S402, WebSocket push is based on the RFC6455 standard, with a frame size of 1KB. The server deploys a RabbitMQ message queue with a queue depth of 1000, and HAProxy is used for load balancing, supporting 1000 concurrent connections. The client sends a heartbeat packet every 30 seconds, and the reconnection interval is 5 seconds. This embodiment ensures low latency and high reliability. Alternative solutions include: using QUIC as the protocol; using Kafka as the message queue; and using Nginx for load balancing. This embodiment is suitable for high real-time scenarios, such as power emergency repairs.

[0095] The steps for decoding and displaying alarm information include:

[0096] S501. Decode alarm information in XML or JSON format on the terminal of the operation and maintenance personnel;

[0097] S502: Displays the fault type, location, and urgency level in a visual manner through an interface adapted to iOS, Android, or Windows operating systems.

[0098] Specifically, the decoding and display of alarm information are performed on the maintenance personnel's terminal. In step S501, XML decoding uses a DOM parser, and JSON decoding uses the Jackson library, with decoding time less than 5ms and support for error recovery mechanisms. In step S502, the visualization interface uses the React Native framework, supports resolutions from 720p to 4K, and displays content including fault type (text), location (map control, using OpenStreetMap), and urgency level (color coding, red, yellow, green). The interface supports offline caching with a cache size of 10MB and a refresh rate of 1Hz. Multi-language support includes Chinese, English, and Spanish, with the NotoSans font. Alternative solutions include: decoding can use the SAX parser; the interface can use the Flutter framework; and the map can use Google Maps. This embodiment supports multi-device compatibility and user-friendly interaction, and is suitable for on-site inspections.

[0099] Example 2:

[0100] Reference Figure 2 In a second embodiment of the present invention, the present invention provides an intelligent alarm and digital push proactive inspection system, comprising:

[0101] The data acquisition and monitoring module is configured to acquire cable operating status data in real time and monitor zero-sequence grounding alarm signals caused by partial discharge;

[0102] The fault analysis and identification module is configured to analyze the fault type, location, and degree of defect based on a database of partial discharge types and degrees, using spectral feature extraction algorithms and pattern recognition algorithms.

[0103] The information encoding and processing module is configured to encode the analysis results into alarm information in XML or JSON format, and process the data through lossless compression and encryption technology;

[0104] The communication and push module is configured to transmit alarm information to the maintenance personnel's terminal through an adaptive communication protocol and real-time push technology.

[0105] The client receiving and display module is configured to decode and display alarm information on the terminals of operation and maintenance personnel.

[0106] Specifically, the data acquisition and monitoring module uses an embedded device with an ARM Cortex-A7 processor, 512MB of memory, and runs a Linux system. The sensor accuracy is 0.1 pC. The fault analysis and identification module is deployed on a cloud server with a 16-core CPU, 64GB of memory, running Ubuntu 20.04, and using SciPy and scikit-learn as its algorithm libraries. The information encoding and processing module uses the Java Spring Boot framework, with a HikariCP database connection pool and a processing throughput of 1000 records per second. The communication and push module supports 10,000 concurrent connections, a network bandwidth of 1Gbps, and a latency of less than 50ms. The client receiving and display module runs on a smartphone or tablet with an operating system version of Android 9.0 or iOS 13.0 or higher, requiring 100MB of storage space. This embodiment supports large-scale deployment. Alternative solutions include: the acquisition module can use a Raspberry Pi; the server can run CentOS; and the client can support Windows 10. This embodiment is applicable to various power system scenarios.

[0107] Example 3

[0108] In the third embodiment of the present invention, based on the same inventive concept, the present invention proposes a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the automated protection alarm and digital push advanced inspection method of the above embodiment.

[0109] Example 4

[0110] In the fourth embodiment of the present invention, based on the same inventive concept, a computer device is proposed, comprising: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory to execute the automated protection alarm and digital push advanced inspection method of the above embodiment.

[0111] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0112] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart alarm and digital push early inspection method applied to an operation and maintenance personnel terminal for monitoring and preventing zero sequence ground faults caused by partial discharge of power distribution cables, characterized in that, The method comprises the following steps: S1, receiving cable operation state data sent by a data acquisition device, the data acquisition device monitoring a zero sequence ground alarm signal caused by partial discharge in real time; S2, sending the zero sequence ground alarm signal to a server, the alarm signal being used to trigger the server to generate an analysis result of fault type, position and defect degree based on a pre-constructed partial discharge type and degree database through a spectrum feature extraction algorithm and a pattern recognition algorithm; S3, receiving alarm information sent by the server, the alarm information being generated after the server encodes the analysis result into an XML or JSON format according to a standardized structure and processes the analysis result through lossless compression and encryption technology; S4, decoding the alarm information and displaying the fault type, position and defect degree.

2. The method of claim 1, wherein, The partial discharge type and degree database is constructed by the server through the following steps: S101, generating data of different partial discharge types and degrees based on field experiment simulation; S102, collecting spectrum feature signals containing partial discharge types, degrees and cable parameters; S103, storing the data in a relational database.

3. The method of claim 1, wherein the method further comprises: The spectrum feature extraction algorithm is executed by the server and comprises the following steps: S201, pre-processing the zero sequence ground alarm signal to filter out noise interference and generating a pre-processed signal x(t), wherein t represents time; S202, applying fast Fourier transform to convert the pre-processed signal x(t) into a frequency domain signal X(f), the formula being: X(f) = ∫x(t)e^(-j2πft)dt wherein f represents frequency, j represents an imaginary unit, e^(-j2πft) represents a complex exponential function, and ∫ represents an integral from negative infinity to positive infinity; S203, calculating spectrum feature parameters, including peak frequency f_p, harmonic component H_k and energy distribution E, the formulas being as follows: peak frequency f_p = argmax|X(f)|, wherein |X(f)| represents the amplitude of the frequency domain signal, and argmax represents the frequency corresponding to the maximum value; harmonic component H_k = |X(kf_0)|, wherein k represents the harmonic order, and f_0 represents the fundamental frequency; energy distribution E = ∫|X(f)|^2df, ∫ representing an integral in a specified frequency range; S204, storing the peak frequency f_p, the harmonic component H_k and the energy distribution E as a feature vector V = [f_p, H_1, H_2, …, H_n, E], wherein n represents the number of harmonic components.

4. The method of claim 1, wherein the method further comprises: The pattern recognition algorithm is executed by the server and comprises the following steps: S301, comparing the feature vector V with a partial discharge feature template T_i in the database, wherein T_i = [f_p,i, H_1,i, H_2,i, …, H_n,i, E_i], i representing a template serial number; S302, adopting a support vector machine algorithm to map the feature vector V to a high-dimensional space through a kernel function K(V, T_i) to construct a classification hyperplane, the kernel function formula being: K(V, T_i) = exp(-γ||V-T_i||^2), wherein γ represents a kernel function parameter, and ||V-T_i||^2 represents the squared Euclidean distance between the feature vector V and the template T_i; S303, calculating a classification result C according to the classification hyperplane, and the formula is: C = sign(∑α_iy_iK(V,T_i)+b), wherein α_i represents a Lagrange multiplier, y_i represents the class label of the template T_i, b represents a bias, sign represents a sign function, and ∑ represents summation of all support vectors; S304, determining the fault type according to the classification result C, and estimating the defect degree in combination with the amplitude |V| and the distribution characteristic E of the feature vector V to generate an analysis result of the fault type, position and defect degree.

5. The method of claim 1, wherein, The alarm information is encoded by the server into a standardized structure containing the following fields: fault type, fault position, emergency degree, occurrence time, cable parameters and historical record; and is processed by the server using the Gzip lossless compression algorithm and the AES-256 encryption algorithm.

6. The automated protection alarm and digitized push advanced patrol method according to claim 1, wherein, The receiving of the alarm information transmitted by the server comprises: S401, receiving the alarm information transmitted through an adaptive communication protocol selected by the server according to the network environment, including using a UDP-based protocol in a local area network or using a TCP-based HTTP / HTTPS protocol in a wide area network; S402, receiving the alarm information pushed in real time through WebSocket technology, and the WebSocket push is performed by the server through message queuing and load balancing technology.

7. The automated protection alarm and digitized push advanced patrol method according to claim 1, wherein, The decoding and displaying of the alarm information comprises: S501, decoding the alarm information in XML or JSON format on the terminal of the operation and maintenance personnel; S502, displaying the fault type, position and emergency degree in a visual manner through an interface adapted to the iOS, Android or Windows operating system.

8. An intelligent alarm and digital push early inspection system, characterized in that, The automatic protection alarm and digitalized push early inspection method according to any one of claims 1-7 comprises: a data acquisition and monitoring module configured to acquire cable operation state data in real time and monitor a zero sequence ground alarm signal caused by partial discharge; a fault analysis and identification module configured to analyze the fault type, position and defect degree based on a partial discharge type and degree database through a frequency spectrum feature extraction algorithm and a pattern recognition algorithm; an information encoding and processing module configured to encode the analysis result into alarm information in XML or JSON format and process the data through lossless compression and encryption technology; a communication and push module configured to transmit the alarm information to the terminal of the operation and maintenance personnel through an adaptive communication protocol and real-time push technology; a client receiving and displaying module configured to decode and display the alarm information on the terminal of the operation and maintenance personnel.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the automatic protection alarm and digitalized push early inspection method according to any one of claims 1-7 when executing the computer program.

10. A readable storage medium, characterized by, The readable storage medium stores the computer program, and the computer program is executed by the processor to implement the automatic protection alarm and digitalized push early inspection method according to any one of claims 1-7.

Images